Skip to main content
Springer logoLink to Springer
. 2021 Sep 9;81(9):800. doi: 10.1140/epjc/s10052-021-09538-2

Precision luminosity measurement in proton–proton collisions at s=13TeV in 2015 and 2016 at CMS

A M Sirunyan 1, A Tumasyan 1, W Adam 2, J W Andrejkovic 2, T Bergauer 2, S Chatterjee 2, M Dragicevic 2, A Escalante Del Valle 2, R Frühwirth 2,195, M Jeitler 2,195, N Krammer 2, L Lechner 2, D Liko 2, I Mikulec 2, F M Pitters 2, J Schieck 2,195, R Schöfbeck 2, M Spanring 2, S Templ 2, W Waltenberger 2, C-E Wulz 2,195, V Chekhovsky 3, A Litomin 3, V Makarenko 3, M R Darwish 4,196, E A De Wolf 4, X Janssen 4, T Kello 4,197, A Lelek 4, H Rejeb Sfar 4, P Van Mechelen 4, S Van Putte 4, N Van Remortel 4, F Blekman 5, E S Bols 5, J D’Hondt 5, J De Clercq 5, M Delcourt 5, S Lowette 5, S Moortgat 5, A Morton 5, D Müller 5, A R Sahasransu 5, S Tavernier 5, W Van Doninck 5, P Van Mulders 5, D Beghin 6, B Bilin 6, B Clerbaux 6, G De Lentdecker 6, L Favart 6, A Grebenyuk 6, A K Kalsi 6, K Lee 6, M Mahdavikhorrami 6, I Makarenko 6, L Moureaux 6, L Pétré 6, A Popov 6, N Postiau 6, E Starling 6, L Thomas 6, M Vanden Bemden 6, C Vander Velde 6, P Vanlaer 6, D Vannerom 6, L Wezenbeek 6, T Cornelis 7, D Dobur 7, M Gruchala 7, G Mestdach 7, M Niedziela 7, C Roskas 7, K Skovpen 7, M Tytgat 7, W Verbeke 7, B Vermassen 7, M Vit 7, A Bethani 8, G Bruno 8, F Bury 8, C Caputo 8, P David 8, C Delaere 8, I S Donertas 8, A Giammanco 8, V Lemaitre 8, K Mondal 8, J Prisciandaro 8, A Taliercio 8, M Teklishyn 8, P Vischia 8, S Wertz 8, S Wuyckens 8, G A Alves 9, C Hensel 9, A Moraes 9, W L Aldá Júnior 10, M Barroso Ferreira Filho 10, H Brandao Malbouisson 10, W Carvalho 10, J Chinellato 10,198, E M Da Costa 10, G G Da Silveira 10,199, D De Jesus Damiao 10, S Fonseca De Souza 10, D Matos Figueiredo 10, C Mora Herrera 10, K Mota Amarilo 10, L Mundim 10, H Nogima 10, P Rebello Teles 10, L J Sanchez Rosas 10, A Santoro 10, S M Silva Do Amaral 10, A Sznajder 10, M Thiel 10, F Torres Da Silva De Araujo 10, A Vilela Pereira 10, C A Bernardes 11, L Calligaris 11, T R Fernandez Perez Tomei 11, E M Gregores 11, D S Lemos 11, P G Mercadante 11, S F Novaes 11, Sandra S Padula 11, A Aleksandrov 12, G Antchev 12, I Atanasov 12, R Hadjiiska 12, P Iaydjiev 12, M Misheva 12, M Rodozov 12, M Shopova 12, G Sultanov 12, A Dimitrov 13, T Ivanov 13, L Litov 13, B Pavlov 13, P Petkov 13, A Petrov 13, T Cheng 14, W Fang 14,197, Q Guo 14, T Javaid 14,200, M Mittal 14, H Wang 14, L Yuan 14, M Ahmad 15, G Bauer 15, C Dozen 15,201, Z Hu 15, J Martins 15,202, Y Wang 15, K Yi 15,203,204, E Chapon 16, G M Chen 16,200, H S Chen 16,200, M Chen 16, A Kapoor 16, D Leggat 16, H Liao 16, Z-A LIU 16,200, R Sharma 16, A Spiezia 16, J Tao 16, J Thomas-Wilsker 16, J Wang 16, H Zhang 16, S Zhang 16,200, J Zhao 16, A Agapitos 17, Y Ban 17, C Chen 17, Q Huang 17, A Levin 17, Q Li 17, M Lu 17, X Lyu 17, Y Mao 17, S J Qian 17, D Wang 17, Q Wang 17, J Xiao 17, Z You 18, X Gao 19,197, H Okawa 19, M Xiao 20, C Avila 21, A Cabrera 21, C Florez 21, J Fraga 21, A Sarkar 21, M A Segura Delgado 21, J Jaramillo 22, J Mejia Guisao 22, F Ramirez 22, J D Ruiz Alvarez 22, C A Salazar González 22, N Vanegas Arbelaez 22, D Giljanovic 23, N Godinovic 23, D Lelas 23, I Puljak 23, Z Antunovic 24, M Kovac 24, T Sculac 24, V Brigljevic 25, D Ferencek 25, D Majumder 25, M Roguljic 25, A Starodumov 25,205, T Susa 25, A Attikis 26, E Erodotou 26, A Ioannou 26, G Kole 26, M Kolosova 26, S Konstantinou 26, J Mousa 26, C Nicolaou 26, F Ptochos 26, P A Razis 26, H Rykaczewski 26, H Saka 26, M Finger 27,206, M Finger Jr 27,206, A Kveton 27, E Ayala 28, E Carrera Jarrin 29, S Abu Zeid 30,207, S Khalil 30,208, E Salama 30,207,209, M A Mahmoud 31, Y Mohammed 31, S Bhowmik 32, A Carvalho Antunes De Oliveira 32, R K Dewanjee 32, K Ehataht 32, M Kadastik 32, J Pata 32, M Raidal 32, C Veelken 32, P Eerola 33, L Forthomme 33, H Kirschenmann 33, K Osterberg 33, M Voutilainen 33, E Brücken 34, F Garcia 34, J Havukainen 34, V Karimäki 34, M S Kim 34, R Kinnunen 34, T Lampén 34, K Lassila-Perini 34, S Lehti 34, T Lindén 34, H Siikonen 34, E Tuominen 34, J Tuominiemi 34, P Luukka 35, H Petrow 35, T Tuuva 35, C Amendola 36, M Besancon 36, F Couderc 36, M Dejardin 36, D Denegri 36, J L Faure 36, F Ferri 36, S Ganjour 36, A Givernaud 36, P Gras 36, G Hamel de Monchenault 36, P Jarry 36, B Lenzi 36, E Locci 36, J Malcles 36, J Rander 36, A Rosowsky 36, M Ö Sahin 36, A Savoy-Navarro 36,210, M Titov 36, G B Yu 36, S Ahuja 37, F Beaudette 37, M Bonanomi 37, A Buchot Perraguin 37, P Busson 37, C Charlot 37, O Davignon 37, B Diab 37, G Falmagne 37, S Ghosh 37, R Granier de Cassagnac 37, A Hakimi 37, I Kucher 37, A Lobanov 37, M Nguyen 37, C Ochando 37, P Paganini 37, J Rembser 37, R Salerno 37, J B Sauvan 37, Y Sirois 37, A Zabi 37, A Zghiche 37, J-L Agram 38,211, J Andrea 38, D Apparu 38, D Bloch 38, G Bourgatte 38, J-M Brom 38, E C Chabert 38, C Collard 38, D Darej 38, J-C Fontaine 38,211, U Goerlach 38, C Grimault 38, A-C Le Bihan 38, P Van Hove 38, E Asilar 39, S Beauceron 39, C Bernet 39, G Boudoul 39, C Camen 39, A Carle 39, N Chanon 39, D Contardo 39, P Depasse 39, H El Mamouni 39, J Fay 39, S Gascon 39, M Gouzevitch 39, B Ille 39, Sa Jain 39, I B Laktineh 39, H Lattaud 39, A Lesauvage 39, M Lethuillier 39, L Mirabito 39, K Shchablo 39, L Torterotot 39, G Touquet 39, M Vander Donckt 39, S Viret 39, A Khvedelidze 40,206, Z Tsamalaidze 40,206, L Feld 41, K Klein 41, M Lipinski 41, D Meuser 41, A Pauls 41, M P Rauch 41, J Schulz 41, M Teroerde 41, D Eliseev 42, M Erdmann 42, P Fackeldey 42, B Fischer 42, S Ghosh 42, T Hebbeker 42, K Hoepfner 42, H Keller 42, L Mastrolorenzo 42, M Merschmeyer 42, A Meyer 42, G Mocellin 42, S Mondal 42, S Mukherjee 42, D Noll 42, A Novak 42, T Pook 42, A Pozdnyakov 42, Y Rath 42, H Reithler 42, J Roemer 42, A Schmidt 42, S C Schuler 42, A Sharma 42, S Wiedenbeck 42, S Zaleski 42, C Dziwok 43, G Flügge 43, W Haj Ahmad 43,212, O Hlushchenko 43, T Kress 43, A Nowack 43, C Pistone 43, O Pooth 43, D Roy 43, H Sert 43, A Stahl 43,213, T Ziemons 43, H Aarup Petersen 44, M Aldaya Martin 44, P Asmuss 44, I Babounikau 44, S Baxter 44, O Behnke 44, A Bermúdez Martínez 44, A A Bin Anuar 44, K Borras 44,214, V Botta 44, D Brunner 44, A Campbell 44, A Cardini 44, P Connor 44, S Consuegra Rodríguez 44, V Danilov 44, M M Defranchis 44, L Didukh 44, G Eckerlin 44, D Eckstein 44, L I Estevez Banos 44, E Gallo 44,215, A Geiser 44, A Giraldi 44, A Grohsjean 44, M Guthoff 44, A Harb 44, A Jafari 44,216, N Z Jomhari 44, H Jung 44, A Kasem 44,214, M Kasemann 44, H Kaveh 44, C Kleinwort 44, J Knolle 44, D Krücker 44, W Lange 44, T Lenz 44, J Leonard 44, J Lidrych 44, K Lipka 44, W Lohmann 44,217, T Madlener 44, R Mankel 44, I-A Melzer-Pellmann 44, J Metwally 44, A B Meyer 44, M Meyer 44, J Mnich 44, A Mussgiller 44, V Myronenko 44, Y Otarid 44, D Pérez Adán 44, D Pitzl 44, A Raspereza 44, B Ribeiro Lopes 44, J Rübenach 44, A Saggio 44, A Saibel 44, M Savitskyi 44, V Scheurer 44, C Schwanenberger 44,215, A Singh 44, R E Sosa Ricardo 44, N Tonon 44, O Turkot 44, A Vagnerini 44, M Van De Klundert 44, R Walsh 44, D Walter 44, Y Wen 44, K Wichmann 44, C Wissing 44, S Wuchterl 44, R Zlebcik 44, R Aggleton 45, S Bein 45, L Benato 45, A Benecke 45, K De Leo 45, T Dreyer 45, M Eich 45, F Feindt 45, A Fröhlich 45, C Garbers 45, E Garutti 45, P Gunnellini 45, J Haller 45, A Hinzmann 45, A Karavdina 45, G Kasieczka 45, R Klanner 45, R Kogler 45, V Kutzner 45, J Lange 45, T Lange 45, A Malara 45, A Nigamova 45, K J Pena Rodriguez 45, O Rieger 45, P Schleper 45, M Schröder 45, J Schwandt 45, D Schwarz 45, J Sonneveld 45, H Stadie 45, G Steinbrück 45, A Tews 45, B Vormwald 45, I Zoi 45, J Bechtel 46, T Berger 46, E Butz 46, R Caspart 46, T Chwalek 46, W De Boer 46, A Dierlamm 46, A Droll 46, K El Morabit 46, N Faltermann 46, K Flöh 46, M Giffels 46, J O Gosewisch 46, A Gottmann 46, F Hartmann 46,213, C Heidecker 46, U Husemann 46, I Katkov 46,218, P Keicher 46, R Koppenhöfer 46, S Maier 46, S Mallows 46, M Metzler 46, S Mitra 46, Th Müller 46, M Musich 46, M Neukum 46, G Quast 46, K Rabbertz 46, J Rauser 46, D Savoiu 46, D Schäfer 46, M Schnepf 46, D Seith 46, I Shvetsov 46, H J Simonis 46, R Ulrich 46, J Van Der Linden 46, R F Von Cube 46, M Wassmer 46, M Weber 46, S Wieland 46, R Wolf 46, S Wozniewski 46, S Wunsch 46, G Anagnostou 47, P Asenov 47, G Daskalakis 47, T Geralis 47, A Kyriakis 47, D Loukas 47, A Stakia 47, M Diamantopoulou 48, D Karasavvas 48, G Karathanasis 48, P Kontaxakis 48, C K Koraka 48, A Manousakis-Katsikakis 48, A Panagiotou 48, I Papavergou 48, N Saoulidou 48, K Theofilatos 48, E Tziaferi 48, K Vellidis 48, E Vourliotis 48, G Bakas 49, K Kousouris 49, I Papakrivopoulos 49, G Tsipolitis 49, A Zacharopoulou 49, I Evangelou 50, C Foudas 50, P Gianneios 50, P Katsoulis 50, P Kokkas 50, N Manthos 50, I Papadopoulos 50, J Strologas 50, M Csanad 51, M M A Gadallah 51,219, S Lökös 51,220, P Major 51, K Mandal 51, A Mehta 51, G Pasztor 51, A J Rádl 51, O Surányi 51, G I Veres 51, M Bartók 52,221, G Bencze 52, C Hajdu 52, D Horvath 52,222, F Sikler 52, V Veszpremi 52, G Vesztergombi 52, S Czellar 53, J Karancsi 53,221, J Molnar 53, Z Szillasi 53, D Teyssier 53, P Raics 54, Z L Trocsanyi 54,223, B Ujvari 54, T Csorgo 55,224, F Nemes 55,224, T Novak 55, S Choudhury 56, J R Komaragiri 56, D Kumar 56, L Panwar 56, P C Tiwari 56, S Bahinipati 57,225, D Dash 57, C Kar 57, P Mal 57, T Mishra 57, V K Muraleedharan Nair Bindhu 57,226, A Nayak 57,226, P Saha 57, N Sur 57, S K Swain 57, S Bansal 58, S B Beri 58, V Bhatnagar 58, G Chaudhary 58, S Chauhan 58, N Dhingra 58,227, R Gupta 58, A Kaur 58, S Kaur 58, P Kumari 58, M Meena 58, K Sandeep 58, J B Singh 58, A K Virdi 58, A Ahmed 59, A Bhardwaj 59, B C Choudhary 59, R B Garg 59, M Gola 59, S Keshri 59, A Kumar 59, M Naimuddin 59, P Priyanka 59, K Ranjan 59, A Shah 59, M Bharti 60,228, R Bhattacharya 60, S Bhattacharya 60, D Bhowmik 60, S Dutta 60, B Gomber 60,229, M Maity 60,230, S Nandan 60, P Palit 60, P K Rout 60, G Saha 60, B Sahu 60, S Sarkar 60, M Sharan 60, B Singh 60,228, S Thakur 60,228, P K Behera 61, S C Behera 61, P Kalbhor 61, A Muhammad 61, R Pradhan 61, P R Pujahari 61, A Sharma 61, A K Sikdar 61, D Dutta 62, V Jha 62, V Kumar 62, D K Mishra 62, K Naskar 62,231, P K Netrakanti 62, L M Pant 62, P Shukla 62, T Aziz 63, S Dugad 63, G B Mohanty 63, U Sarkar 63, S Banerjee 64, S Bhattacharya 64, R Chudasama 64, M Guchait 64, S Karmakar 64, S Kumar 64, G Majumder 64, K Mazumdar 64, S Mukherjee 64, D Roy 64, S Dube 65, B Kansal 65, S Pandey 65, A Rane 65, A Rastogi 65, S Sharma 65, H Bakhshiansohi 66,232, M Zeinali 66,233, S Chenarani 67,234, S M Etesami 67, M Khakzad 67, M Mohammadi Najafabadi 67, M Felcini 68, M Grunewald 68, M Abbrescia 69, R Aly 69,235, C Aruta 69, A Colaleo 69, D Creanza 69, N De Filippis 69, M De Palma 69, A Di Florio 69, A Di Pilato 69, W Elmetenawee 69, L Fiore 69, A Gelmi 69, M Gul 69, G Iaselli 69, M Ince 69, S Lezki 69, G Maggi 69, M Maggi 69, I Margjeka 69, V Mastrapasqua 69, J A Merlin 69, S My 69, S Nuzzo 69, A Pompili 69, G Pugliese 69, A Ranieri 69, G Selvaggi 69, L Silvestris 69, F M Simone 69, R Venditti 69, P Verwilligen 69, G Abbiendi 70, C Battilana 70, D Bonacorsi 70, L Borgonovi 70, S Braibant-Giacomelli 70, L Brigliadori 70, R Campanini 70, P Capiluppi 70, A Castro 70, F R Cavallo 70, C Ciocca 70, M Cuffiani 70, G M Dallavalle 70, T Diotalevi 70, F Fabbri 70, A Fanfani 70, E Fontanesi 70, P Giacomelli 70, L Giommi 70, C Grandi 70, L Guiducci 70, F Iemmi 70, S Lo Meo 70,236, S Marcellini 70, G Masetti 70, F L Navarria 70, A Perrotta 70, F Primavera 70, A M Rossi 70, T Rovelli 70, G P Siroli 70, N Tosi 70, S Albergo 71,237, S Costa 71,237, A Di Mattia 71, R Potenza 71, A Tricomi 71,237, C Tuve 71, G Barbagli 72, A Cassese 72, R Ceccarelli 72, V Ciulli 72, C Civinini 72, R D’Alessandro 72, F Fiori 72, E Focardi 72, G Latino 72, P Lenzi 72, M Lizzo 72, M Meschini 72, S Paoletti 72, R Seidita 72, G Sguazzoni 72, L Viliani 72, L Benussi 73, S Bianco 73, D Piccolo 73, M Bozzo 74, F Ferro 74, R Mulargia 74, E Robutti 74, S Tosi 74, A Benaglia 75, F Brivio 75, F Cetorelli 75, V Ciriolo 75,213, F De Guio 75, M E Dinardo 75, P Dini 75, S Gennai 75, A Ghezzi 75, P Govoni 75, L Guzzi 75, M Malberti 75, S Malvezzi 75, A Massironi 75, D Menasce 75, F Monti 75, L Moroni 75, M Paganoni 75, D Pedrini 75, S Ragazzi 75, T Tabarelli de Fatis 75, D Valsecchi 75,213, D Zuolo 75, S Buontempo 76, F Carnevali 76, N Cavallo 76, A De Iorio 76, F Fabozzi 76, A O M Iorio 76, L Lista 76, S Meola 76,213, P Paolucci 76,213, B Rossi 76, C Sciacca 76, P Azzi 77, N Bacchetta 77, D Bisello 77, P Bortignon 77, A Bragagnolo 77, R Carlin 77, P Checchia 77, P De Castro Manzano 77, T Dorigo 77, F Gasparini 77, U Gasparini 77, S Y Hoh 77, L Layer 77,238, M Margoni 77, A T Meneguzzo 77, M Presilla 77, P Ronchese 77, R Rossin 77, F Simonetto 77, G Strong 77, M Tosi 77, H YARAR 77, M Zanetti 77, P Zotto 77, A Zucchetta 77, G Zumerle 77, C Aime‘ 78, A Braghieri 78, S Calzaferri 78, D Fiorina 78, P Montagna 78, S P Ratti 78, V Re 78, M Ressegotti 78, C Riccardi 78, P Salvini 78, I Vai 78, P Vitulo 78, G M Bilei 79, D Ciangottini 79, L Fanò 79, P Lariccia 79, G Mantovani 79, V Mariani 79, M Menichelli 79, F Moscatelli 79, A Piccinelli 79, A Rossi 79, A Santocchia 79, D Spiga 79, T Tedeschi 79, P Azzurri 80, G Bagliesi 80, V Bertacchi 80, L Bianchini 80, T Boccali 80, E Bossini 80, R Castaldi 80, M A Ciocci 80, R Dell’Orso 80, M R Di Domenico 80, S Donato 80, A Giassi 80, M T Grippo 80, F Ligabue 80, E Manca 80, G Mandorli 80, A Messineo 80, F Palla 80, G Ramirez-Sanchez 80, A Rizzi 80, G Rolandi 80, S Roy Chowdhury 80, A Scribano 80, N Shafiei 80, P Spagnolo 80, R Tenchini 80, G Tonelli 80, N Turini 80, A Venturi 80, P G Verdini 80, F Cavallari 81, M Cipriani 81, D Del Re 81, E Di Marco 81, M Diemoz 81, E Longo 81, P Meridiani 81, G Organtini 81, F Pandolfi 81, R Paramatti 81, C Quaranta 81, S Rahatlou 81, C Rovelli 81, F Santanastasio 81, L Soffi 81, R Tramontano 81, N Amapane 82, R Arcidiacono 82, S Argiro 82, M Arneodo 82, N Bartosik 82, R Bellan 82, A Bellora 82, J Berenguer Antequera 82, C Biino 82, A Cappati 82, N Cartiglia 82, S Cometti 82, M Costa 82, R Covarelli 82, N Demaria 82, B Kiani 82, F Legger 82, C Mariotti 82, S Maselli 82, E Migliore 82, V Monaco 82, E Monteil 82, M Monteno 82, M M Obertino 82, G Ortona 82, L Pacher 82, N Pastrone 82, M Pelliccioni 82, G L Pinna Angioni 82, M Ruspa 82, R Salvatico 82, K Shchelina 82, F Siviero 82, V Sola 82, A Solano 82, D Soldi 82, A Staiano 82, M Tornago 82, D Trocino 82, S Belforte 83, V Candelise 83, M Casarsa 83, F Cossutti 83, A Da Rold 83, G Della Ricca 83, G Sorrentino 83, F Vazzoler 83, S Dogra 84, C Huh 84, B Kim 84, D H Kim 84, G N Kim 84, J Lee 84, S W Lee 84, C S Moon 84, Y D Oh 84, S I Pak 84, B C Radburn-Smith 84, S Sekmen 84, Y C Yang 84, H Kim 85, D H Moon 85, T J Kim 86, J Park 86, S Cho 87, S Choi 87, Y Go 87, B Hong 87, K Lee 87, K S Lee 87, J Lim 87, J Park 87, S K Park 87, J Yoo 87, J Goh 88, A Gurtu 88, H S Kim 89, Y Kim 89, J Almond 90, J H Bhyun 90, J Choi 90, S Jeon 90, J Kim 90, J S Kim 90, S Ko 90, H Kwon 90, H Lee 90, S Lee 90, B H Oh 90, M Oh 90, S B Oh 90, H Seo 90, U K Yang 90, I Yoon 90, D Jeon 91, J H Kim 91, B Ko 91, J S H Lee 91, I C Park 91, Y Roh 91, D Song 91, I J Watson 91, S Ha 92, H D Yoo 92, Y Choi 93, Y Jeong 93, H Lee 93, Y Lee 93, I Yu 93, T Beyrouthy 94, Y Maghrbi 94, V Veckalns 95,239, M Ambrozas 96, A Juodagalvis 96, A Rinkevicius 96, G Tamulaitis 96, A Vaitkevicius 96, W A T Wan Abdullah 97, M N Yusli 97, Z Zolkapli 97, J F Benitez 98, A Castaneda Hernandez 98, J A Murillo Quijada 98, L Valencia Palomo 98, G Ayala 99, H Castilla-Valdez 99, E De La Cruz-Burelo 99, I Heredia-De La Cruz 99,240, R Lopez-Fernandez 99, C A Mondragon Herrera 99, D A Perez Navarro 99, A Sanchez-Hernandez 99, S Carrillo Moreno 100, C Oropeza Barrera 100, M Ramirez-Garcia 100, F Vazquez Valencia 100, I Pedraza 101, H A Salazar Ibarguen 101, C Uribe Estrada 101, J Mijuskovic 102,241, N Raicevic 102, D Krofcheck 103, S Bheesette 104, A P H Butler 104, P H Butler 104, A Lokhovitskiy 104, P Lujan 104, A Ahmad 105, M I Asghar 105, A Awais 105, M I M Awan 105, H R Hoorani 105, W A Khan 105, M A Shah 105, M Shoaib 105, M Waqas 105, V Avati 106, L Grzanka 106, M Malawski 106, H Bialkowska 107, M Bluj 107, B Boimska 107, T Frueboes 107, M Górski 107, M Kazana 107, M Szleper 107, P Traczyk 107, P Zalewski 107, K Bunkowski 108, K Doroba 108, A Kalinowski 108, M Konecki 108, J Krolikowski 108, M Walczak 108, M Araujo 109, P Bargassa 109, D Bastos 109, A Boletti 109, P Faccioli 109, M Gallinaro 109, J Hollar 109, N Leonardo 109, T Niknejad 109, J Seixas 109, O Toldaiev 109, J Varela 109, S Afanasiev 110, D Budkouski 110, P Bunin 110, M Gavrilenko 110, I Golutvin 110, I Gorbunov 110, A Kamenev 110, V Karjavine 110, A Lanev 110, A Malakhov 110, V Matveev 110,242,243, V Palichik 110, V Perelygin 110, M Savina 110, D Seitova 110, V Shalaev 110, S Shmatov 110, S Shulha 110, V Smirnov 110, O Teryaev 110, N Voytishin 110, A Zarubin 110, I Zhizhin 110, G Gavrilov 111, V Golovtcov 111, Y Ivanov 111, V Kim 111,244, E Kuznetsova 111,245, V Murzin 111, V Oreshkin 111, I Smirnov 111, D Sosnov 111, V Sulimov 111, L Uvarov 111, S Volkov 111, A Vorobyev 111, Yu Andreev 112, A Dermenev 112, S Gninenko 112, N Golubev 112, A Karneyeu 112, M Kirsanov 112, N Krasnikov 112, A Pashenkov 112, G Pivovarov 112, D Tlisov 112, A Toropin 112, V Epshteyn 113, V Gavrilov 113, N Lychkovskaya 113, A Nikitenko 113,246, V Popov 113, G Safronov 113, A Spiridonov 113, A Stepennov 113, M Toms 113, E Vlasov 113, A Zhokin 113, T Aushev 114, O Bychkova 115, M Danilov 115,247, P Parygin 115, E Popova 115, V Rusinov 115, V Andreev 116, M Azarkin 116, I Dremin 116, M Kirakosyan 116, A Terkulov 116, A Belyaev 117, E Boos 117, M Dubinin 117,248, L Dudko 117, A Ershov 117, A Gribushin 117, A Kaminskiy 117,249, V Klyukhin 117, O Kodolova 117, I Lokhtin 117, S Obraztsov 117, S Petrushanko 117, V Savrin 117, V Blinov 118,250, T Dimova 118,250, L Kardapoltsev 118,250, I Ovtin 118,250, Y Skovpen 118,250, I Azhgirey 119, I Bayshev 119, V Kachanov 119, A Kalinin 119, D Konstantinov 119, V Petrov 119, R Ryutin 119, A Sobol 119, S Troshin 119, N Tyurin 119, A Uzunian 119, A Volkov 119, A Babaev 120, V Okhotnikov 120, L Sukhikh 120, V Borchsh 121, V Ivanchenko 121, E Tcherniaev 121, P Adzic 122,251, M Dordevic 122, P Milenovic 122, J Milosevic 122, V Milosevic 122, M Aguilar-Benitez 123, J Alcaraz Maestre 123, A Álvarez Fernández 123, I Bachiller 123, M Barrio Luna 123, Cristina F Bedoya 123, C A Carrillo Montoya 123, M Cepeda 123, M Cerrada 123, N Colino 123, B De La Cruz 123, A Delgado Peris 123, J P Fernández Ramos 123, J Flix 123, M C Fouz 123, O Gonzalez Lopez 123, S Goy Lopez 123, J M Hernandez 123, M I Josa 123, J León Holgado 123, D Moran 123, Á Navarro Tobar 123, A Pérez-Calero Yzquierdo 123, J Puerta Pelayo 123, I Redondo 123, L Romero 123, S Sánchez Navas 123, M S Soares 123, L Urda Gómez 123, C Willmott 123, J F de Trocóniz 124, R Reyes-Almanza 124, B Alvarez Gonzalez 125, J Cuevas 125, C Erice 125, J Fernandez Menendez 125, S Folgueras 125, I Gonzalez Caballero 125, E Palencia Cortezon 125, C Ramón Álvarez 125, J Ripoll Sau 125, V Rodríguez Bouza 125, A Trapote 125, J A Brochero Cifuentes 126, I J Cabrillo 126, A Calderon 126, B Chazin Quero 126, J Duarte Campderros 126, M Fernandez 126, C Fernandez Madrazo 126, P J Fernández Manteca 126, A García Alonso 126, G Gomez 126, C Martinez Rivero 126, P Martinez Ruiz del Arbol 126, F Matorras 126, J Piedra Gomez 126, C Prieels 126, F Ricci-Tam 126, T Rodrigo 126, A Ruiz-Jimeno 126, L Scodellaro 126, N Trevisani 126, I Vila 126, J M Vizan Garcia 126, M K Jayananda 127, B Kailasapathy 127,252, D U J Sonnadara 127, D D C Wickramarathna 127, W G D Dharmaratna 128, K Liyanage 128, N Perera 128, N Wickramage 128, T K Aarrestad 129, D Abbaneo 129, J Alimena 129, E Auffray 129, G Auzinger 129, J Baechler 129, P Baillon 129, A H Ball 129, D Barney 129, J Bendavid 129, N Beni 129, M Bianco 129, A Bocci 129, E Brondolin 129, T Camporesi 129, M Capeans Garrido 129, G Cerminara 129, S S Chhibra 129, L Cristella 129, D d’Enterria 129, A Dabrowski 129, N Daci 129, A David 129, A De Roeck 129, M Deile 129, R Di Maria 129, M Dobson 129, M Dünser 129, N Dupont 129, A Elliott-Peisert 129, N Emriskova 129, F Fallavollita 129,253, D Fasanella 129, S Fiorendi 129, A Florent 129, G Franzoni 129, J Fulcher 129, W Funk 129, S Giani 129, D Gigi 129, K Gill 129, F Glege 129, L Gouskos 129, M Haranko 129, J Hegeman 129, Y Iiyama 129, V Innocente 129, T James 129, P Janot 129, J Kaspar 129, J Kieseler 129, M Komm 129, N Kratochwil 129, C Lange 129, S Laurila 129, P Lecoq 129, K Long 129, C Lourenço 129, L Malgeri 129, S Mallios 129, M Mannelli 129, F Meijers 129, S Mersi 129, E Meschi 129, F Moortgat 129, M Mulders 129, S Orfanelli 129, L Orsini 129, F Pantaleo 129, L Pape 129, E Perez 129, M Peruzzi 129, A Petrilli 129, G Petrucciani 129, A Pfeiffer 129, M Pierini 129, H Qu 129, T Quast 129, D Rabady 129, A Racz 129, M Rieger 129, M Rovere 129, H Sakulin 129, J Salfeld-Nebgen 129, S Scarfi 129, C Schäfer 129, C Schwick 129, M Selvaggi 129, A Sharma 129, P Silva 129, W Snoeys 129, P Sphicas 129,254, S Summers 129, V R Tavolaro 129, D Treille 129, A Tsirou 129, P Tsrunchev 129, G P Van Onsem 129, M Verzetti 129, J Wanczyk 129,255, K A Wozniak 129, W D Zeuner 129, L Caminada 130,256, A Ebrahimi 130, W Erdmann 130, R Horisberger 130, Q Ingram 130, H C Kaestli 130, D Kotlinski 130, U Langenegger 130, M Missiroli 130, T Rohe 130, K Androsov 131,255, M Backhaus 131, P Berger 131, A Calandri 131, N Chernyavskaya 131, A De Cosa 131, G Dissertori 131, M Dittmar 131, M Donegà 131, C Dorfer 131, F Eble 131, T Gadek 131, T A Gómez Espinosa 131, C Grab 131, D Hits 131, W Lustermann 131, A-M Lyon 131, R A Manzoni 131, C Martin Perez 131, M T Meinhard 131, F Micheli 131, F Nessi-Tedaldi 131, J Niedziela 131, F Pauss 131, V Perovic 131, G Perrin 131, S Pigazzini 131, M G Ratti 131, M Reichmann 131, C Reissel 131, T Reitenspiess 131, B Ristic 131, D Ruini 131, D A Sanz Becerra 131, M Schönenberger 131, V Stampf 131, J Steggemann 131,255, R Wallny 131, D H Zhu 131, C Amsler 132,257, C Botta 132, D Brzhechko 132, M F Canelli 132, A De Wit 132, R Del Burgo 132, J K Heikkilä 132, M Huwiler 132, A Jofrehei 132, B Kilminster 132, S Leontsinis 132, A Macchiolo 132, P Meiring 132, V M Mikuni 132, U Molinatti 132, I Neutelings 132, G Rauco 132, A Reimers 132, P Robmann 132, S Sanchez Cruz 132, K Schweiger 132, Y Takahashi 132, C Adloff 133,258, C M Kuo 133, W Lin 133, A Roy 133, T Sarkar 133,230, S S Yu 133, L Ceard 134, P Chang 134, Y Chao 134, K F Chen 134, P H Chen 134, W-S Hou 134, Y y Li 134, R-S Lu 134, E Paganis 134, A Psallidas 134, A Steen 134, E Yazgan 134, P r Yu 134, B Asavapibhop 135, C Asawatangtrakuldee 135, N Srimanobhas 135, F Boran 136, S Damarseckin 136,259, Z S Demiroglu 136, F Dolek 136, I Dumanoglu 136,260, E Eskut 136, G Gokbulut 136, Y Guler 136, E Gurpinar Guler 136,261, I Hos 136,262, C Isik 136, E E Kangal 136,263, O Kara 136, A Kayis Topaksu 136, U Kiminsu 136, G Onengut 136, K Ozdemir 136,264, A Polatoz 136, A E Simsek 136, B Tali 136,265, U G Tok 136, S Turkcapar 136, I S Zorbakir 136, C Zorbilmez 136, B Isildak 137,266, G Karapinar 137,267, K Ocalan 137,268, M Yalvac 137,269, B Akgun 138, I O Atakisi 138, Y C Cekmecelioglu 138, E Gülmez 138, M Kaya 138,270, O Kaya 138,271, Ö Özçelik 138, S Tekten 138,272, E A Yetkin 138,273, A Cakir 139, K Cankocak 139,260, Y Komurcu 139, S Sen 139,274, F Aydogmus Sen 140, S Cerci 140,265, B Kaynak 140, S Ozkorucuklu 140, D Sunar Cerci 140,265, B Grynyov 141, L Levchuk 142, E Bhal 143, S Bologna 143, J J Brooke 143, A Bundock 143, E Clement 143, D Cussans 143, H Flacher 143, J Goldstein 143, G P Heath 143, H F Heath 143, L Kreczko 143, B Krikler 143, S Paramesvaran 143, T Sakuma 143, S Seif El Nasr-Storey 143, V J Smith 143, N Stylianou 143,275, J Taylor 143, A Titterton 143, K W Bell 144, A Belyaev 144,276, C Brew 144, R M Brown 144, D J A Cockerill 144, K V Ellis 144, K Harder 144, S Harper 144, J Linacre 144, K Manolopoulos 144, D M Newbold 144, E Olaiya 144, D Petyt 144, T Reis 144, T Schuh 144, C H Shepherd-Themistocleous 144, A Thea 144, I R Tomalin 144, T Williams 144, R Bainbridge 145, P Bloch 145, S Bonomally 145, J Borg 145, S Breeze 145, O Buchmuller 145, V Cepaitis 145, G S Chahal 145,277, D Colling 145, P Dauncey 145, G Davies 145, M Della Negra 145, S Fayer 145, G Fedi 145, G Hall 145, M H Hassanshahi 145, G Iles 145, J Langford 145, L Lyons 145, A-M Magnan 145, S Malik 145, A Martelli 145, J Nash 145,278, V Palladino 145, M Pesaresi 145, D M Raymond 145, A Richards 145, A Rose 145, E Scott 145, C Seez 145, A Shtipliyski 145, A Tapper 145, K Uchida 145, T Virdee 145,213, N Wardle 145, S N Webb 145, D Winterbottom 145, A G Zecchinelli 145, J E Cole 146, A Khan 146, P Kyberd 146, C K Mackay 146, I D Reid 146, L Teodorescu 146, S Zahid 146, S Abdullin 147, A Brinkerhoff 147, B Caraway 147, J Dittmann 147, K Hatakeyama 147, A R Kanuganti 147, B McMaster 147, N Pastika 147, S Sawant 147, C Smith 147, C Sutantawibul 147, J Wilson 147, R Bartek 148, A Dominguez 148, R Uniyal 148, A M Vargas Hernandez 148, A Buccilli 149, O Charaf 149, S I Cooper 149, D Di Croce 149, S V Gleyzer 149, C Henderson 149, C U Perez 149, P Rumerio 149, C West 149, A Akpinar 150, A Albert 150, D Arcaro 150, C Cosby 150, Z Demiragli 150, D Gastler 150, J Rohlf 150, K Salyer 150, D Sperka 150, D Spitzbart 150, I Suarez 150, A Tsatsos 150, S Yuan 150, D Zou 150, G Benelli 151, B Burkle 151, X Coubez 151,214, D Cutts 151, Y t Duh 151, M Hadley 151, U Heintz 151, J M Hogan 151,279, E Laird 151, G Landsberg 151, K T Lau 151, J Lee 151, J Luo 151, M Narain 151, S Sagir 151,280, E Usai 151, W Y Wong 151, X Yan 151, D Yu 151, W Zhang 151, C Brainerd 152, R Breedon 152, M Calderon De La Barca Sanchez 152, M Chertok 152, J Conway 152, P T Cox 152, R Erbacher 152, F Jensen 152, O Kukral 152, R Lander 152, M Mulhearn 152, D Pellett 152, B Regnery 152, D Taylor 152, M Tripathi 152, Y Yao 152, F Zhang 152, M Bachtis 153, R Cousins 153, A Dasgupta 153, A Datta 153, D Hamilton 153, J Hauser 153, M Ignatenko 153, M A Iqbal 153, T Lam 153, N Mccoll 153, W A Nash 153, S Regnard 153, D Saltzberg 153, C Schnaible 153, B Stone 153, V Valuev 153, K Burt 154, Y Chen 154, R Clare 154, J W Gary 154, G Hanson 154, G Karapostoli 154, O R Long 154, N Manganelli 154, M Olmedo Negrete 154, W Si 154, S Wimpenny 154, Y Zhang 154, J G Branson 155, P Chang 155, S Cittolin 155, S Cooperstein 155, N Deelen 155, J Duarte 155, R Gerosa 155, L Giannini 155, D Gilbert 155, J Guiang 155, R Kansal 155, V Krutelyov 155, R Lee 155, J Letts 155, M Masciovecchio 155, S May 155, S Padhi 155, M Pieri 155, B V Sathia Narayanan 155, V Sharma 155, M Tadel 155, A Vartak 155, F Würthwein 155, Y Xiang 155, A Yagil 155, N Amin 156, C Campagnari 156, M Citron 156, A Dorsett 156, V Dutta 156, J Incandela 156, M Kilpatrick 156, B Marsh 156, H Mei 156, A Ovcharova 156, M Quinnan 156, J Richman 156, U Sarica 156, D Stuart 156, S Wang 156, A Bornheim 157, O Cerri 157, I Dutta 157, J M Lawhorn 157, N Lu 157, J Mao 157, H B Newman 157, J Ngadiuba 157, T Q Nguyen 157, M Spiropulu 157, J R Vlimant 157, C Wang 157, S Xie 157, Z Zhang 157, R Y Zhu 157, J Alison 158, M B Andrews 158, T Ferguson 158, T Mudholkar 158, M Paulini 158, I Vorobiev 158, J P Cumalat 159, W T Ford 159, E MacDonald 159, R Patel 159, A Perloff 159, K Stenson 159, K A Ulmer 159, S R Wagner 159, J Alexander 160, Y Cheng 160, J Chu 160, D J Cranshaw 160, K Mcdermott 160, J Monroy 160, J R Patterson 160, D Quach 160, J Reichert 160, A Ryd 160, W Sun 160, S M Tan 160, Z Tao 160, J Thom 160, P Wittich 160, M Zientek 160, M Albrow 161, M Alyari 161, G Apollinari 161, A Apresyan 161, A Apyan 161, S Banerjee 161, L A T Bauerdick 161, A Beretvas 161, D Berry 161, J Berryhill 161, P C Bhat 161, K Burkett 161, J N Butler 161, A Canepa 161, G B Cerati 161, H W K Cheung 161, F Chlebana 161, M Cremonesi 161, K F Di Petrillo 161, V D Elvira 161, J Freeman 161, Z Gecse 161, L Gray 161, D Green 161, S Grünendahl 161, O Gutsche 161, R M Harris 161, R Heller 161, T C Herwig 161, J Hirschauer 161, B Jayatilaka 161, S Jindariani 161, M Johnson 161, U Joshi 161, P Klabbers 161, T Klijnsma 161, B Klima 161, M J Kortelainen 161, K H M Kwok 161, S Lammel 161, D Lincoln 161, R Lipton 161, T Liu 161, J Lykken 161, C Madrid 161, K Maeshima 161, C Mantilla 161, D Mason 161, P McBride 161, P Merkel 161, S Mrenna 161, S Nahn 161, V O’Dell 161, V Papadimitriou 161, K Pedro 161, C Pena 161,248, O Prokofyev 161, F Ravera 161, A Reinsvold Hall 161, L Ristori 161, B Schneider 161, E Sexton-Kennedy 161, N Smith 161, A Soha 161, L Spiegel 161, S Stoynev 161, J Strait 161, L Taylor 161, S Tkaczyk 161, N V Tran 161, L Uplegger 161, E W Vaandering 161, H A Weber 161, A Woodard 161, D Acosta 162, P Avery 162, D Bourilkov 162, L Cadamuro 162, V Cherepanov 162, F Errico 162, R D Field 162, D Guerrero 162, B M Joshi 162, M Kim 162, J Konigsberg 162, A Korytov 162, K H Lo 162, K Matchev 162, N Menendez 162, G Mitselmakher 162, D Rosenzweig 162, K Shi 162, J Sturdy 162, J Wang 162, E Yigitbasi 162, X Zuo 162, T Adams 163, A Askew 163, D Diaz 163, R Habibullah 163, S Hagopian 163, V Hagopian 163, K F Johnson 163, R Khurana 163, T Kolberg 163, G Martinez 163, H Prosper 163, C Schiber 163, R Yohay 163, J Zhang 163, M M Baarmand 164, S Butalla 164, T Elkafrawy 164,207, M Hohlmann 164, R Kumar Verma 164, D Noonan 164, M Rahmani 164, M Saunders 164, F Yumiceva 164, M R Adams 165, L Apanasevich 165, H Becerril Gonzalez 165, R Cavanaugh 165, X Chen 165, S Dittmer 165, O Evdokimov 165, C E Gerber 165, D A Hangal 165, D J Hofman 165, C Mills 165, G Oh 165, T Roy 165, M B Tonjes 165, N Varelas 165, J Viinikainen 165, X Wang 165, Z Wu 165, Z Ye 165, M Alhusseini 166, K Dilsiz 166,281, S Durgut 166, R P Gandrajula 166, M Haytmyradov 166, V Khristenko 166, O K Köseyan 166, J-P Merlo 166, A Mestvirishvili 166,282, A Moeller 166, J Nachtman 166, H Ogul 166,283, Y Onel 166, F Ozok 166,284, A Penzo 166, C Snyder 166, E Tiras 166,285, J Wetzel 166, O Amram 167, B Blumenfeld 167, L Corcodilos 167, M Eminizer 167, A V Gritsan 167, S Kyriacou 167, P Maksimovic 167, J Roskes 167, M Swartz 167, TÁ Vámi 167, C Baldenegro Barrera 168, P Baringer 168, A Bean 168, A Bylinkin 168, T Isidori 168, S Khalil 168, J King 168, G Krintiras 168, A Kropivnitskaya 168, C Lindsey 168, N Minafra 168, M Murray 168, C Rogan 168, C Royon 168, S Sanders 168, E Schmitz 168, J D Tapia Takaki 168, Q Wang 168, J Williams 168, G Wilson 168, S Duric 169, A Ivanov 169, K Kaadze 169, D Kim 169, Y Maravin 169, T Mitchell 169, A Modak 169, K Nam 169, F Rebassoo 170, D Wright 170, E Adams 171, A Baden 171, O Baron 171, A Belloni 171, S C Eno 171, Y Feng 171, N J Hadley 171, S Jabeen 171, R G Kellogg 171, T Koeth 171, A C Mignerey 171, S Nabili 171, M Seidel 171, A Skuja 171, S C Tonwar 171, L Wang 171, K Wong 171, D Abercrombie 172, G Andreassi 172, R Bi 172, S Brandt 172, W Busza 172, I A Cali 172, Y Chen 172, M D’Alfonso 172, G Gomez Ceballos 172, M Goncharov 172, P Harris 172, M Hu 172, M Klute 172, D Kovalskyi 172, J Krupa 172, Y-J Lee 172, B Maier 172, A C Marini 172, C Mironov 172, C Paus 172, D Rankin 172, C Roland 172, G Roland 172, Z Shi 172, G S F Stephans 172, K Tatar 172, J Wang 172, Z Wang 172, B Wyslouch 172, R M Chatterjee 173, A Evans 173, P Hansen 173, J Hiltbrand 173, Sh Jain 173, M Krohn 173, Y Kubota 173, Z Lesko 173, J Mans 173, M Revering 173, R Rusack 173, R Saradhy 173, N Schroeder 173, N Strobbe 173, M A Wadud 173, J G Acosta 174, S Oliveros 174, K Bloom 175, M Bryson 175, S Chauhan 175, D R Claes 175, C Fangmeier 175, L Finco 175, F Golf 175, J R González Fernández 175, C Joo 175, I Kravchenko 175, J E Siado 175, G R Snow 175, W Tabb 175, F Yan 175, G Agarwal 176, H Bandyopadhyay 176, L Hay 176, I Iashvili 176, A Kharchilava 176, C McLean 176, D Nguyen 176, J Pekkanen 176, S Rappoccio 176, A Williams 176, G Alverson 177, E Barberis 177, C Freer 177, Y Haddad 177, A Hortiangtham 177, J Li 177, G Madigan 177, B Marzocchi 177, D M Morse 177, V Nguyen 177, T Orimoto 177, A Parker 177, L Skinnari 177, A Tishelman-Charny 177, T Wamorkar 177, B Wang 177, A Wisecarver 177, D Wood 177, S Bhattacharya 178, J Bueghly 178, Z Chen 178, A Gilbert 178, T Gunter 178, K A Hahn 178, N Odell 178, M H Schmitt 178, K Sung 178, M Velasco 178, R Band 179, R Bucci 179, N Dev 179, R Goldouzian 179, M Hildreth 179, K Hurtado Anampa 179, C Jessop 179, K Lannon 179, N Loukas 179, N Marinelli 179, I Mcalister 179, F Meng 179, K Mohrman 179, Y Musienko 179,242, R Ruchti 179, P Siddireddy 179, M Wayne 179, A Wightman 179, M Wolf 179, M Zarucki 179, L Zygala 179, B Bylsma 180, B Cardwell 180, L S Durkin 180, B Francis 180, C Hill 180, A Lefeld 180, B L Winer 180, B R Yates 180, F M Addesa 181, B Bonham 181, P Das 181, G Dezoort 181, P Elmer 181, A Frankenthal 181, B Greenberg 181, N Haubrich 181, S Higginbotham 181, A Kalogeropoulos 181, G Kopp 181, S Kwan 181, D Lange 181, M T Lucchini 181, D Marlow 181, K Mei 181, I Ojalvo 181, J Olsen 181, C Palmer 181, D Stickland 181, C Tully 181, Z Xie 181, S Malik 182, S Norberg 182, A S Bakshi 183, V E Barnes 183, R Chawla 183, S Das 183, L Gutay 183, M Jones 183, A W Jung 183, S Karmarkar 183, M Liu 183, G Negro 183, N Neumeister 183, G Paspalaki 183, C C Peng 183, S Piperov 183, A Purohit 183, J F Schulte 183, M Stojanovic 183,210, J Thieman 183, F Wang 183, R Xiao 183, W Xie 183, J Dolen 184, N Parashar 184, A Baty 185, S Dildick 185, K M Ecklund 185, S Freed 185, F J M Geurts 185, A Kumar 185, W Li 185, B P Padley 185, R Redjimi 185, J Roberts 185, W Shi 185, A G Stahl Leiton 185, A Bodek 186, P de Barbaro 186, R Demina 186, J L Dulemba 186, C Fallon 186, T Ferbel 186, M Galanti 186, A Garcia-Bellido 186, O Hindrichs 186, A Khukhunaishvili 186, E Ranken 186, R Taus 186, B Chiarito 187, J P Chou 187, A Gandrakota 187, Y Gershtein 187, E Halkiadakis 187, A Hart 187, M Heindl 187, E Hughes 187, S Kaplan 187, O Karacheban 187,217, I Laflotte 187, A Lath 187, R Montalvo 187, K Nash 187, M Osherson 187, S Salur 187, S Schnetzer 187, S Somalwar 187, R Stone 187, S A Thayil 187, S Thomas 187, H Wang 187, H Acharya 188, A G Delannoy 188, S Spanier 188, O Bouhali 189,286, M Dalchenko 189, A Delgado 189, R Eusebi 189, J Gilmore 189, T Huang 189, T Kamon 189,287, H Kim 189, S Luo 189, S Malhotra 189, R Mueller 189, D Overton 189, D Rathjens 189, A Safonov 189, N Akchurin 190, J Damgov 190, V Hegde 190, S Kunori 190, K Lamichhane 190, S W Lee 190, T Mengke 190, S Muthumuni 190, T Peltola 190, S Undleeb 190, I Volobouev 190, Z Wang 190, A Whitbeck 190, E Appelt 191, S Greene 191, A Gurrola 191, W Johns 191, C Maguire 191, A Melo 191, H Ni 191, K Padeken 191, F Romeo 191, P Sheldon 191, S Tuo 191, J Velkovska 191, M W Arenton 192, B Cox 192, G Cummings 192, J Hakala 192, R Hirosky 192, M Joyce 192, A Ledovskoy 192, A Li 192, C Neu 192, B Tannenwald 192, E Wolfe 192, P E Karchin 193, N Poudyal 193, P Thapa 193, K Black 194, T Bose 194, J Buchanan 194, C Caillol 194, S Dasu 194, I De Bruyn 194, P Everaerts 194, F Fienga 194, C Galloni 194, H He 194, M Herndon 194, A Hervé 194, U Hussain 194, A Lanaro 194, A Loeliger 194, R Loveless 194, J Madhusudanan Sreekala 194, A Mallampalli 194, A Mohammadi 194, D Pinna 194, A Savin 194, V Shang 194, V Sharma 194, W H Smith 194, D Teague 194, S Trembath-Reichert 194, W Vetens 194; CMS Collaboration288
PMCID: PMC8550658  PMID: 34781320

Abstract

The measurement of the luminosity recorded by the CMS detector installed at LHC interaction point 5, using proton–proton collisions at s=13TeV in 2015 and 2016, is reported. The absolute luminosity scale is measured for individual bunch crossings using beam-separation scans (the van der Meer method), with a relative precision of 1.3 and 1.0% in 2015 and 2016, respectively. The dominant sources of uncertainty are related to residual differences between the measured beam positions and the ones provided by the operational settings of the LHC magnets, the factorizability of the proton bunch spatial density functions in the coordinates transverse to the beam direction, and the modeling of the effect of electromagnetic interactions among protons in the colliding bunches. When applying the van der Meer calibration to the entire run periods, the integrated luminosities when CMS was fully operational are 2.27 and 36.3 fb-1 in 2015 and 2016, with a relative precision of 1.6 and 1.2%, respectively. These are among the most precise luminosity measurements at bunched-beam hadron colliders.

Introduction

Luminosity, L, is a key parameter at particle colliders. Along with the energy available in the center-of-mass system, it is one of the two main figures of merit that quantify the potential for delivering large data samples and producing novel massive particles. The instantaneous luminosity L(t) is the process-independent ratio of the rate R(t) of events produced per unit of time dt to the cross section σ for a given process. The fundamental limitations on precise predictions for these cross sections (e.g., from quantum chromodynamics) motivate the techniques used for luminosity measurements at various types of colliders. The precise determination of the integrated luminosity, L(t)dt, has proven particularly challenging at hadron colliders, with an achieved precision typically ranging from 1 to 15% [1]. The “precision frontier” target of 1% [2] does not reflect a fundamental limitation, but rather results from a variety of uncorrelated sources of systematic uncertainty with typical magnitudes of 0.1–0.5%. In this paper, we report the precise determination of the absolute luminosity at the CERN LHC interaction point (IP) 5 with the CMS detector [3], using data from proton–proton (pp ) collisions at s=13TeV collected in 2015 and 2016.

A central component of the physics program at the LHC consists of measurements that can precisely test the validity of standard model (SM) predictions, e.g., cross sections for the production of electroweak gauge bosons [4, 5] or top quark pairs [6, 7]. A good understanding of the luminosity is critical to minimize the systematic uncertainty in these measurements. The uncertainty in the luminosity measurement is often the dominant systematic uncertainty [57], motivating an effort to improve its precision.

Stable luminosity information is also crucial to the efforts of the LHC operators to optimize the performance of the accelerator [8, 9]. In this context, it is important to provide luminosity information in real time at a high enough frequency to facilitate rapid optimization. The ability to measure the luminosity of individual bunch crossings (bunch-by-bunch luminosity) is also necessary so that the distribution of number of collisions per crossings is known to the experiments. This information is important when preparing simulations as well as optimization of thresholds to keep event-recording rates near data acquisition design targets.

An absolute luminosity scale is obtained with good accuracy using the direct method of van der Meer (vdM) scans [1013]. In these scans, the transverse separation of the two beams is varied over time and the resulting rate of some physical observables (e.g., number of charged particles passing through a silicon detector or energy deposited in a calorimeter) as a function of separation is used to extract the effective beam size. The absolute luminosity at one point in time can then be calculated from measurable beam parameters – namely, the transverse spatial widths of the overlap of the beams and the number of protons in each beam. To achieve the desired accuracy in the absolute luminosity calibration, the vdM scans are typically performed under carefully tailored conditions and with beam parameters optimized for that purpose [1], in conjunction with processing the input from accelerator instrumentation and multiple detector systems. A relative normalization method is then needed to transfer the absolute luminosity calibration to the complete data-taking period. To this end, for a given subdetector, the cross section σvis in the “visible” phase space region, defined by its acceptance, is measured for several observables. The integrated luminosity is obtained from the σvis-calibrated counts accumulated for a given period of data taking. Changes in the detector response over time can result in variations in σvis, which could appear as nonlinearity and/or long-term instability in the measured luminosity.

To address these challenges, CMS employs a multifaceted approach, in which measurements from various individual subsystems are used to produce a final luminosity value with high precision, good linearity, and stability. Several methods and independent detectors are used to provide redundancy and to minimize any bias originating from detector effects.

The LHC orbit is divided into a total of 3564 time windows 25ns long (bunch crossing slots), each of which can potentially contain a colliding bunch. However, the total number of filled bunch crossings is limited by design to a maximum of 2808 by the choice of the beam production scheme in the injectors and constraints from the rise times of injection and extraction kicker magnets in the various accelerators involved [14]. Furthermore, the length of the injections in 2015 and 2016 was limited by the maximal tolerable heat load in the arcs due to electron clouds (2015) and safety considerations in the LHC injection system with very luminous beams (2016) [15]. The bunch crossings are numbered with an identification number (BCID) in the range 1–3564. The specific pattern of filled and empty bunch crossings used in a single fill is known as the “filling scheme”; a typical filling scheme is composed of long strings of consecutive bunches, up to 72 bunches long, called a “train”, with the individual trains separated by gaps of varying lengths. Generally, filling schemes also include some number of noncolliding bunch crossings, where one beam is filled but the other remains empty; these can be used to study effects from beam-induced background. The two LHC beams are designated “beam 1” and “beam 2”, where beam 1 (beam 2) circulates in the clockwise (counterclockwise) direction, as viewed from above [14].

For Run 2 of the LHC, the period from 2015 to 2018 featuring pp collisions at s=13TeV, the CMS luminosity systems were significantly upgraded and expanded. We report the results for the first two years [16], in which the operational conditions feature a wide range in the number of colliding bunches nb and instantaneous luminosity, reaching a maximum of 2232 and 2208, and 0.5×1034 and 1.5×1034cm-2s-1 in 2015 and 2016, respectively. In the majority of pp LHC fills in Run 2, the bunches are spaced 25ns apart. The initial Run 2 data set delivered with a bunch spacing of 50ns is negligibly small [17], and hence not included in this paper. In this paper, “pileup” refers to the total number of pp interactions in a single bunch crossing, and “out-of-time pileup” refers to additional pp collisions in nearby bunches. For a total inelastic pp cross section of 80mb  [18, 19], the pileup during nominal physics data-taking conditions in 2015 (2016) extended from 5 to 35 (10 to 50) with an expected average (μ) of about 14 (27) pp interactions.

This paper is structured as follows. In Sect. 2 the CMS detector is described with special emphasis on the subdetectors used to derive observables for luminosity estimation, and in Sect. 3 we review the methods to obtain the luminosity information. Section 4 describes the vdM scan calibration method and the associated systematic uncertainty. Sections 5 and 6 outline the corrections applied to the luminosity algorithms and their resulting performance, respectively. Finally, Sect. 7 outlines the sources of corrections and the associated systematic uncertainties, and presents the main results. A summary is given in Sect. 8.

The CMS detector

The CMS detector is a multipurpose apparatus designed to study high-pT physics processes in pp collisions, as well as a broad range of phenomena in heavy ion collisions. The central element of CMS is a 3.8T superconducting solenoid, 13m in length and 6m in diameter. Within the solenoid volume are – in order of increasing radius from the beam pipe – a silicon pixel and strip tracker of high granularity for measuring charged particles up to pseudorapidity (η) of ±2.5; a lead tungstate crystal electromagnetic calorimeter for measurements of the energy of photons, electrons, and the electromagnetic component of hadronic showers (“jets”); and a brass and scintillator hadron calorimeter, each composed of a barrel and two endcap sections, for jet energy measurements. The forward hadron (HF) calorimeter uses steel as an absorber and quartz fibers as the sensitive material. The two halves of the HF are located 11.2m from the interaction region, one on each end, and together they provide coverage in the range 3.0<|η|<5.2, hence extending the pseudorapidity coverage provided by the barrel and endcap detectors. Outside the magnet, and within the range |η|<2.4, is the muon system [20], which is embedded in the iron flux-return yoke. It is composed of detection planes made using three technologies: drift tubes (DTs) in the barrel, cathode strip chambers (CSCs) in the endcaps, and resistive plate chambers (RPCs) both in the barrel and in the endcaps.

Events of interest for physics are selected using a two-tiered trigger system [21]. The first-level trigger, composed of custom hardware processors, uses information from the calorimeters and muon detectors to select events at a rate of around 100kHz. The second level, known as the high-level trigger, consists of a farm of processors running a version of the full event reconstruction software optimized for fast processing, and reduces the event rate to around 1kHz before data storage.

Several subdetectors, although not part of the main CMS data acquisition (DAQ) system, provide additional inputs (e.g., binary logic signals) to the triggering system. The two beam monitors closest to the IP for each LHC experiment, the Beam Pick-up Timing for eXperiments (BPTX) detectors [22], are reserved for timing measurements. They are located on either side of IP 5 at a distance of approximately 175m. The BPTX system can be used to provide a set of zero-bias events (i.e., events from nominally colliding bunch crossings but without a requirement for any specific activity in the event) by requiring a coincidence between the two BPTX sides. To suppress noise in triggers with high background, the presence of this coincidence is typically required [21].

The knowledge of the integrated luminosity requires stability over long periods of time, and hence benefits greatly from redundant measurements whose combination can lead to an improved precision. To that end, several upgrades were completed during the first LHC long shutdown (LS1), the transition period between LHC Run 1 (2009–2012) and Run 2. The main luminosity subdetectors (luminometers) in Run 1 were the silicon pixel detector and the HF. The HF back-end electronics, which were upgraded during LS1, consist of two independent readout systems: a primary readout over optical links for physics data taking, and a secondary readout using Ethernet links, explicitly reserved for luminosity data. In addition, two other luminometers were designed, constructed, and commissioned: the Pixel Luminosity Telescope (PLT) [23] and the Fast Beam Conditions Monitor (BCM1F) [24]. Finally, a separate DAQ system was developed that is independent of the central DAQ system [21, 25], so that HF, PLT, and BCM1F data, as well as LHC beam-related data, are collected and stored in a time- rather than event-based manner.

The luminometers, along with the accompanying algorithms used to estimate the instantaneous luminosity in Run 2, are briefly described in the following. Figure 1 shows an overview of the position of these luminometers within CMS. A more detailed description of the rest of the CMS detector, together with a definition of the coordinate system used and the relevant kinematic variables, is reported in Ref. [3].

Fig. 1.

Fig. 1

Schematic cross section through the CMS detector in the r-z plane. The main luminometers in Run 2, as described in the text, are highlighted, showing the silicon pixel detector, PLT, BCM1F, DTs, and HF. The two RAMSES monitors used as a luminometer in Run 2 are located directly behind HF. In this view, the detector is symmetric about the horizontal and vertical axes, so only one quarter is shown here. The center of the detector, corresponding to the approximate position of the pp collision point, is located at the origin. Solid lines represent distinct η values

Silicon pixel cluster counting

The pixel cluster counting (PCC) method, which uses the mean number of pixel clusters in the silicon pixel detector, exploits the very large number of pixels in the inner part of the CMS tracking system. The number of pixels in 2015–2016 was about 7×107, which means that the probability of a given pixel being hit by two different charged particles from the same bunch crossing is exceedingly small. The mean number of pixel clusters in simulated zero-bias events is of the order of 100 per pp collision, although the precise mean depends on the fraction of the detector used for a given data set. Assuming each pixel cluster comprises five pixels and using a typical pileup for the 2016 running of μ=27, the fraction of pixels hit in a typical bunch crossing is roughly:

f=NpixelhitNpixeltotal100×5×277×107=0.02%. 1

The probability of accidental overlap between pixel clusters is correspondingly small, and, as a consequence, the number of pixel clusters per bunch crossing is linearly dependent on pileup, and therefore an accurate measure of instantaneous luminosity. Simulated pp collision events that contain only in-time pileup and detector noise are generated using pythia version 8.223 [19] with the CUETP8M1 [18, 26] tune. The simulated events include a full simulation of the CMS detector response based on Geant4  [27]. For the sake of simplicity, the number of pileup interactions present in each simulated event is randomly generated from a Poisson distribution with μ up to 50. Figure 2 shows a representative PCC distribution at μ=45 and the average PCC as a function of μ. The latter distribution is fitted with a first-order polynomial, assuming no correlations among different values of μ. Good agreement is seen based on the estimated goodness-of-fit χ2 per degree of freedom (dof) value of about 0.5 [28], indicating linearity under simulated conditions.

Fig. 2.

Fig. 2

The upper plot shows the number of pixel clusters and their statistical uncertainty from simulation of pileup following a Poisson distribution with a mean of 45. The lower plot shows the mean number of pixel clusters from simulation as a function of mean pileup. The red curve is a first-order polynomial fit with slope and χ2/dof values shown in the legend. Only pixel modules considered for the PCC measurement in data are included. The lower panel of the lower plot shows the difference between the simulation and the linear fit in black points. The green band is the final linearity uncertainty for the 2016 data set

Only the components (modules) of the pixel subdetector that are stable for the entire period of data taking are used for the PCC rate measurements, excluding pixel modules known to be defective or significantly affected by the limited size of the readout buffer [29]. The measured σvis for PCC, σvisPCC, therefore depends on the data-taking period (i.e., one calibration per year).

Primary vertex counting

The primary vertex counting (PVC) method uses the vertices that have been reconstructed using the tracks in the CMS detector. For this method, a good primary vertex is defined to be one with 11 or more tracks. This requirement is sufficient to suppress spurious vertices [29], and results in better vertex resolution.

The PVC method is simple and robust, but suffers from mild nonlinearity effects when there are many collisions in a single bunch crossing. There are two competing effects. In one effect, primary vertices from two collisions occurring close to one another in space are merged, leading to an undercounting of vertices. In the other effect, the very large numbers of tracks associated with numerous collisions can produce spurious vertices, leading to overcounting. The precision with which these effects are understood falls short of the 1% level needed for luminosity studies. However, during vdM scans these effects are minimal because of the very low pileup, and so PVC is very useful as a validation tool for the vdM analysis in the measurement of beam-dependent parameters.

Forward hadron calorimeter

The HF luminosity measurement uses the separate readout described above, so the measurement can be performed at the full 40MHz bunch crossing rate. The back-end electronics upgrade during LS1 added new electronics using field-programmable gate array (FPGA) technology such that several features of the readout were separately programmable for luminosity histogramming, i.e., identifying and counting the readout channels. Although the whole HF is capable of being read out for luminosity use, only the two outer rings in η are used to ensure uniform occupancy and minimize minor nonlinearities expected from simulation.

The computation of the HF observable is based on the occupancy method (HFOC). In this method, the fraction of channel measurements above an analog-to-digital converter (ADC) threshold is used for each bunch slot in a configurable time window. The ADC threshold is set high enough to avoid most noise and as low as possible otherwise. Both the ADC threshold and the integration time of the histograms between readouts are configurable, but they were fixed during data taking in 2015 and 2016. The number of valid measurements is also stored, so the fraction of events with hits above threshold can be computed.

Pixel Luminosity Telescope

The PLT is a dedicated system for measuring luminosity using silicon pixel sensors, installed in Run 2, at the beginning of 2015. There are a total of 48 sensors arranged into 16 “telescopes”, eight at either end of CMS outside the pixel endcap. Each telescope contains three sensor planes which are arranged in a triplet that faces the IP. The sensors measure 8×8mm2, divided into 80 rows and 52 columns, although only the central region of the sensors is used to reduce the contribution from background. The PLT measures the rate of triple coincidences, where a hit is observed in all three planes, typically corresponding to a track from a particle originating at the IP. The overall mean rate for PLT is estimated using the fraction of events where no triple coincidences are observed (as described in Sect. 3) in order to avoid potential systematic effects from overlapping tracks being counted as a single hit.

Fast beam conditions monitor

The BCM1F measures luminosity and beam-induced background separately. It consists of a total of 24 sensors mounted on the same carriage as the PLT. Single-crystal diamond sensors are used with split-pad metallization. Each sensor has two readout channels to keep the overall occupancy low, given the experimental conditions in Run 2. The BCM1F features a fast readout with 6.25ns time resolution. The precise time measurement allows hits from collision products to be separated from beam-induced background hits, while the incoming background is separated in time from the outgoing collision products due to the position of BCM1F 1.8 m from the center of CMS.

Drift tube muon detector

The luminosity measurement based on the DT muon detector [20] is based on an efficient trigger on a low-background physics object: muons produced in the CMS barrel. Muon track segments from barrel muon DT stations are sent every bunch crossing to track finder hardware, where tracks are built and later used to generate first level triggers. The number of tracks in time windows of approximately 23 s is read out and stored in a database. These data are used to estimate luminosity. The rate of muons in the DTs is significantly lower than the rate for most other observables from other luminometers. Thus, there are not enough muon tracks during the vdM scans to provide a precise measurement of σvis, and so the system must be calibrated to the normalized PCC luminosity measurement. On the other hand, the muon candidate rate has been observed to be linear with luminosity and rather stable over time. The luminosity data of this system are integrated over all bunches.

Radiation monitoring system for the environment and safety

The Radiation Monitoring System for the Environment and Safety (RAMSES) is a monitoring subsystem of the unified supervisory CERN system [30, 31]. There are 10 ionization chambers filled with air at atmospheric pressure that are used as monitors installed in the CMS experimental cavern. They are sensitive to ionizing radiation and can monitor the ambient dose equivalent rate. Thus, they generate alarms and interlocks to ensure the safety of the personnel. This system is maintained and calibrated by the LHC radiation protection group.

While not designed as a luminometer, the two chambers with the highest rates (designated PMIL55X14 and PMIL55X15) have been used to produce a luminosity measurement with good linearity and stability over time. However, similarly to the DT luminosity measurement, the overall rates are too low for bunch-by-bunch measurements or extracting an absolute calibration during vdM scans. The RAMSES luminosity is thus calibrated to the normalized PCC luminosity measurement and is used as an additional measurement for assessing the luminometer stability with time.

Luminosity determination algorithms

Each bunch crossing gives rise to a certain number of pp interactions. In a given luminometer, each interaction results in some number of observables (e.g., hits, tracks, or clusters). If one averages over several unbiased measurements, the mean number of observables is

Nobservables=Nobservables/interactionNinteractionsNobservables/interactionμ, 2

where the average number of interactions per bunch crossing is denoted by μ, in keeping with the Poisson nature of the underlying probability distribution. Typically these observables are averaged over seconds or tens of seconds.

To measure the instantaneous luminosity, we use the fact that μ is proportional to the single-bunch crossing instantaneous luminosity Lb via:

μ=σLbνr, 3

where νr=11245.6Hz is the LHC revolution frequency during collisions, and σ is the total interaction cross section. At the LHC, Lb is typically expressed in units of Hz/μb1030cm-2s-1.

Two algorithms have been developed for extracting the instantaneous luminosity. One method is rate-scaling, where the raw rate of observables is scaled with calibration constants to the luminosity. Rearranging Eqs. (2) and (3), one can estimate the instantaneous luminosity using the average number of observables at a given time:

Lb=NobservablesNobservables/interactionνrσNobservablesνrσvis. 4

Luminosity is estimated from PCC, PVC, DT, and RAMSES data using the rate-scaling algorithm.

The second method (zero counting) uses the average fraction of bunch crossings where no observables in a detector are produced. This zero fraction is then used to infer the mean number of observables per bunch crossing. The principal advantage of the zero-counting method is that it is not affected by cases where two or more separate signals overlap in the detector and produce only one reconstructed observable.

Assuming that the probability of no observables in a single collision is p, then the probability of no observables seen in a bunch crossing with k interactions is thus simply pk. Averaged over a large number of bunch crossings, with the number of interactions per bunch crossing distributed according to a Poisson distribution of mean μ, the expected fraction of events with zero observables recorded, f0, can be expressed as:

f0=k=0e-μμkk!pk=e-μ1-p. 5

The logarithm of Eq. (5) is proportional to the mean number of pp interactions per bunch crossing, and hence to the Lb according to Eq. (3):

Lb=μνrσ=-lnf011-pνrσ-lnf0νrσvis. 6

The actual value of p does not need to be known beforehand, since it is effectively absorbed in σvis, although it could be extracted from the measured σvis value. The raw inputs from HFOC, PLT, and BCM1F are converted to luminosity using the zero-counting method.

Absolute luminosity calibration

Any luminometer requires an externally determined absolute calibration. Approximate σvis values can be obtained using Monte Carlo (MC) simulation, but these ultimately rely on theory, i.e., the inelastic pp cross section, and are not expected to be reliable at the percent level that represents the target accuracy for the CMS luminosity measurement. At the LHC, the precision of theoretical predictions for SM processes is typically limited by the knowledge of the parton distribution functions in the proton. Although methods independent of theoretical assumptions have been proposed at the expense of introducing correlations between low- and high-μ data-taking periods [32], a more precise and purely experimental method to determine the luminosity is based on the vdM scan technique, which is used in this paper.

Beam-separation scans are therefore performed to obtain calibrated σvis for the luminosity measurement. These were pioneered by Simon van der Meer at the ISR [10], extended by Carlo Rubbia to the case of a collider with bunched beams [11], and have been extensively used by all four major LHC experiments [12, 13]. The key principle of the vdM scan method is to infer the beam-overlap integral from the rates measured at different beam separations – provided the beam displacements are calibrated as absolute distances – as opposed to measuring the bunch density functions directly. The basic formalism is described in the following.

The van der Meer method

The instantaneous luminosity for a single colliding bunch pair in a colliding-beam accelerator is given by:

Lb=νrn1n2Aeff, 7

where n1 and n2 are the numbers of particles in each of the two bunches, and Aeff is the effective area of overlap between the bunches. In general, each of the bunches will be distributed in the plane transverse to the beam direction, in which case 1/Aeff can be replaced by an overlap integral of the bunch densities, i.e.,

Lb=νrn1n2ρ1(x,y)ρ2(x,y)dxdy, 8

where x and y represent the horizontal and vertical coordinates in the plane transverse to the beams, and ρ1 and ρ2 are the normalized two-dimensional density distributions for the two bunches. Here, we have integrated over time and the longitudinal coordinate z.

If one assumes that the bunch profiles can be factorized into terms depending only on x and y [10, 11], then ρi can be written as the product of one-dimensional density functions of the form ρi(x,y)=fi(x)gi(y) (i=1,2), and 1/Aeff can be written

1Aeff=f1(x)f2(x)dxg1(y)g2(y)dy1Weff1Heff, 9

where Weff and Heff are the effective width and the effective height of the luminous region. For the ideal case of Gaussian-distributed bunches with the same width in both beams and undergoing head-on collisions, Eq. (8) reduces to:

Lb=νrn1n24πσxσy, 10

where σx and σy are the root-mean-square (RMS) widths of the horizontal and vertical bunch profiles in either beam, respectively. In the case of round beams, σx=σyσbϵNβ/γ, where ϵN is the so-called normalized emittance, γ the relativistic Lorentz factor, and β corresponds to the value of the optical function β at the IP [33].

We designate the luminosity when the beams are displaced with respect to each other by an amount w in the x direction, or an amount h in the y direction, as L(w,h). As shown in Ref. [10], when a separation scan is performed in the x direction, in which w is varied in a systematic way from - to +, the effective width can be determined from:

Weff=f1(x)f2(x-w)dxdwf1(x)f2(x)dx=Lb(w,0)dwLb(0,0), 11

where common normalization factors have been canceled in the second step. Similarly, if a scan is performed in the y direction, the effective beam-overlap height is given by

Heff=g1(y)g2(y-h)dydhg1(y)g2(y)dy=Lb(0,h)dhLb(0,0). 12

For Gaussian-distributed bunches, the resulting scan curves, L(w,0) and L(0,h), are also Gaussian with RMS widths of Σx=Weff=2σx and Σy=Heff=2σy, yielding

Lb=νrn1n22πΣxΣy. 13

Equations (11) and (12) are quite general, and do not depend on the assumption of Gaussian-distributed bunches. Indeed, it is frequently the case that simple Gaussians do not provide an adequate description of the scan-curve data. In such cases, we use double-Gaussian functions of the form

f(x)=12π[ϵxσ1xexp(-x22σ1x2)+1-ϵxσ2xexp(-x22σ2x2)], 14

where ϵx is the fraction of the Gaussian with width σ1x. Normally the Gaussian with the smaller width σ1x is considered the core Gaussian, while the Gaussian with the larger width σ2x is used to fit the tails of the scan curve. Similar relations apply for the y coordinate. The effective value of Σi (i=x,y) is then given by

Σi=σ1iσ2iϵiσ2i+(1-ϵi)σ1i. 15

To calibrate a given luminosity algorithm, the absolute luminosity computed from beam parameters via Eq. (13) is used in conjunction with Eq. (3) to obtain

σvis=μvis2πΣxΣyn1n2, 16

where μvis is the visible interaction rate. In this analysis, μvis is taken as the arithmetic mean of the peak values from L(w,0) and L(0,h) in scans that are performed sufficiently close in time to minimize the impact of varying bunch distributions over the course of a fill. Equation (16) therefore provides a direct calibration of the visible cross section for each algorithm in terms of ΣxΣy and n1n2.

In the LHC, bunches typically cross at a small angle ϕ in the horizontal plane at IP 5. This introduces a reduction in the luminosity relative to the case of head-on collisions [1], given by:

LL0=1+σzσxtanϕ22-1/2, 17

where σx is the width of the luminous region in the crossing plane and σz is the width in the longitudinal direction. For typical LHC physics running conditions in 2016, ϕ140μrad, σx12μm, and σz8cm, and so the reduction from Eq. (17) is around 10% [34]. The vdM scans are typically carried out under special conditions, where ϕ=0, as described in the following. The values of σvis do not depend on the crossing angle.

Analysis of vdM scan data

While σvis does not depend on beam conditions, the LHC delivers beams under special conditions to improve the precision of measurements and to reduce systematic effects. The vdM filling schemes are characterized by a low number of colliding bunch pairs at IP 5 (nb=30–50). The bunches are widely separated from each other in the LHC orbit, to reduce the effect of afterglow (as discussed in Sect. 5.1). Special beam optics with β19m and transverse emittance of ϵN3.0μm are implemented to produce a relatively large bunch size of approximately σb=100μm. Large bunches reduce the impact of vertex reconstruction resolution in analyses where vertex positions are utilized. A crossing angle of 0 is used for collisions at IP 5 in vdM scans. To minimize the effect of potential nonlinear response in the luminometers, the target pileup is set to μ0.6, which is 1–2 orders of magnitude lower than typical physics fills. To achieve that goal, in addition to the large beam size, the beams have relatively low intensities, which typically begin at (8–9)×1010 protons per filled bunch, resulting in a total intensity of (3.5–4.0)×1012 per beam for 44 bunches.

The total beam intensities are measured with the DC current transformers (DCCT) [35], and the bunch currents measured with the fast beam current transformers (FBCT) [36], and cross-checked with the longitudinal density monitors (LDMs) [37, 38] and the beam quality monitors [39]. Because of the low beam intensity and low collision rate, the luminosity remains nearly constant over the course of time, in contrast to typical physics fills [9]. The beam orbit is monitored using two systems, the Diode Orbit and Oscillation (DOROS) beam position monitors (BPMs) [40] located near IP 5, and the BPMs located in the LHC arcs adjacent to CMS (referred to as “LHC arc BPMs”). The latter are transformed to a beam position at IP 5 using the LHC optics files that are centrally provided by LHC operators [41]. The orbit is also tracked using the movements of the luminous region at IP 5 based on the vertices reconstructed with the CMS tracker.

The vdM scan program at IP 5 consists of a series of x-y scan pairs. Figure 3 shows the progression of these scans in a calibration fill, with the beam displacement measured by the DOROS BPMs [40, 42].

Fig. 3.

Fig. 3

Relative change in the positions of beams 1 and 2 measured by the DOROS BPMs during fill 4954 in the horizontal (x) or vertical (y) directions, as a function of the time elapsed from the beginning of the program. The gray vertical lines delineate vdM, BI, or LSC scans

Typical scan sessions consist of at least three vdM scan pairs, with one scan in each of the transverse coordinates per pair. There are two at the start of the fill and another at the end of fill. In the absence of systematic effects, all scans are expected to produce compatible results. In each pair, the scans are typically performed first in the x and then in the y direction, although sometimes the pair is performed in the opposite order. In the vdM scans, the two beams are separated by up to 6σb, and scanned across one another in a sequence of 25 steps of 30s each to obtain a statistically significant measurement.

Dedicated length scale calibration (LSC) scans (described in Sect. 4.3.4), which are used to calibrate the distance by which the steering magnets displace the beams, are also performed typically close in time to the rest of the scans and using the same collision optics configuration. The LSC implemented at IP 5 is of a constant-separation type, in which the two beams are positioned at -2.5 and -1.5σb relative to nominal and moved together forward in steps of 1σb, maintaining the 1σb separation between the two beams, until they reach the +2.5σb point. Then, their positions are swapped, and they are moved together backward in -1σb steps back to -2.5σb. The scan is performed once in the x direction and once in the y direction, with a total of 10 steps of 60s in each direction. The LSC scans are performed with successive forward and backward displacements for multiple measurements under slightly different conditions in case there are compounding effects that limit precision. The transverse position of the luminous region is needed for this calibration and is measured using reconstructed primary vertices in CMS data.

To test the assumption of transversely factorizable bunch profiles in Eq. (9), four dedicated beam-imaging (BI) scans are performed, one for each beam and each transverse direction. One beam is kept fixed at its head-on position, while the other is moved and scanned in 19 steps from -4.5 to +4.5σb along x or y with a duration of 40s per step. Primary vertices are reconstructed, and their positions are then analyzed to perform a global fit to derive the transverse bunch density distributions of the beams (as discussed in Sect. 4.4.1). The BI scans are also analyzed as regular beam-separation scans. During both BI and regular vdM scans, the transverse bunch density distributions are also determined by simultaneously fitting the beam-separation dependence (“evolution”) on the luminosity and the luminous region position, orientation, and spatial width, as reflected in the reconstructed primary vertices (as discussed in Sect. 4.4.2).

The LHC conditions at IP 5 for the luminosity calibration fills discussed in this paper for 2015 and 2016 are summarized in Table 1.

Table 1.

Summary of the LHC conditions at IP 5 for the scan sessions in pp collisions in 2015 and 2016. The column labeled μ is the average pileup corresponding to Linit, the latter denoting the initial instantaneous luminosity. The columns corresponding to “No. of scans” indicate the total number of vdM, BI, and LSC scans that were performed in either transverse coordinate, counting only scans used for analysis

Fill s (TeV ) Date nb ϕ (μrad) β (cm) μ Linit (×1030cm-2s-1) No. of scans
vdM BI LSC
4266 13 Aug. 2015 30 0 1917 0.6 2.7 6 4 3
4945 13 May 2016 32 0 1917 0.6 2.5 2
4954 13 May 2016 32 0 1917 0.6 2.5 6 4 2

Pixel data are collected for PCC and for methods involving collision vertices using a zero-bias trigger, which collects data from five BCIDs with a total rate of approximately 20kHz. Figure 4 shows vdM scan data from PCC recorded in the fifth scan pair of the session in fill 4954. The fit function corresponds to the double-Gaussian formalism of Eq. (14), and the parameters are estimated by simultaneously fitting the PCC and PVC rate measurements. An additional constant term is included to estimate the background originating from noncollision sources. This function provides a good description of the data in a range that extends over nearly three orders of magnitude in rate (χ2/dof1 in Fig. 4). For other luminometers, background rates are either negligible (PLT and PVC) or estimated and subtracted (BCM1F and HFOC) prior to the beam parameter fit. Since the instantaneous luminosity is relatively low, any nonlinear effect has a negligibly small impact in any method. The beam-width parameters (Eq. (15)) measured using different luminometers are in excellent agreement, which is shown in Fig. 5 with comparisons of Aeff with the nominal PCC+PVC results.

Fig. 4.

Fig. 4

Example vdM scans for PCC for BCID 41, from the last scan pair in fill 4954, showing the rate normalized by the product of beam currents and its statistical uncertainty as a function of the beam separation in the x (left) and y (right) direction, and the fitted curves. The purple curve shows the overall double-Gaussian fit, while the blue, yellow, and green curves show the first and second Gaussian components and the constant component, respectively. All corrections described in Sect. 4.3 are applied. The lower panels display the difference between the measured and fitted values divided by the statistical uncertainty

Fig. 5.

Fig. 5

The two figures show comparisons of effective area (Aeff) of cross-check luminometers with respect to the nominal PCC+PVC for fills 4266 (upper) and 4954 (lower). The points are the ratio of the Aeff of the labeled luminometer to PCC+PVC. There are 25 Aeff values because there are five scan pairs with five BCIDs analyzed for each scan pair. The solid lines are the average of all the Aeff while the bands are the standard deviations. In both sets of data the average comparison is compatible with unity within or near the standard deviation

Although the accelerator parameters, such as bunch transverse sizes or intensities, vary during the course of a fill, such changes cancel in the calculation of σvis, which should remain invariant. This is shown in Fig. 6 for the measured σvisPCC as a function of time for vdM scans taken in fills 4266 and 4954. After including all the effects described in Sect. 4.3, σvisPCC=9.166±0.056(stat) and 8.429±0.029(stat) barns in 2015 and 2016, respectively, where the bunch-by-bunch fit uncertainty in Σx, Σy, and μvis is propagated to the measured σvisPCC per scan. Since these uncertainties are statistical in nature, they contribute to the scan-to-scan combination in an uncorrelated way. The assumption of factorizable proton bunch densities limits the level of accuracy in the luminosity scale inferred from Eq. (13). A common approach is thus adopted at the LHC that includes a dedicated tailoring of the proton bunch injection chain to minimize the emergence of non-Gaussian bunch density distributions [43]. Since the factorizability between the x and y distributions could impact the vdM scan result of the different IPs differently, CMS reconstructs the individual proton bunch densities during the BI and vdM scans, as described in Sect. 4.4.

Fig. 6.

Fig. 6

The measured σvisPCC, corrected for all the effects described in Sect. 4.3, shown chronologically for all vdM scan pairs (where 3 and 4 are BI scans) taken in fills 4266 (upper) and 4954 (lower), respectively. Each of the five colliding bunch pairs is marked with a different color. The error bars correspond to the statistical uncertainty propagated from the vdM fit to σvisPCC. The band is the standard deviation of all fitted σvisPCC values

Corrections to vdM scan data

Several systematic effects can change the measurement of σvis, and the following sections describe the measurement of these effects, the corrections used, and the resulting systematic uncertainty in σvis.

Adjustments to the bunch-by-bunch charge measurement are made to correct for spurious charge that is present outside the nominally filled part of the slot (Sect. 4.3.1). Then, we correct for potential sources of bias associated with the beam position monitoring at the scale of μm. We distinguish between “orbit drifts”, which we model with smooth, linear functions, and residual differences relative to the nominal beam positions, where corrections per scan step are assessed. Since both effects are time dependent, thereby biasing σvis incoherently, they are monitored continuously during each scan (Sect. 4.3.2).

Another source of correction originates from the electromagnetic interaction between charged particles in the colliding bunches (beam-beam effects); when the beams are displaced, rather than being head-on, a beam deflection and change in β may be induced. The former causes the beams to be more separated than the nominal value from LHC beam position estimates, whereas the latter influences the spatial distributions of proton bunches and thus the observed rate. The resulting corrections to σvis are evaluated at IP 5 [44, 45], and depend on the LHC optics, beam parameters, and filling scheme (as discussed in Sect. 4.3.3).

The vdM method requires an accurate knowledge of the beam separation. Possible differences in the absolute scale between the nominal beam separation produced by the steering of the LHC magnets, as used in Eqs. (11) and (12), and the actual separation are determined by using the LSC procedure (Sect. 4.3.4).

Beam current calibration and spurious charge

The LHC beam currents are measured by dedicated devices. The FBCT system is used to measure the current of individual bunches in 25ns bunch slots. The DCCT system provides a precise (0.2%) measurement of the total current for each of the two beams; since it is more precise than the FBCT sum, its scale is used to normalize the sum of the FBCT measurements.

Both the DCCT and FBCT measurements are sensitive to additional charges outside the actual colliding bunch. These components must be measured and subtracted. The LHC radio frequency (RF) cavities operate at 400MHz, so a single 25ns wide bunch slot contains ten 2.5ns wide “RF buckets”. Only one RF bucket in a given bunch slot is filled with protons, and, in principle, the other nine RF buckets are empty. Similarly, of the total 3564 bunch slots, only a predefined subset is filled, according to the filling scheme. In practice, however, a small amount of spurious charge is present in the nominally empty RF buckets and bunch slots, which should be subtracted from the n1 and n2 values in Eq. (13). The amount of “ghost” charge in the nominally empty bunch slots is included in the DCCT but not in the FBCT measurement, since the latter is insensitive to bunch charges below a certain threshold. The out-of-time (satellite) charge occupies RF buckets adjacent to the main bunch. As such, it can experience long-range interactions with the main bunch in the other beam and is visible in the FBCT measurement. The corrected value for nj (where j denotes the BCID) is therefore given by:

nj=nFBCTj1-fsatjjnFBCTjNDCCT1-fghost, 18

where fsatj represents the per-bunch correction due to the satellite bunch population and fghost is the correction for the ghost charge.

The spurious charge is measured by the LHC LDM system, which provides a precise longitudinal distribution of the beam charge with a time resolution of 90ps. The data from the LDMs for fills 4266 and 4954 indicate that both the ghost and satellite charges are small. The latter is estimated to be <0.1% for each of the two beams and is neglected. No particular time dependence for either beam is observed, and the resulting overall spurious-charge correction in σvis amounts to +0.2 and +0.3% in 2015 and 2016, respectively. This is applied as a correction to the beam currents in Eq. (16).

The ghost charge is also measured using the beam-gas imaging method [12, 46, 47], which compares the beam-gas rates in bunch crossings at IP 8 (the location of the LHCb detector) where only one beam contains protons, or where neither beam contains protons, leading to consistent results with the LDM measurement. The systematic uncertainty of 0.1% is assigned to cover the difference between the two estimates of the ghost contributions to the beam current.

Beam position monitoring

Although the LHC beam orbits are generally stable during a fill, even a small variation (either random or systematic in nature) in the beam positions during scans can significantly affect the resulting calibrations. The beam positions are measured primarily using the DOROS BPM system. The LHC arc BPMs, when possible, are used to confirm the stability of the orbits during the scan.

To measure the orbit drift, we use the beam position measurements in x and y in three 15s periods when the beams are nominally colliding head-on: immediately before and after each scan, as well as at the middle point of the scan, where the beams are also head-on. For each scan, a fit using a first-order polynomial is performed from the point before the scan to the middle point, and it is used to derive the correction for the first half of the scan. Similarly, a fit from the middle to the point after the scan is used to correct the second half of the scan. Figure 7 shows the measured positions along with the resulting fits. In general, the orbit drift during the 2015 and 2016 vdM scans is less than about 5μm for most of the scans. However, in the third scan of both series, the orbit drift was significant enough to shift σvis by approximately +1.0%. The corrections are derived using the average of the two BPM systems, and the largest deviation of the correction from each individual system from the nominal correction is taken as the value of the systematic uncertainty due to orbit drift. This is typically 0.1–0.2 % overall.

Fig. 7.

Fig. 7

Effect of orbit drift in the horizontal (upper) and vertical (lower) beam-separation directions during fill 4954. The dots correspond to the beam positions measured by the DOROS or LHC arc BPMs in μm at times when the beams nominally collide head-on and in three periods per scan (before, during, and after) represented by the vertical lines. First-order polynomial fits are subsequently made to the input from BPMs (dots) and are used to estimate the orbit drift at each scan step. Slow, linear orbit drifts are corrected exactly in this manner, and more discrete discontinuities are corrected on average

At each scan step, the actual beam separation can be also affected by systematic or random deviations of the beam positions from their nominal settings, which, in turn, impact the observed rate at each scan point. The magnitude of this potential bias is evaluated from consecutive single-beam orbit measurements at IP 5, provided by the DOROS BPMs and with a duration of a few seconds each. They are further corrected for the beam-beam effects (as discussed in Sect. 4.3.3) and the length scale (as described in Sect. 4.3.4) using the position of reconstructed vertices as the calibration target. The impact from beam-beam deflection at the location of the DOROS BPMs (zDOROS=±21.5m away from IP 5) is magnified by a factor of 1+tanπQx/yzDOROS/β, where Qx and Qy are the betatron tune values in the x and y directions [33]. Because these values are different, the resulting factors are 2.7 in the x direction and 2.8 in the y direction. The measurements from the DOROS BPMs are integrated over all bunches. Therefore, the observed beam-beam deflection may be overestimated because of the inclusion of noncolliding, nondeflected bunches. In this analysis, a reduction factor of 0.6 is thus applied on top of the geometric factor in both years, which is the approximate fraction of the total number of bunches in the vdM fills that collide at IP 5. The orbit drift, as described above, is also subtracted from the single-beam DOROS measurements before forming the actual beam separation. Finally, an additional length scale correction is made to DOROS data for each beam and in both of the two transverse directions. The calibration using vertices, both for DOROS and nominal LHC positions, determines only the average length scale for the two beams. The calibrations of each beam are also not necessarily the same for the two sets of data. Therefore, a final, relative calibration of the DOROS data is made to align each beam in both transverse directions to the scale of the LHC beams. Figure 8 shows the residual difference in beam separation in all y scans in 2015 and 2016 as well as the residuals per beam in a single scan, which shows symmetric behavior. The resulting impact on σvis is in the range -0.6 to +0.4 and -0.5 to -0.2%, with average values of -0.1 and -0.3%, in 2015 and 2016, respectively. Corrections are applied for each scan, and the uncertainty comes from the reduction factor in the beam-beam deflection correction at the location of the DOROS BPMs.

Fig. 8.

Fig. 8

The beam-separation residuals in y during all scans in fills 4266 (upper) and 4954 (lower) are shown on the left. The dots correspond to the difference (in terms of beam separation in μm) between the corrected beam positions measured by the DOROS BPMs and the beam separation provided by LHC magnets (“nominal”). The error bars denote the standard deviation in the measurements. The figures on the right show the residual position differences per beam between the DOROS BPMs and LHC positions for the first vdM scans in y in fills 4266 (upper) and 4954 (lower)

Beam-beam effects

We distinguish two types of beam-beam interactions that affect the vdM and BI scan measurements: coherent and incoherent beam-beam effects. The total correction originates from the combination of both effects, which affect the nominal beam separation (coherent) and the detector rate (incoherent) via the change of the beam shapes.

The closed orbits of the bunches in the scans are shifted coherently by the angular kick induced by their electromagnetic repulsion, resulting in an increase in the absolute beam separation. The size of this additional beam-beam deflection depends on the transverse beam size, bunch intensities, collision optics, and separation between the orbits of colliding bunches. It is calculated based on the Bassetti–Erskine formalism for the electric field of elliptically distributed bunches, as discussed in Ref. [48]. The orbit shift depends linearly on the separation for small nominal beam separations, reaches a maximum near 2σb (0.2mm in fills 4266 and 4954), and decreases nonlinearly towards zero at larger separations. Figure 9 (left) [44, 45] shows the resulting correction as a function of nominal beam separation, for the conditions during the scans in fill 4954 (Table 1). The beam-beam deflection correction increases the Σx and Σy values, impacting the σvis measurement by about +2.0 (+1.6)% in 2015 (2016).

Fig. 9.

Fig. 9

Calculated beam-beam deflection due to closed-orbit shift (left) and the multiplicative rate correction for PLT due to the dynamic-β effect (right) as a function of the nominal beam separation for the beam parameters associated with fill 4954 (first scan, BCID 992). Lines represent first-order polynomial interpolations between any two adjacent values

The incoherent effect corresponds to the change of the proton bunch density distribution functions ρ(x,y) at the IP due to deflection at the per-particle level. It causes a change in the effective β, and thus results in a change in the measured luminosity. This dynamic evolution of β is usually referred to as the “dynamic-β” effect. The correction for the dynamic-β effect is evaluated numerically by using a dedicated particle tracking program that calculates Aeff under different hypotheses [44, 45]. Considering the dynamic-β effect independently of the beam-beam deflection, we obtain the ratio of the detector rate as shown in Fig. 9 (right). At vdM conditions, the dynamic-β correction can be up to about -2% at large values of beam separation. Figure 9 shows the effect is typically larger at higher beam separation. In contrast to the beam-beam deflection, the dynamic-β correction thus decreases the original Σx and Σy values. The corresponding impact on the calculated σvis is about −1.7 (−1.4)% in 2015 (2016).

The total beam-beam correction (i.e., when both the beam-beam deflection and dynamic-β effects are included) results in an increase in the calculated σvis of about 0.3 (0.2)% in 2015 (2016) at IP 5. In addition, when considering further head-on collisions at the IP at the opposite side of the ring (IP 1 at ATLAS), the effect is approximated as a single-IP simulation but with shifted betatron tune values. The impact on σvis is enhanced by a factor of about two, leading to a total beam-beam correction of +0.6 (+0.4)% in 2015 (2016). The uncertainty in this calculation is dominated by the uncertainty in the betatron tune values, which was estimated taking into account the symmetric tune spread as well as the full shift due to head-on collisions at a second interaction point (in ATLAS at IP 1). These considerations translate into an uncertainty of 0.5 % in the corrected σvis  [44, 45].

Length scale calibration

In the canonical vdM formalism described in Sect. 4.1, it is implicitly assumed that the beam separation is perfectly known. Operationally, the nominal displacement of the beams at the IP is achieved based on a local distortion (bump) of the orbit using a pair of steering dipoles located on either side of the IP [49]. The size of the nominal separation is subject to potential uncertainty associated with the response of the steering dipoles themselves (e.g., magnet hysteresis) or lattice imperfection [41], i.e., higher multipole components in the quadrupoles located within those orbit bumps. For a given IP, there are four possible bumps, for the two possible displacement directions of the two beams.

An accurate calibration for the size of the bumps can be obtained using the CMS tracker. In particular, for small vertex displacements, the uncertainty in the reconstructed vertex position in x or y is 20μm for zero-bias collisions [29]. During LSC scans, the data for each separation distance contains several hundred thousand reconstructed vertices, yielding a position measurement with submicron precision.

The vdM scans described in Sect. 4.2 are typically done by moving the beams in equal steps in opposite directions. Since the two beams have independent length scales, the full separation correction is obtained from the mean of the length scale corrections per beam. Separate scans, wherein both beams are moved in steps in the same direction, are thus required to obtain the LSC. A more detailed description on the relationship between the calibration constant associated with the “offset” (i.e., the arithmetic mean between the transverse beam positions) and the observed quantities during LSC scans can be found in Ref. [12]. Here, for each scan step, the centroid of the luminous region is measured as the mean from a Gaussian fit to the observed vertex positions. A calibration constant for each transverse direction is extracted with a first-order polynomial fit to the difference between the measured mean position and the nominal offset as a function of the latter. This constant corresponds to the average calibration of the bumps of the two beams. It is then applied as a scale factor to correct the nominal beam displacement.

The nominal offset is also affected by the random and systematic beam position deviations described in Sect. 4.3.2. The beam positions at each step are monitored using DOROS BPMs. We estimate the arithmetic mean of the measured step sizes as a good representation of the nominal settings, after excluding outlier step sizes based on an iterative procedure. The difference of the remaining step sizes from the mean is used to correct the nominal offsets, and their standard deviation is the uncertainty due to beam position deviations. The correction improves the quality of the first-order polynomial fits and the forward-to-backward scan agreement. Consistent results are also found using the LHC arc BPMs to derive the correction for beam position deviations.

The fit results for the x and y LSC scans are shown in Fig. 10 for fill 4945. The difference between the measured displacement of the beam centroid and the nominal displacement of the beams, corrected for the estimated beam position deviation, is plotted as a function of the latter. In all cases, the data are well described by first-order polynomial fits with calibration constants differing on average from zero by -0.3 and -0.8% in the horizontal plane in 2015 and 2016, respectively, and by -0.1 and -0.5% in the vertical plane. The combined correction to the visible cross section is (-0.4±0.2) and (-1.3±0.3)%. The total uncertainty, equal to the uncertainty contributions from the x and y planes added in quadrature, includes the statistical uncertainty in the first-order polynomial fits (<0.1%), the variation between the two scan directions and the different scans (0.1%), a tracker alignment uncertainty (<0.1%), and the uncertainty from the estimated beam position deviations (0.1–0.2%).

Fig. 10.

Fig. 10

Fits to LSC forward (purple) and backward (green) scan data for the x (upper) and y (lower) LSC scans in fill 4945. The error bars denote the statistical uncertainty in the fitted luminous region centroid

Transverse factorizability

The use of the vdM scan technique to measure Aeff relies on the assumption that the proton bunch density functions are factorizable into x- and y-dependent components, as described in Sect. 4.1. If this condition is not met exactly, the measurements of Aeff and σvis will be biased. To correct for this potential bias, the bunch density distributions are measured independently with two methods, which are used in a combined way to evaluate Aeff. In both methods, primary vertices are reconstructed from tracks measured in the CMS silicon tracker.

Beam-imaging method

In the BI method [50, 51], the distributions of reconstructed vertices during BI scans are used to obtain an image of the transverse bunch profiles integrated over the scanning direction. A primary vertex resolution comparable to or smaller than the transverse beam sizes is necessary to extract the beam profiles from the measured distributions. The two-dimensional distribution in x and y of the reconstructed vertices depends on the overlap of the bunch density functions, their transverse separations Δx and Δy, and the vertex resolution V of the CMS tracker system as:

Nvtx(x,y;Δx,Δy)ρ1(x,y)ρ2(x+Δx,y+Δy)V. 19

The combination of the vertex distributions from all steps of the BI scan in the x direction is approximated as:

Δx=-4.5σb+4.5σbNvtx(x,y;Δx,Δy)[ρ1(x,y)ρ2(x+Δx,y)d(Δx)]V=ρ1(x,y)(Mxρ2)(y)V. 20

Here, (Mxρ2)(y)=ρ2(x,y)dx denotes that the proton bunch density of the second beam appears marginalized in the direction of the scan. This results from the assumption that the step size is small enough with respect to the width of the bunch densities, so we can replace the sum over discrete scan points with a continuous integral over Δx. This two-dimensional vertex distribution can be exploited to constrain the transverse correlations of the bunch density of the first beam.

Combining four such vertex distributions accumulated during the BI scan set, we reconstruct the two-dimensional proton bunch densities of the two beams from a simultaneous fit. This requires knowledge of the primary vertex resolution, which is modeled with a two-dimensional Gaussian function. Convolving with the primary vertex resolution is then analytically possible for bunch density models built from Gaussian functions.

Models for the proton bunch density are built from Gaussian distributions parameterized with an additional correlation parameter ϱ:

gj(x,y)=12πσjxσjy1-ϱj2exp(-12(1-ϱj2)×[x2σjx2+y2σjy2-2ϱjxyσjxσjy]), 21

where j indicates the beam number (j=1 or 2). More complicated models are constructed with sums of these individual correlated Gaussian distributions. Distributions with a wide tail are better described by adding a Gaussian component with a small weight and a large width. Distributions with a flattened central part can be modeled with an additional component with a small negative weight and a narrow width. Typically, both nonzero correlation parameters and different widths are required to describe the nonfactorizability observed in data.

The best description of the BI data collected in 2015 and 2016 for the five bunch crossings used is achieved consistently with a sum of three Gaussian distributions, where the narrow component has a negative weight:

ρj(x,y)=-wj,1gj,1(x,y)+wj,2gj,2(x,y)+(1+wj,1-wj,2)gj,3(x,y). 22

Figure 11 shows the two-dimensional pull distribution, i.e., (Ndatavtx-Nfitvtx)/σdata, and the one-dimensional projections for the vertex distributions collected in the BI scan where the first beam is moved vertically for one bunch crossing in fill 4954. In these fits, the effects from the beam-beam deflection and dynamic-β are included in the positions of the reconstructed vertices and as per-vertex weights, respectively, whereas the impact of orbit drift is negligibly small.

Fig. 11.

Fig. 11

Example of the pull distributions of the fit model of Eq. (22) with respect to the vertex distribution that constrains beam 2 in the y direction recorded in fill 4954. The upper plot shows the two-dimensional pull distributions, and the lower plots show the per-bin pulls averaged over the same radial distance (lower left) or angle (lower right). The error bars in the lower plot denote the standard error in the mean of the pulls in each bin. The fluctuations observed in the radial projection of the residuals are included in the uncertainty estimation

The value of Aeff can then be calculated from an integration of the overlap of the bunch densities directly (i.e., Aeff=ρ1(x,y)ρ2(x,y)dxdy). This is compared to the value of Aeff obtained from an MC simulated vdM scan pair generated with the reconstructed bunch densities as input, and analyzed with the vdM method (i.e., Aeff=1/(2πΣxMCΣyMC)). The difference between the two values yields the bias of the vdM results, and is applied as a correction to σvis values. The bias is computed separately for each bunch crossing, and the results are shown in Fig. 12. The values for the estimated bias are averaged, resulting in a correction of +1.3 (0.9)% in σvis for 2015 (2016) because of the assumption of x-y factorization.

Fig. 12.

Fig. 12

Factorization bias estimated from the fits to the BI bunch-by-bunch data in fills 4266 (upper) and 4954 (lower). The error bars denote sources of uncertainty (statistical and systematic), added in quadrature, in the factorization bias estimates

To estimate the uncertainty in the measured bias, the MC simulation of the vdM scans is repeated multiple times and the RMS of the resulting biases is 0.1% for both years, which is considered as the statistical uncertainty in the vdM scans. Additionally, a systematic uncertainty is evaluated with a closure test: simulated models are constructed by randomly drawing parameters of the fit model in Eq. (22). These are used to simulate MC pseudo-experiments by generating BI scan data, which are then fitted with the same model and procedure. Comparing simulated models with fit quality and fitted correction values similar to the data fits, the bias obtained from the bunch densities reconstructed from the fit agrees well on average with the true bias of the simulated model. The RMS of the distributions of deviations is 0.5 % for both years. We assign this RMS as the systematic uncertainty.

Luminous region evolution

In this method, which was inspired by Ref. [13], the luminosity and luminous region geometry are used to reconstruct the bunch density distributions in three dimensions and as a function of time. Using single-beam parameters, described in the following, bunch profiles are then generated for simulated vdM scans and treated as genuine vdM scan data. Similar to the BI method, the impact of factorization is extracted by comparing the “measured” luminosity extracted from the one-dimensional vdM simulated bunch profiles with the “true” luminosity from the computed four-dimensional (x, y, z, t) overlap integral of the single-bunch distributions. The luminous region is modeled by a three-dimensional ellipsoid whose parameters (nine in total) are extracted from an unbinned maximum likelihood fit of a three-dimensional Gaussian function to the spatial distribution of the primary vertices [29]. The vertex resolution is determined from data as part of the fitting procedure.

The bunch profiles ρj(x,y,z), parameterized per beam j, are the sum of three individual Gaussian distributions gj,13(x,y,z) with common mean, but arbitrary width and orientation parameters (referred to as “bunch parameters” in the following):

ρj(x,y,z)=wj,1gj,1(x,y,z)+wj,2gj,2(x,y,z)+(1-wj,1-wj,2)gj,3(x,y,z). 23

The overlap integral of Eq. (23) is evaluated at each scan step to predict the true luminosity and the geometry of the luminous region for a given set of bunch parameters. In this calculation, we consider the impact of beam-beam effects, LSC, and orbit drifts. The bunch parameters are then adjusted according to a χ2 minimization procedure to determine the best-fit centroid position, orientation, and the widths (corrected for the primary vertex resolution) of the luminous region measured at each step of a BI or vdM scan. An example of a fit to the PCC luminosity and luminous region geometry is illustrated in Fig. 13 for one of the horizontal scans in fill 4954 and a subset of the three-dimensional ellipsoid parameters. One of the four figures shows the variation in the beam width in y during the x-separation beam scan, which is indicative of nonfactorization. The goodness of fit is better than χ2/dof=1.8 for both years, with some systematic deviations being apparent mainly in the tails of the scan. The fits are repeated by substituting PLT as the luminosity input, but no particular dependence is seen.

Fig. 13.

Fig. 13

Beam-separation dependence of the luminosity and some luminous region parameters during the first horizontal vdM scan in fill 4954. The points represent the luminosity normalized by the beam current product (upper left), the horizontal position of the luminous centroid (upper right), and the horizontal and vertical luminous region widths (lower left and right). The error bars represent the statistical uncertainty in the luminosity, and the fit uncertainty in the luminous region parameters. The line is the result of the three-Gaussian (g1+g2±g3) fit described in the text. In all cases, the lower panels show the one-dimensional pulls

This procedure is applied to all (i.e., BI and vdM) scans in fills 4266 and 4954, and the results are summarized in Fig. 14. The σvis extracted from the standard vdM analysis with the assumption that factorization is valid is smaller by 0.6–1.1 (0.2)% than that computed from the reconstructed single-bunch parameters in fill 4266 (4954). Similar to the evaluation in the BI method, the uncertainty amounts to 0.6 %. This uncertainty is dominated by the standard deviation in simulation-driven closure tests, and includes the fit uncertainty in data and the contributions from beam-beam effects, length scale, and orbit drift. These observations are thus consistent with the ones obtained in Sect. 4.4.1 in terms of absolute magnitude during the BI scans. The two results are combined to produce the final correction in σvis of +(0.8--1.3±0.5) and +(0.6±0.5)% in 2015 and 2016, respectively. The final corrections retain the time evolution derived uniquely from the luminous region evolution method.

Fig. 14.

Fig. 14

Ratio of the σvis evaluated from the overlap integral of the reconstructed single-bunch profiles in two (BI method) or three (luminous region evolution) spatial dimensions to that determined by the vdM method, assuming factorization, and their combination. The central values are displayed as points or with a line while the corresponding full uncertainties are shown as hatched areas. Different methods (including the combination) are color coded. Each point corresponds to one scan pair in fills 4266 (left) and 4954 (right). The statistical uncertainty is shown by the error bars

Rate corrections under physics running conditions

The calibration scans described in the previous sections are performed with a small number of well-separated proton bunches with low bunch intensity. In contrast, during nominal conditions, the collision rate is generally maximized to produce large data sets for physics measurements and searches. This section describes the corrections that are applied to uncalibrated luminometer rates to ensure that the final luminosity values are accurate. These corrections, summarized in Table 2 for 2016, compensate for out-of-time pileup, efficiency, and nonlinearity effects for each individual luminometer.

Table 2.

Summary of the rate corrections under physics running conditions in 2016 applied separately to each luminometer. For HFOC, two distinct sources of out-of-time pileup corrections are provided. In the first and second columns, the vdM calibration condition and the relative agreement of the luminometers in terms of Aeff relative to PCC during fill 4954 are given, respectively. The DT luminosity is also corrected for a very small additional muon rate from beam halo and cosmic sources, which is treated as a constant per fill

vdM calibrated vdM calibration agreement to PCC (%) Out-of-time pileup corrections (%) Efficiency corrections (%) Nonlinear response (%)
PCC Yes 0–4 1
DT No
HFOC Yes 0.2 0–15, 1–5 1 0–10
PLT Yes 0.1 0–10 −0.2 to +1.4/(Hz/μb)
PVC Yes <0.1
RAMSES No

Out-of-time pileup corrections

The measurements in most detectors have out-of-time pileup contributions that do not arise from the in-time pp collision within the 25 ns window of the bunch crossing. Ideally, these contributions should be subtracted from all bunch crossings before the total instantaneous luminosity is computed. There are generally two types of effects that are considered: spillover of electronic signals and real additional response from material activation. These are denoted as type 1 (T1) and 2 (T2) afterglow, respectively.

The T1 afterglow generally only impacts the following bunch crossing because electronic signals tend to decline exponentially and hence two bunches later (50ns) the signal is again below threshold. The T1 contribution in bunch n+1 from bunch n is proportional to Lb(n). Thus, the model for the correction is:

Lb,corr(n+1)=Lb,uncorr(n+1)-αT1Lb,corr(n), 24

where αT1 is detector dependent and sometimes time dependent; αT1 ranges from 0.005 for BCM1F to 0.02 for HFOC to as large as 0.09 for PCC.

In contrast, T2 afterglow tends to impact all bunch crossings, because the half-life of the activated material can be longer than several bunch crossings. The response can be modeled with a single- or double-exponential distribution. The impact of T2 afterglow varies by filling scheme and by detector. In fills where nb is low and where the bunches are well separated, the T2 corrections are very small and often completely negligible, as is the case by design in the vdM calibration fills. When LHC fills contain several hundred bunches, the corrections start to contribute at the percent level in most bunches. With maximally full filling schemes, the corrections can be up to about 4 (15)% for PCC (HFOC).

Although there are clearly two distinct components, a combined (T1 and T2) model can be constructed that gives the response for a specific bunch crossing, accounting for contributions from all other 3563 bunch crossing slots. This model is referred to as the single-bunch response (SBR). The SBR for HFOC luminosity is taken directly from data in a reference fill with nb=2 for approximately the first half of the bunch crossings, and the bunches in the second half are smoothly extrapolated using an exponential model. The SBR is normalized to Lb(n) and it is then subtracted from all other bunch slots. This procedure is repeated for all bunch crossings.

After the corrections from the SBR are applied, empty bunch slots, where there are no collisions, should have a rate of zero. For PCC, the SBR is determined by optimizing αT1, which is time dependent and measured in intervals of about 20min, and the parameters of the exponential used for T2 corrections, such that there is minimal residual rate in the noncolliding bunch slots. Figure 15 shows per-bunch data in a fill from 2016 before and after the afterglow corrections for PCC are applied.

Fig. 15.

Fig. 15

The upper plot shows the instantaneous luminosity measured from PCC as a function of BCID before (filled blue points) and after (open red points) afterglow corrections are applied for each colliding bunch. The upper panel shows a subset of bunch crossings colliding at IP 5, and the lower panel shows empty bunch crossings (the scale is different in the two panels to show differences more clearly). The open red points in the lower panel lie close to 0, indicating that any residual PCC response is small in empty bunch slots. The lower plot shows the estimated residual T1 and T2 afterglow as a function of time during the full range of 2016 data for both PCC and HFOC, which use the same afterglow subtraction methodology

The empty bunch slots are also used to estimate the residual afterglow after the full set of corrections is applied. The corrected rate in the first empty bunch slot after a colliding bunch slot is used to estimate the residual T1 response. Likewise, the 2nd to 30th empty bunch slots are used to estimate the residual T2 effect. This procedure is performed for the entire 2015 and 2016 data sets for PCC and HFOC luminosity measurements. A window covering all residuals over the course of each data set is used as the systematic uncertainty in the final corrections. The resulting uncertainty for PCC in the two corrections is 0.30.1 (0.30.3)% in 2015 (2016).

These types of per-bunch luminosity corrections are applied for PCC, HFOC, and BCM1F, whereas PLT is almost completely background free and no such correction is needed. Since the DT and RAMSES measurements integrate over all bunch crossings, out-of-time pileup corrections can only be applied on average to the integrated rates. For DT these amount to 0–1%, while no corrections are applied to RAMSES.

A second type of T1 afterglow affects the HFOC luminosity. This is the case where the afterglow from a preceding bunch and the signal from the current bunch are both under the threshold to be counted as a hit, but their sum exceeds the threshold. This effect is referred to as the “bunch train effect”, because it affects only active bunches preceded by other active bunches (that is, bunches within a train, as opposed to “leading” bunches at the beginning of a train). The method previously described for estimating T1 afterglow does not include this contribution. This effect is measured in a dedicated study comparing the double ratio of the leading bunch in a train relative to the second bunch for HFOC divided by the same ratio for PCC. A single correction model with magnitude 1–5%, linearly increasing with instantaneous luminosity, is determined utilizing most valid data from 2016.

Efficiency corrections

Radiation damage can affect the detector response by reducing efficiency, increasing noise, or both. Noise is typically a small effect for most luminometers, but reduced response in detectors due to radiation damage can have significant (percent-level) effects, and so corrections are required. Corrections are measured against a stable benchmark relative to the performance at or near the vdM scans, and are applied to σvis. Shifts of 0–10% in detector response in the PLT in 2016 are corrected using RAMSES as a benchmark, whereas the impact of radiation damage on the HF efficiency is corrected using a parameterization derived from a model of aging. An HF efficiency correction of 1% is derived by measuring the average energy deposits in the HF in events characterized by the presences of Zbosons that decay to two muons with large transverse momentum.

A further efficiency correction is necessary for the PCC measurement. The pixel detector has a static internal memory buffer for data storage before the trigger decision is taken. When the buffer is filled, the oldest data overflows and are lost. This effect is proportional to the total instantaneous luminosity, and it can be estimated by studying the frequency of missing pixel clusters in otherwise well-reconstructed tracks [29]. In 2016, the effect was 1.0% at 1.4×1034 cm-2s-1. A correction proportional to total instantaneous luminosity is applied, and the total impact on integrated luminosity is 0.2%. Since the total luminosity in 2015 is substantially lower, no correction is applied. The PCC also has very small noise corrections.

Nonlinear response

In the absence of out-of-time pileup, the PCC luminosity is expected to be linear, according to simulations, so no corrections are applied. Moreover, the ratios of PCC to both DT and RAMSES luminosity measurements are highly compatible as a function of the instantaneous luminosity without any corrections. The HFOC response in 2015 and 2016, on the other hand, exhibits significant nonlinearity compared to the other luminometers. The main source of nonlinearity is the uncalibrated ADC-to-charge conversion applied at the time of data taking. Data from fill 5416, which exhibit a wide range of instantaneous luminosity, are used to model the correction for HFOC with a fourth-order polynomial. This smooth function extrapolates to the σvis calibration at low pileup within uncertainty. This single model is used to correct the nonlinear behavior of HFOC (0–10% higher response when compared to PCC) throughout 2016.

As described in Sect. 3, nonlinearity corrections are also needed for PLT. The corrections are modeled with a first-order polynomial. The parameters are time dependent, because of changes in the PLT operating conditions during the course of 2015 and 2016. These corrections, amounting to −0.2 to +1.4/(Hz/μb), are derived by comparing with RAMSES data in five different periods.

Detector stability and linearity

After the rate corrections are applied (as discussed in Sect. 5), comparisons between different luminometers are performed to assess remaining systematic effects impacting the luminosity measurement. Since PCC is expected to suffer the least from nonlinearities, as described in Sect. 2.1, once out-of-time pileup effects are corrected, PCC is the preferred luminometer for these data sets in the following estimates for stability and potential nonlinearity.

Upper bounds on stability

One measurement of potential instability in the PCC luminosity comes from intrinsic monitoring (i.e., comparing rates from different sections/parts of the subdetector over time). In a perfectly stable system, the fractional rates among different subcomponents would exhibit no variation with time. Figure 16 shows the result of applying out-of-time pileup corrections (as discussed in Sect. 5.1) separately to each pixel layer or disk. The sum of the per-region corrections matches the total, nominal correction made for all PCC regions to better than 0.1%. After the corrections are applied, the relative rates are quite stable over the course of 2015 and 2016. This is also shown in Fig. 16, where the relative PCC rates over time are simultaneously fit to a first-order polynomial.

Fig. 16.

Fig. 16

The relative contribution to the total number of observed pixel clusters from the four regions of the pixel detector used in the luminosity measurement (barrel layers 2 and 3, and inner and outer forward pixel disks), as a function of time throughout 2016. The lines represent first-order polynomial fits to the relative contributions from each region

However, this method cannot detect global shifts in σvis, and so it is crucial to make comparisons with completely independent systems. With multiple independent systems available for comparison, luminometers displaying brief periods of instability can be clearly identified. The cross-detector comparison is repeated for the entire data set for each year to detect periods where a single luminometer experiences transient effects (e.g., data quality issues, some detector components off, anomalous signals, etc.). Figure 17 shows the ratio of the luminosity measurements for different pairs of detectors throughout 2016, highlighting (in red) periods where the ratios significantly deviate from unity and so the associated data are invalidated.

Fig. 17.

Fig. 17

The luminosity measurements from PCC, HFOC, and RAMSES are compared as a function of the integrated luminosity in 2016. Comparison among three luminometers facilitates the identification of periods where a single luminometer suffers from transient stability issues. The ratios that are plotted in red contain invalidated data. The dashed line delineates the vdM calibration (fill 4954)

After the exclusion of invalidated data, which amount to 5% for each luminometer, the remaining input from different luminometers is used to assess an upper limit on the stability of the luminosity. PCC measurements are valid for 98.3 (94.3)% of the data set in 2015 (2016). The rest of the luminosity is provided by the next most stable luminometer, which is RAMSES (HFOC) for 2015 (2016). The primary luminosity, which is PCC or luminosity from the next most stable detector when PCC is unavailable, is compared with the next-best available luminometer (secondary). In Fig. 18, the latter is selected using the lowest standard deviation in the ratio relative to PCC over fixed time intervals of approximately 20min each. The position of the mean shows the agreement between the luminometers on the integrated luminosity. The width reflects stability effects, as well as residual statistical uncertainty in the luminosity measurement in each interval. From the distribution over the course of each year, the width is an upper limit on the uncertainty due to time dependencies in the luminometers. For 2015 (2016) a systematic uncertainty due to detector stability of 0.6 (0.5)% is derived.

Fig. 18.

Fig. 18

The ratio of the primary (best available) to secondary (next-best available) luminosity as computed in time windows of approximately 20min each. The left plot shows the 2015 results (principally PCC/RAMSES), and the right plot shows the 2016 results (principally PCC/HFOC). Each entry is weighted by the integrated luminosity for the time period

Time dependence of linearity

We make use of two methods for assessing the detector linearity. The primary method compares the ratio of the instantaneous luminosity from two luminometers per fill as a function of the instantaneous luminosity, which is estimated from the numerator. A first-order polynomial fit is performed and the slope is extracted. The slopes per fill are then studied as a function of time. No significant deviation over time is observed between DT/PCC or RAMSES/PCC and HFOC/PCC, DT/PCC, or RAMSES/PCC in 2015 and 2016, respectively.

To estimate the uncertainty, the fitted slopes are weighted according to the per-fill integrated luminosity. The mean values deviate slightly from 0, and the largest deviation is the systematic uncertainty in the linearity of PCC luminosity. Figure 19 shows the summary of these slopes for 2015 and 2016 at s=13TeV both for the whole year, and for subsets of each data set with equal luminosity. The largest average slope is 0.26 (0.08)%/(Hz/μb) in 2015 (2016), which translates into a 0.5 (0.3)% uncertainty in the integrated luminosity of the 2015 (2016) data set, where the average Lb is approximately 2.0 (3.3) Hz/μb.

Fig. 19.

Fig. 19

Linearity summary for 2015 (upper) and 2016 (lower) at s=13TeV. The slopes are plotted for each detector relative to PCC. The markers are averages of fill-by-fill slopes from fits binned in roughly equal fractions of the total integrated luminosity through the year. The error bars on the markers are the propagated statistical uncertainty from fitted slope parameters in each fill, which are weighted by integrated luminosities of each fill. The dashed lines and corresponding hatched areas show the average from the entire data set and its uncertainty

The alternative method makes use of the entire data set throughout the year, and extracts a single relative slope with a first-order polynomial fit. To remove effects from variations in the absolute luminosity scale over time, the per-fill ratios are shifted such that their extrapolation at zero luminosity is unity. The results are consistent with the primary method described above.

Total luminosity correction and uncertainty

For each data set, final rate corrections and final calibrations are applied to data in small time windows of 218 LHC orbits, approximately 23 s. All the measurements are summed to derive a total integrated luminosity measurement. The contributions to the systematic uncertainty in the integrated luminosity are divided into two general categories:

  • “normalization” uncertainty in the absolute luminosity scale, σvis, determined from the vdM scan procedure

  • “integration” uncertainty associated with σvis variations over time (stability) and pileup (linearity and out-of-time pileup corrections).

The magnitudes of the corrections applied to the absolute normalization from the vdM calibration are listed in Table 3, and Table 4 summarizes the sources of uncertainty. The dominant sources of normalization uncertainty are associated with the beam position monitoring (as discussed in Sect. 4.3.2), transverse factorizability (as explained in Sect. 4.4), and beam-beam effects (as described in Sect. 4.3.3).

Table 3.

Summary of the BCID-averaged corrections to σvis (in %) obtained with the vdM scan calibrations at s=13TeV in 2015 and 2016. When a range is shown, it is because of possible scan-to-scan variations. To obtain the impact on σvis, each correction is consecutively included, the fits are redone following the order below, and the result is compared with the baseline. The impact from transverse factorizability is obtained separately (as discussed in Sect. 4.4)

Source Impact on σvis (%)
2015 2016
Ghost and satellite charge +0.2 +0.3
Orbit drift +0.6 to +1.0 +0.2 to +1.0
Residual beam position corrections -0.6 to +0.4 -0.5 to -0.2
Beam-beam effects +0.6 +0.4
Length scale calibration −0.4 −1.3
Transverse factorizability +0.8 to +1.3 +0.6

Table 4.

Summary of contributions to the relative systematic uncertainty in σvis (in %) at s=13TeV in 2015 and 2016. The systematic uncertainty is divided into groups affecting the description of the vdM profile and the bunch population product measurement (normalization), and the measurement of the rate in physics running conditions (integration). The fourth column indicates whether the sources of uncertainty are correlated between the two calibrations at s=13TeV

Source 2015 (%) 2016 (%) Corr
Normalization uncertainty
Bunch population
Ghost and satellite charge 0.1 0.1 Yes
Beam current normalization 0.2 0.2 Yes
Beam position monitoring
Orbit drift 0.2 0.1 No
Residual differences 0.8 0.5 Yes
Beam overlap description
Beam-beam effects 0.5 0.5 Yes
Length scale calibration 0.2 0.3 Yes
Transverse factorizability 0.5 0.5 Yes
Result consistency
Other variations in σvis 0.6 0.3 No
Integration uncertainty
Out-of-time pileup corrections
Type 1 corrections 0.3 0.3 Yes
Type 2 corrections 0.1 0.3 Yes
Detector performance
Cross-detector stability 0.6 0.5 No
Linearity 0.5 0.3 Yes
Data acquisition
CMS deadtime 0.5 < 0.1 No
Total normalization uncertainty 1.3 1.0
Total integration uncertainty 1.0 0.7
Total uncertainty 1.6 1.2

The dominant sources of integration uncertainty arise from the linearity and stability of the primary relative to secondary luminosity measurements over the course of each year (as discussed in Sect. 6). In addition, the subleading systematic uncertainty due to out-of-time pileup corrections is considered for the PCC method since it is primarily used for the luminosity estimate.

Several sources of normalization uncertainty are considered to be correlated for the years studied because the scan procedures and analysis methodology are identical between the two vdM calibrations. The sources of the normalization uncertainty that are not correlated between the two vdM programs, and are partly statistical in nature, are the orbit drift, along with the scan-to-scan and bunch-to-bunch variations in the measured σvis. The latter are collectively referred to as “other variations in σvis ” in Table 4.

Among the sources of integration uncertainty, the afterglow corrections are treated identically in the two data sets, and so this source of systematic uncertainty is correlated. The estimate of the uncertainty due to linearity is considered to be correlated, since it is derived from the PCC linearity in both years. On the other hand, the stability assessment is based on cross-detector comparisons. Although PCC is the primary luminometer in each data set, the secondary luminometer is different for each year. Since the source of instability cannot be assessed and contains time-dependent features, the uncertainty is not correlated.

The tool used for providing luminosity values to physics analyses applies the corrections to the raw luminosity values using the average per-bunch luminosity, rather than the individual bunch-by-bunch values. This potentially introduces an error in the case where these corrections include a nonlinear term and the bunch-by-bunch luminosity varies significantly among bunches. We evaluated the effect of this approximation on 2016 data, and found that the overall impact on the integrated luminosity was <0.1%.

Finally, the quantity measured by the luminometers is the luminosity delivered to CMS; however, the quantity of interest to most physics analyses is the luminosity corresponding to the data actually recorded by the CMS DAQ system. These are related by the deadtime, as obtained from the trigger and clock system of CMS [21]. In 2015 this measurement was affected by an algorithm issue in the trigger system and has an uncertainty of 0.5%, but this problem was resolved before data taking began in 2016, so in 2016 the impact is negligible (<0.1%) and uncorrelated with 2015.

When applying the vdM calibration to the entire periods, the total integrated luminosity is 2.27 fb-1 with a relative precision of 1.6% in 2015, and 36.3 fb-1 with a relative precision of 1.2% in 2016. The combined 2015+2016 luminosity measurement has a precision of 1.2%, which is the same as the 2016 precision since it is the significantly larger data set and the precision in 2015 is similar.

Summary

The luminosity calibration using beam-separation (van der Meer, vdM) scans has been presented for data from proton–proton collisions recorded by the CMS experiment in 2015 and 2016 when all subdetectors were fully operational. The main sources of systematic uncertainty are related to residual differences between the measured beam positions and the ones provided by the operational settings of the LHC magnets, the factorizability of the transverse spatial distributions of proton bunches, and the modeling of effects on the proton distributions due to electromagnetic interactions among protons in the colliding bunches. When applying the vdM calibration to the entire data-taking period, the relative stability and linearity of luminosity subdetectors (luminometers) are included in the uncertainty in the integrated luminosity measurement as well.

The resulting relative precision in the calibration from the vdM scans is 1.3 (1.0)% in 2015 (2016) at s=13TeV; the integration uncertainty due to luminometer-specific effects contributes 1.0 (0.7)%, resulting in a total uncertainty of 1.6   (1.2)%; when applying the vdM calibration to the entire periods, the total integrated luminosity is 2.27 (36.3)fb-1.

The final precision is among the best achieved at bunched-beam hadron colliders. Advanced techniques are used to estimate and correct for the bias associated with the beam position monitoring at the scale of μm, the factorizability of the transverse beam distribution, and beam-beam effects. In addition, detailed luminometer rate corrections and the inclusion of novel measurements (such as the data from the Radiation Monitoring System for the Environment and Safety) lead to precise estimates of the stability and linearity over time.

In the coming years, a similarly precise calibration of the real-time luminosity delivered to the LHC will become increasingly important for standard operations. Under those conditions, the impact of out-of-time pileup effects is expected to be larger, but in principle they can be mitigated using techniques described in this paper.

Acknowledgements

Bundes-ministerium Forschungs-gemeinschaft Forschungs-zentren Rachada-pisek We congratulate our colleagues in the CERN accelerator departments for the excellent performance of the LHC machine. We thank all contributors to the LHC Luminosity Calibration and Monitoring Working Group and, in particular, its co-coordinator Witold Kozanecki for the discussions on beam instrumentation and beam physics. We also thank the technical and administrative staffs at CERN and at other CMS institutes for their contributions to the success of the CMS effort. In addition, we gratefully acknowledge the computing c enters and personnel of the Worldwide LHC Computing Grid and other centers for delivering so effectively the computing infrastructure essential to our analyses. Finally, we acknowledge the enduring support for the construction and operation of the LHC, the CMS detector, and the supporting computing infrastructure provided by the following funding agencies: the Austrian Federal Ministry of Education, Science and Research and the Austrian Science Fund; the Belgian Fonds de la Recherche Scientifique, and Fonds voor Wetenschappelijk Onderzoek; the Brazilian Funding Agencies (CNPq, CAPES, FAPERJ, FAPERGS, and FAPESP); the Bulgarian Ministry of Education and Science; CERN; the Chinese Academy of Sciences, Ministry of Science and Technology, and National Natural Science Foundation of China; the Ministerio de Ciencia Tecnología e Innovación (MINCIENCIAS), Colombia; the Croatian Ministry of Science, Education and Sport, and the Croatian Science Foundation; the Research and Innovation Foundation, Cyprus; the Secretariat for Higher Education, Science, Technology and Innovation, Ecuador; the Ministry of Education and Research, Estonian Research Council via PRG780, PRG803 and PRG445 and European Regional Development Fund, Estonia; the Academy of Finland, Finnish Ministry of Education and Culture, and Helsinki Institute of Physics; the Institut National de Physique Nucléaire et de Physique des Particules/CNRS, and Commissariat à l’Énergie Atomique et aux Énergies Alternatives/CEA, France; the Bundesministerium für Bildung und Forschung, the Deutsche Forschungsgemeinschaft (DFG), under Germany’s Excellence Strategy – EXC 2121 “Quantum Universe” – 390833306, and under project number 400140256 - GRK2497, and Helmholtz-Gemeinschaft Deutscher Forschungszentren, Germany; the General Secretariat for Research and Technology, Greece; the National Research, Development and Innovation Fund, Hungary; the Department of Atomic Energy and the Department of Science and Technology, India; the Institute for Studies in Theoretical Physics and Mathematics, Iran; the Science Foundation, Ireland; the Istituto Nazionale di Fisica Nucleare, Italy; the Ministry of Science, ICT and Future Planning, and National Research Foundation (NRF), Republic of Korea; the Ministry of Education and Science of the Republic of Latvia; the Lithuanian Academy of Sciences; the Ministry of Education, and University of Malaya (Malaysia); the Ministry of Science of Montenegro; the Mexican Funding Agencies (BUAP, CINVESTAV, CONACYT, LNS, SEP, and UASLP-FAI); the Ministry of Business, Innovation and Employment, New Zealand; the Pakistan Atomic Energy Commission; the Ministry of Science and Higher Education and the National Science Center, Poland; the Fundação para a Ciência e a Tecnologia, Portugal; JINR, Dubna; the Ministry of Education and Science of the Russian Federation, the Federal Agency of Atomic Energy of the Russian Federation, Russian Academy of Sciences, the Russian Foundation for Basic Research, and the National Research Center “Kurchatov Institute”; the Ministry of Education, Science and Technological Development of Serbia; the Secretaría de Estado de Investigación, Desarrollo e Innovación, Programa Consolider-Ingenio 2010, Plan Estatal de Investigación Científica y Técnica y de Innovación 2017–2020, research project IDI-2018-000174 del Principado de Asturias, and Fondo Europeo de Desarrollo Regional, Spain; the Ministry of Science, Technology and Research, Sri Lanka; the Swiss Funding Agencies (ETH Board, ETH Zurich, PSI, SNF, UniZH, Canton Zurich, and SER); the Ministry of Science and Technology, Taipei; the Thailand Center of Excellence in Physics, the Institute for the Promotion of Teaching Science and Technology of Thailand, Special Task Force for Activating Research and the National Science and Technology Development Agency of Thailand; the Scientific and Technical Research Council of Turkey, and Turkish Atomic Energy Authority; the National Academy of Sciences of Ukraine; the Science and Technology Facilities Council, UK; the US Department of Energy, and the US National Science Foundation. Individuals have received support from the Marie-Curie program and the European Research Council and Horizon 2020 Grant, contract nos. 675440, 724704, 752730, 765710, and 824093 (European Union); the Leventis Foundation; the Alfred P. Sloan Foundation; the Alexander von Humboldt Foundation; the Belgian Federal Science Policy Office; the Fonds pour la Formation à la Recherche dans l’Industrie et dans l’Agriculture (FRIA-Belgium); the Agentschap voor Innovatie door Wetenschap en Technologie (IWT-Belgium); the F.R.S.-FNRS and FWO (Belgium) under the “Excellence of Science – EOS” – be.h project n. 30820817; the Beijing Municipal Science and Technology Commission, no. Z191100007219010; the Ministry of Education, Youth and Sports (MEYS) of the Czech Republic; the Lendület (“Momentum”) Program and the János Bolyai Research Scholarship of the Hungarian Academy of Sciences, the New National Excellence Program ÚNKP, the NKFIA research grants 123842, 123959, 124845, 124850, 125105, 128713, 128786, and 129058 (Hungary); the Council of Scientific and Industrial Research, India; the National Science Center (Poland), contracts Opus 2014/15/B/ST2/03998 and 2015/19/B/ST2/02861; the National Priorities Research Program by Qatar National Research Fund; the Ministry of Science and Higher Education, project no. 0723-2020-0041 (Russia); the Programa de Excelencia María de Maeztu, and the Programa Severo Ochoa del Principado de Asturias; the Thalis and Aristeia programs cofinanced by EU-ESF, and the Greek NSRF; the Rachadapisek Sompot Fund for Postdoctoral Fellowship, Chulalongkorn University, and the Chulalongkorn Academic into Its 2nd Century Project Advancement Project (Thailand); the Kavli Foundation; the Nvidia Corporation; the SuperMicro Corporation; the Welch Foundation, contract C-1845; and the Weston Havens Foundation (USA).

Data Availability Statement

This manuscript has no associated data or the data will not be deposited. [Authors’ comment: Release and preservation of data used by the CMS Collaboration as the basis for publications is guided by the CMS policy as written in its document “CMS data preservation, re-use and open access policy” (https://cms-docdb.cern.ch/cgi-bin/PublicDocDB/RetrieveFile?docid=6032&filename=CMSDataPolicyV1.2.pdf&version=2).]

Declarations

Conflict of interest

The authors declare that they have no conflict of interest.

References

  • 1.Grafström P, Kozanecki W. Luminosity determination at proton colliders. Prog. Part. Nucl. Phys. 2015;81:97. doi: 10.1016/j.ppnp.2014.11.002. [DOI] [Google Scholar]
  • 2.ATLAS and CMS Collaborations, Addendum to the report on the physics at the HL-LHC, and perspectives for the HE-LHC: collection of notes from ATLAS and CMS. CERN (2019). 10.23731/CYRM-2019-007.Addendum. arXiv:1902.10229
  • 3.CMS Collaboration, The CMS experiment at the CERN LHC. JINST 3, S08004 (2008). 10.1088/1748-0221/3/08/S08004
  • 4.CMS Collaboration, Measurement of the differential cross sections for the associated production of a W boson and jets in proton–proton collisions at s=13 TeV. Phys. Rev. D 96, 072005 (2017). 10.1103/PhysRevD.96.072005. arXiv:1707.05979
  • 5.CMS Collaboration, Measurement of the differential Drell–Yan cross section in proton–proton collisions at s=13 TeV. JHEP 12, 059 (2019). 10.1007/JHEP12(2019)059. arXiv:1812.10529
  • 6.CMS Collaboration, Measurement of the tt¯ production cross section using events with one lepton and at least one jet in pp collisions at s=13 TeV. JHEP 09, 051 (2017). 10.1007/JHEP09(2017)051. arXiv:1701.06228
  • 7.CMS Collaboration, Measurement of the tt¯ production cross section, the top quark mass, and the strong coupling constant using dilepton events in pp collisions at s=13 TeV. Eur. Phys. J. C 79, 368 (2019). 10.1140/epjc/s10052-019-6863-8. arXiv:1812.10505 [DOI] [PMC free article] [PubMed]
  • 8.Hostettler M, et al. Luminosity scans for beam diagnostics. Phys. Rev. Accel. Beams. 2018;21:102801. doi: 10.1103/PhysRevAccelBeams.21.102801. [DOI] [Google Scholar]
  • 9.F. Antoniou et al., Can we predict luminosity?, in Proceedings of the 7th Evian Workshop on LHC Beam Operation, p. 125 (2016). http://cds.cern.ch/record/2293678
  • 10.S. van der Meer, Calibration of the effective beam height in the ISR. ISR Report CERN-ISR-PO-68-31 (1968). https://cds.cern.ch/record/296752
  • 11.C. Rubbia, Measurement of the luminosity of pp¯ collider with a (generalized) van der Meer method. CERN pp¯ Note 38 (1977). https://cds.cern.ch/record/1025746
  • 12.LHCb Collaboration, Precision luminosity measurements at LHCb. JINST 9, P12005 (2014). 10.1088/1748-0221/9/12/P12005. arXiv:1410.0149
  • 13.ATLAS Collaboration, Luminosity determination in pp collisions at s=8 TeV using the ATLAS detector at the LHC. Eur. Phys. J. C 76, 653 (2016). 10.1140/epjc/s10052-016-4466-1. arXiv:1608.03953 [DOI] [PMC free article] [PubMed]
  • 14.Evans L, Bryant P. LHC machine. JINST. 2008;3:S08001. doi: 10.1088/1748-0221/3/08/S08001. [DOI] [Google Scholar]
  • 15.B. Salvachua, Overview of proton–proton physics during Run 2, in Proceedings of the 9th Evian Workshop on LHC Beam Operation (2019), p. 7. http://cds.cern.ch/record/2750272
  • 16.J. Boyd, C. Schwick, Experiments—experience and future, in Proceedings of the 7th Evian Workshop on LHC Beam Operation (2017), p. 229. https://cds.cern.ch/record/2293538
  • 17.CMS Collaboration, Measurement of the double-differential inclusive jet cross section in proton–proton collisions at s=13 TeV. Eur. Phys. J. C 76, 451 (2016). 10.1140/epjc/s10052-016-4286-3. arXiv:1605.04436 [DOI] [PMC free article] [PubMed]
  • 18.Skands P, Carrazza S, Rojo J. Tuning PYTHIA 8.1: the Monash 2013 tune. Eur. Phys. J. C. 2014;74:3024. doi: 10.1140/epjc/s10052-014-3024-y. [DOI] [Google Scholar]
  • 19.Sjöstrand T, et al. An introduction to PYTHIA 8.2. Comput. Phys. Commun. 2015;191:159. doi: 10.1016/j.cpc.2015.01.024. [DOI] [Google Scholar]
  • 20.CMS Collaboration, Performance of the CMS muon detector and muon reconstruction with proton–proton collisions at s=13 TeV. JINST 13, P06015 (2018). 10.1088/1748-0221/13/06/p06015. arXiv:1804.04528
  • 21.CMS Collaboration, The CMS trigger system. JINST 12, P01020 (2017). 10.1088/1748-0221/12/01/P01020. arXiv:1609.02366
  • 22.A.J. Bell, Beam and radiation monitoring for CMS, in Proceedings, 2008 IEEE Nuclear Science Symposium (2008), p. 2322. 10.1109/NSSMIC.2008.4774822
  • 23.Lujan P. Performance of the Pixel Luminosity Telescope for luminosity measurement at CMS during Run 2. PoS. 2017;314:504. doi: 10.22323/1.314.0504. [DOI] [Google Scholar]
  • 24.M. Hempel, Development of a novel diamond based detector for machine induced background and luminosity measurements. PhD thesis, Brandenburg University of Technology Cottbus-Senftenberg, 2017. DESY-THESIS-2017-030. 10.3204/PUBDB-2017-06875
  • 25.CMS Collaboration, CMS technical design report for the level-1 trigger upgrade. CMS Technical Design Report CERN-LHCC-2013-011, CMS-TDR-012 (2013). https://cds.cern.ch/record/1556311
  • 26.CMS Collaboration, Event generator tunes obtained from underlying event and multiparton scattering measurements. Eur. Phys. J. C 76, 155 (2016). 10.1140/epjc/s10052-016-3988-x. arXiv:1512.00815 [DOI] [PMC free article] [PubMed]
  • 27.Geant4 Collaboration, Geant4—a simulation toolkit. Nucl. Instrum. Methods A 506, 250 (2003). 10.1016/S0168-9002(03)01368-8
  • 28.W. Verkerke, D.P. Kirkby, The RooFit toolkit for data modeling, in Proceedings of the 13th International Conference for Computing in High-Energy and Nuclear Physics (CHEP03) (2003), p. MOLT007. arXiv:physics/0306116. http://inspirehep.net/literature/621398
  • 29.CMS Collaboration, Description and performance of track and primary-vertex reconstruction with the CMS tracker. JINST 9, P10009 (2014). 10.1088/1748-0221/9/10/P10009. arXiv:1405.6569
  • 30.G. Segura Millan, D. Perrin, L. Scibile, RAMSES: the LHC radiation monitoring system for the environment and safety, in Proceedings of the 10th International Conference on Accelerator and Large Experimental Physics Control Systems (ICALEPCS 2005) (2005). TH3B.1-3O. https://accelconf.web.cern.ch/ica05/proceedings/pdf/O4_011.pdf
  • 31.A. Ledeul et al., CERN supervision, control, and data acquisition system for radiation and environmental protection, in Proceedings of the 12th International Workshop on Emerging Technologies and Scientific Facilities Controls (PCaPAC’18) (2019), p. 248 (FRCC3). 10.18429/JACoW-PCaPAC2018-FRCC3
  • 32.Salfeld-Nebgen J, Marlow D. Data-driven precision luminosity measurements with Z bosons at the LHC and HL-LHC. JINST. 2018;13:P12016. doi: 10.1088/1748-0221/13/12/P12016. [DOI] [Google Scholar]
  • 33.O.S. Brüning et al., LHC design report vol. 1: the LHC main ring. CERN (2004). 10.5170/CERN-2004-003-V-1
  • 34.M. Hostettler, LHC luminosity performance. PhD thesis, University of Bern. CERN-THESIS-2018-051 (2018). https://cds.cern.ch/record/2319396
  • 35.C. Barschel et al., Results of the LHC DCCT calibration studies. Technical Report CERN-ATS-Note-2012-026 PERF (2012). https://cds.cern.ch/record/1425904
  • 36.D. Belohrad et al., The LHC fast BCT system: a comparison of design parameters with initial performance. Technical Report CERN-BE-2010-010 (2010). https://cds.cern.ch/record/1267400
  • 37.Jeff A, et al. Longitudinal density monitor for the LHC. Phys. Rev. ST Accel. Beams. 2012;15:032803. doi: 10.1103/PhysRevSTAB.15.032803. [DOI] [Google Scholar]
  • 38.A. Jeff, A longitudinal density monitor for the LHC. PhD thesis, University of Liverpool. CERN-THESIS-2012-240 (2012). https://cds.cern.ch/record/1513180
  • 39.G. Papotti, T. Bohl, F. Follin, U. Wehrle, Longitudinal beam measurements at the LHC: the LHC Beam Quality Monitor, in Proceedings, 2nd International Particle Accelerator Conference (IPAC 2011) (2011), p. 1852 (TUPZ022). https://cds.cern.ch/record/1399087
  • 40.M. Gasior, J. Olexa, R. Steinhagen, BPM electronics based on compensated diode detectors—results from development systems, in Proceedings of the 2012 Beam Instrumentation Workshop (BIW’12). (2012), p. 44 (MOPG010). https://cds.cern.ch/record/1476070
  • 41.T. Persson et al., LHC optics corrections in Run 2, in Proceedings, 9th Evian Workshop on LHC Beam Operation. (2019), p. 59. https://cds.cern.ch/record/2750277
  • 42.M. Gasior, G. Baud, J. Olexa, G. Valentino, First operational experience with the LHC Diode ORbit and OScillation (DOROS) system, in Proceedings of the 5th International Beam Instrumentation Conference (IBIC 2016) (2017), p. 43 (MOPG07). 10.18429/JACoW-IBIC2016-MOPG07
  • 43.H. Bartosik, G. Rumolo, Production of the single bunch for van der Meer scans in the LHC injector chain. CERN Accelerator Note CERN-ACC-NOTE-2013-0008 (2013). https://cds.cern.ch/record/1590405
  • 44.Balagura V. Van der Meer scan luminosity measurement and beam–beam correction. Eur. Phys. J. C. 2021;81:26. doi: 10.1140/epjc/s10052-021-08837-y. [DOI] [Google Scholar]
  • 45.A. Babaev et al., Impact of beam-beam effects on absolute luminosity calibrations at the CERN Large Hadron Collider. Phys. Rev. ST Accel. Beams (2021) (to be submitted)
  • 46.C. Barschel, Precision luminosity measurement at LHCb with beam–gas imaging. PhD thesis, RWTH Aachen University (2014). CERN-THESIS-2013-301. https://cds.cern.ch/record/1693671
  • 47.G. Coombs, M. Ferro-Luzzi, R. Matev, Beam–gas imaging measurements at LHCb, in Proceedings, 7th International Beam Instrumentation Conference (IBIC 2018) (2019), p. 459 (WEPB13). 10.18429/JACoW-IBIC2018-WEPB13
  • 48.W. Kozanecki, T. Pieloni, J. Wenninger, Observation of beam–beam deflections with LHC orbit data. CERN Accelerator Note CERN-ACC-NOTE-2013-0006 (2013). https://cds.cern.ch/record/1581723
  • 49.S.M. White, Determination of the absolute luminosity at the LHC. PhD thesis, Université Paris-Sud 11. CERN-THESIS-2010-139, LAL-10-154 (2010). https://cdsweb.cern.ch/record/1308187
  • 50.Klute M, Medlock C, Salfeld-Nebgen J. Beam imaging and luminosity calibration. JINST. 2017;12:P03018. doi: 10.1088/1748-0221/12/03/P03018. [DOI] [Google Scholar]
  • 51.J. Knolle, Measuring luminosity and the tt¯Z production cross section with the CMS experiment. PhD thesis, Universität Hamburg, 2020. CERN-THESIS-2020-185, DESY-THESIS-2020-020. 10.3204/PUBDB-2020-03187

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

This manuscript has no associated data or the data will not be deposited. [Authors’ comment: Release and preservation of data used by the CMS Collaboration as the basis for publications is guided by the CMS policy as written in its document “CMS data preservation, re-use and open access policy” (https://cms-docdb.cern.ch/cgi-bin/PublicDocDB/RetrieveFile?docid=6032&filename=CMSDataPolicyV1.2.pdf&version=2).]


Articles from The European Physical Journal. C, Particles and Fields are provided here courtesy of Springer

RESOURCES