Skip to main content
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2014 Jan 22.
Published in final edited form as: Commun Stat Theory Methods. 2012 Dec 21;42(6):566–567. doi: 10.1080/03610926.2012.718848

Corrigendum for The Distribution of Family Sizes Under a Time-Homogeneous Birth and Death Process

PMCID: PMC3898512  NIHMSID: NIHMS442843

Corrigendum for The Distribution of Family Sizes Under a Time-Homogeneous Birth and Death Process

PANAGIS MOSCHOPOULOS1 AND MAX SHPAK2

1Department of Mathematical Sciences, University of Texas at El Paso, El Paso, Texas, USA

2Department of Biological Sciences, University of Texas at El Paso, El Paso, Texas, USA

Address correspondence to Max Shpak, Department of Biological Sciences, University of Texas at El Paso, El Paso, Texas 79968, USA; Email: mshpak@utep.edu

Copyright © Taylor & Francis Group, LLC

Regretably, there are a number of significant errors in Moschopoulos and Shpak, (2010). Communications in Statistics—Theory and Methods 39:1761–1775. Most are typographic errors that have the potential to cause confusion, and in some instances, have carried through to several equations.

  1. There should be negative exponents in Eq. (1.2):
    pn(t)=e-λt(1-e-λt)n-1. (1.2)
  2. In the equation defining Qn immediately above Eq. (2.5), there is a missing negative exponent in the denominator exp[−ωt]; it should be:
    Qn=0ρωexp[-(ω+ρ)t]1-θexp[-ωt](1-exp[-ωt]1-θexp[-ωt])n-1dt.
  3. The coefficient of 1 − θ in Eq. (2.5) (and those derived from it) cancels with the change of variable, and should not appear in the equations. Note that this did not lead to errors in the numerics because of normalization. The equations should be:
    Qn=ρ01yn-1(1-y)ρ/ω(1-θy)-ρ/ω-1dy. (2.5)
    qn=ρω01yn-1(1-y)ρ/ω(1-θy)-ρ/ωdy. (2.6)
    F21(a,b,c,θ)=k=0(a)k(b)k(c)kθkk! (2.7a)
    Qn=ρΓ(1+ρω)Ak=0(1+ρω)kBθkk!. (2.8)
    Furthermore, there is an error in the position of the bracket relative to powers of θ in (2.9), which should be
    Qn=ρΓ(1+ρω)n-1-ρω[1-ρω(1+ρω)2n+ρω(1+ρω)(3(ρω)2+ρω)24n2]×k=0(1+ρω)k[1-kn(1+ρω)+φ(ρ,ω,k)n2]θkk!×(1+O(n-3)). (2.9)
  4. The factor 1 − θ is extraneous to the asymptotic 2.11–14 as well. Note the typo in the paragraph immediately before (2.11), i.e., we have θ > 1 rather than θ < 1 in this instance. Moreover, there are a number of other errors with the indices and signs (most significantly, the equations should have θ−1 in place of θ).
    Qn=ρ01/θyn-1(1-y)ρ/ω(1-θy)-ρ/ω-1dy. (2.11)
    Qn=ρθ-n01zn-1(1-zθ)ρ/ω(1-z)-ρ/ω-1dz. (2.12)
    Note that powers of θ should be negative, since θ > 1
    Qn=ρθ-nΓ(-ρω)Ak=0(-ρω)kBθ-kk!. (2.13)
    For the above distribution, the coefficients A and B are
    A=Γ(n)Γ(n-ρω),B(ρ,ω,k,n)=(n)k(n-ρω)k=Γ(n+k)Γ(n-ρω)Γ(n)Γ(n+k-ρω).
    As in (2.9), there is also an error in the position of the bracket relative to the powers of θ in the expansion
    Qn=ρθ-nΓ(-ρω)nρω×{1-ρω(ρω+1)2n+(ρω-1)(ρω-2)[3(1+ρω)2-ρω-1]24n2}×k=0(-ρω)k[1+kn(ρω+1)+φ(ρ,ω,k)n2]θ-kk!+O(n-3). (2.14)
    The asymptotic approximation to this distribution, calculated for n → ∞, is:
    Qnρθ-n(1-θ-1)ρ/ωΓ(-ρω)nρ/ω. (2.15)

RESOURCES