###This model is identical to the main model, 
#except that it lets the same model (or variations on it) be iterated multiple times

##n_its sets the number of iterations
#In order to vary the parameters from one iteration to another, simply use a function of n_its 
#instead of a constant to define the parameter in question

##This code was the base of most sensitivity analyses

#Load required packages
require(corrplot)
require(flexsurv)

#Set the working directory
setwd("") ###Insert your working directory path here
source("S2_File.r") 

########Set parameters
n_its <- 3 #Number of times to repeat the model

#Population size and generations
n_inds <- 10000 #Population size
n_gens <- 500 #Number of generations
###Trade-off parameters
   #The weight of the trade-off to favor fertility versus survival. 
   #Higher values ( > 1) favor fertility, lower (<1) favor survival
   #Should be positive, 1 is "neutral" (conserves default fertility and survival) 
   #t_o1 is trade-off 1, t_o2 is trade-off 2 (only t_o1 is used in the single-trade-off model)
    t_o1_wt <- 0.43 #W1 in article Table 1, Eqns (3) - (6)
    t_o2_wt <- 0.1 #W2 in article Table 1, Eqns (3) - (6)
   #Parameters relating to stochasticity of TO values
    herit <- 0.95 #Heritability, loosely defined (h in article Table 1, Eqn (S1))
    sig_to <- 0.1 #Standard deviation of TOs in the first generation
   #Parameters for trade-off functions 
    b0 <- 0.01 #Minimum possible aging rate, b0 in article Table 1, Eqn (3) 
    yy <- -2 #Lag for TO in the b-TO function, gamma in article Table 1, Eqn (3) 
      aa <- 1 #coefficient of exponent in b-TO function, fixed at 1 (neutral) except in sensitivity analyses
      cc <- 1 #TO division factor in b-TO function, fixed at 1 (neutral) except in sensitivity analyses
   #Variance (stochastic component) in reproductive success, not too important 
    sig1 <- 0.1 #Indicated as "SD of LRS" in article Table 1, Eqn (2)
   #Constraint on maximum fertility (asymptotic limit)
    maxfert <- 1 #fmax in article Table 1, Eqn (4) 
   #Parameters for simulating age at death in various models
    G_aa <- 0.08 #a parameter in the Gompertz model, when fixed, indicated as "a" in article Table 1, Eqn (1)
     ##The following parameters are only used in sensitivity analyses on the mortality function
      GM_aa <- 0.07 #a parameter in the Gompertz-Makeham model, when fixed, Eqn (S4)
      GM_bb <- 0.09 #b parameter in the Gompertz and Gompertz-Makeham models, when fixed, Eqn (S4)
      G_bb <- 0.1 #b parameter in the Gompertz model, when fixed, Eqns (1) and (S3)
      GM_c <- 0.01 #Gompertz-Makeham c parameter, Eqn (S4)
      L_a <- 0.08 #Intercept in the linear model, Eqn (S2)


#Create vectors to store mean data across generations
mean_vecs <- c("meant_o1s", "meant_o2s", "meanbs", "meana_ds", "meanferts", "meanr_ss") 
mods <- c("live1", "live2", "live3", "live4")
vec <- matrix(NA, n_gens, 1) 
for (j in 1:length(mean_vecs)){
  for (k in 1:length(mods)){
    assign(paste(mean_vecs[j], k, sep = ""), vec) 
}}

#Create matrix to store final outputs across iterations
TO1_results <- matrix(NA, n_its, 4)
colnames(TO1_results) <- c("m1", "m2", "m3", "m4") 
TO2_results <- matrix(NA, n_its, 4)
colnames(TO2_results) <- c("m1", "m2", "m3", "m4")
fert_results <- matrix(NA, n_its, 4)
colnames(fert_results) <- c("m1", "m2", "m3", "m4")
AD_results <- matrix(NA, n_its, 4)
colnames(AD_results) <- c("m1", "m2", "m3", "m4") 
LRS_results <- matrix(NA, n_its, 4)
colnames(LRS_results) <- c("m1", "m2", "m3", "m4")

for (xxx in 1:n_its){
   #Generate trade-offs in the first generation and first generation data using functions sourced from other file
   t_o1s <- rnorm(n_inds, 0, sig_to)
   t_o2s <- rnorm(n_inds, 0, sig_to)
   data1_0 <- live1(i = 0, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
   b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy) 
   data2_0 <- live1(i = 0, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
   b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
   data3_0 <- live1(i = 0, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
   b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
   data4_0 <- live1(i = 0, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
   b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
   
   #Run for all generations, model1
   for (i in 1:n_gens){
     k <- 1 #This says it's model 1
     data <- get(paste("data1_", i-1, sep = "")) #Get previous generation data
     tt1 <- data[, 1] #Extract trade-off 1 from previous generation 
     tt2 <- data[, 2] #Extract trade-off 2 from previous generation
     pp <- data[, 8] #Extract weights based on reproductive success from previous generation
     t_o1s <- sample(x = tt1, size = n_inds, replace = T, prob = pp) #Sample previous generation proporional to reproductive success to obtain current generation trade-off values
     t_o1s <- t_o1s + rnorm(n_inds, 0, abs(t_o1s)*(1-herit)) #Perturb random as function of heritability 
     t_o2s <- sample(x = tt2, size = n_inds, replace = T, prob = pp) #Ditto
     t_o2s <- t_o2s + rnorm(n_inds, 0, abs(t_o2s)*(1-herit)) #Ditto
     #Run function to calculate all current generation data
     res <- live1(i = i, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
     b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
     #Save results
     for (j in 1:length(mean_vecs)){
       vec <- get(paste(mean_vecs[j], k, sep = ""))
       vec[i] <- mean(res[, j]) 
       assign(paste(mean_vecs[j], k, sep = ""), vec)
     }
     assign(paste("data1_", i, sep = ""), res)
   }
   TO1_results[xxx, 1] <- meant_o1s1[n_gens]
   TO2_results[xxx, 1] <- meant_o2s1[n_gens]
   fert_results[xxx, 1] <- meanferts1[n_gens]
   AD_results[xxx, 1] <- meana_ds1[n_gens]
   LRS_results[xxx, 1] <- meanr_ss1[n_gens]
   
   #Run for all generations, model2
   for (i in 1:n_gens){
     k <- 2
     data <- get(paste("data2_", i-1, sep = ""))
     tt1 <- data[, 1] 
     tt2 <- data[, 2]
     pp <- data[, 8]
     t_o1s <- sample(x = tt1, size = n_inds, replace = T, prob = pp) 
     t_o1s <- t_o1s + rnorm(n_inds, 0, abs(t_o1s)*(1-herit)) 
     t_o2s <- sample(x = tt2, size = n_inds, replace = T, prob = pp) 
     t_o2s <- t_o2s + rnorm(n_inds, 0, abs(t_o2s)*(1-herit))
     res <- live2(i = i, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
     b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
     for (j in 1:length(mean_vecs)){
       vec <- get(paste(mean_vecs[j], k, sep = ""))
       vec[i] <- mean(res[, j]) 
       assign(paste(mean_vecs[j], k, sep = ""), vec)
     }
     assign(paste("data2_", i, sep = ""), res)
   }
   TO1_results[xxx, 2] <- meant_o1s2[n_gens]
   TO2_results[xxx, 2] <- meant_o2s2[n_gens]
   fert_results[xxx, 2] <- meanferts2[n_gens]
   AD_results[xxx, 2] <- meana_ds2[n_gens]
   LRS_results[xxx, 2] <- meanr_ss2[n_gens]
   
   #Run for all generations, model3
   for (i in 1:n_gens){
     k <- 3
     data <- get(paste("data3_", i-1, sep = ""))
     tt1 <- data[, 1] 
     tt2 <- data[, 2]
     pp <- data[, 8]
     t_o1s <- sample(x = tt1, size = n_inds, replace = T, prob = pp) 
     t_o1s <- t_o1s + rnorm(n_inds, 0, abs(t_o1s)*(1-herit)) 
     t_o2s <- sample(x = tt2, size = n_inds, replace = T, prob = pp) 
     t_o2s <- t_o2s + rnorm(n_inds, 0, abs(t_o2s)*(1-herit))
     res <- live3(i = i, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
     b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
     for (j in 1:length(mean_vecs)){
       vec <- get(paste(mean_vecs[j], k, sep = ""))
       vec[i] <- mean(res[, j]) 
       assign(paste(mean_vecs[j], k, sep = ""), vec)
     }
     assign(paste("data3_", i, sep = ""), res)
   }
   TO1_results[xxx, 3] <- meant_o1s3[n_gens]
   TO2_results[xxx, 3] <- meant_o2s3[n_gens]
   fert_results[xxx, 3] <- meanferts3[n_gens]
   AD_results[xxx, 3] <- meana_ds3[n_gens]
   LRS_results[xxx, 3] <- meanr_ss3[n_gens]
   
   #Run for all generations, model4
   for (i in 1:n_gens){
     k <- 4
     data <- get(paste("data4_", i-1, sep = ""))
     tt1 <- data[, 1] 
     tt2 <- data[, 2]
     pp <- data[, 8]
     t_o1s <- sample(x = tt1, size = n_inds, replace = T, prob = pp) 
     t_o1s <- t_o1s + rnorm(n_inds, 0, abs(t_o1s)*(1-herit)) 
     t_o2s <- sample(x = tt2, size = n_inds, replace = T, prob = pp) 
     t_o2s <- t_o2s + rnorm(n_inds, 0, abs(t_o2s)*(1-herit))
     res <- live4(i = i, t_o1s = t_o1s, t_o2s = t_o2s, t_o1_wt = t_o1_wt, t_o2_wt = t_o2_wt, sig1 = sig1, 
     b0 = b0, maxfert = maxfert, n_inds = n_inds, aa = aa, cc = cc, yy = yy)
     for (j in 1:length(mean_vecs)){
       vec <- get(paste(mean_vecs[j], k, sep = ""))
       vec[i] <- mean(res[, j]) 
       assign(paste(mean_vecs[j], k, sep = ""), vec)
     }
     assign(paste("data4_", i, sep = ""), res)
   }
   TO1_results[xxx, 4] <- meant_o1s4[n_gens]
   TO2_results[xxx, 4] <- meant_o2s4[n_gens]
   fert_results[xxx, 4] <- meanferts4[n_gens]
   AD_results[xxx, 4] <- meana_ds4[n_gens]
   LRS_results[xxx, 4] <- meanr_ss4[n_gens]
}
#write.csv(TO1_results, "100 rpt TO1.csv")
#write.csv(TO2_results, "100 rpt TO2.csv")
#write.csv(fert_results, "100 rpt fert.csv")
#write.csv(AD_results, "100 rpt AD.csv")
#write.csv(LRS_results, "100 rpt LRS.csv")

#Make plots of results
groups <- c("1 TO", "Additive", "Max", "Min")
cols <- c("black", "red", "blue", "green")
windows(height = 10, width = 15)
tiff(filename = "100 repeats.tif", 9, 6, "in", res = 500, compression = "lzw")
par(mfrow = c(2, 3), mar = c(3, 2, 1, 1), mgp = c(2, 0.7, 0))
boxplot(TO1_results, names = groups, horizontal = T, xlab = "Final mean TO1", col = cols)
boxplot(TO2_results, names = groups, horizontal = T, xlab = "Final mean TO2", col = cols)
boxplot(fert_results, names = groups, horizontal = T, xlab = "Final mean fertility", col = cols)
boxplot(AD_results, names = groups, horizontal = T, xlab = "Final mean age at death", col = cols)
boxplot(LRS_results, names = groups, horizontal = T, xlab = "Final mean lifetime reproductive success", col = cols)
dev.off()