-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathfit5.4.mcmc0.R
167 lines (135 loc) · 4.03 KB
/
fit5.4.mcmc0.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
# author : Erik Volz
# date : August 1 2017
# Bayesian MCMC fit model
# Metropolis-hastings
# iterations
ITER <- 10e3
require(akima)
source('fit5.4.helpers.R') # load the model and data
## PRIORS
ln.dprior0 <- function(theta, theta_clade)
{
lnd<- 0
for (n in names(theta)){
if (n=='p12') {
lnd <- lnd + dbeta( theta[n], 2, 2, log=T)
}
if (n=='p22') {
lnd <- lnd + dbeta( theta[n], 2, 2, log=T)
}
if (n=='importRate') {
lnd <- lnd + dexp( theta[n], 1/20, log=T)
}
if (n=='w2') {
lnd <- lnd + dlnorm( theta[n], log(1), 1, log=T)
}
if (n=='wchron') {
lnd <- lnd + dlnorm( theta[n], log(1), 2, log=T)
}
}
for (n in rownames(theta_clade ) ){
if (n == 'I0'){
lnd <- lnd + sum( dexp( theta_clade[n,], 1, log=T) )
}
if (n == 'N'){
lnd <- lnd + sum( dlnorm( theta_clade[n,], log(1e3), 2, log=T) )
}
if (n == 'beta'){
lnd <- lnd + sum( dlnorm( theta_clade[n,], log(.1), 1, log=T) )
}
}
lnd
}
## PROPOSALS
# parameters estimated for each clade:
prop_sd_clade <- c( beta = .0025, N = 500, I0 = .1)
# parameters shared by all clades:
pnames_pooled <- c( 'wchron', 'w2', 'importRate', 'p12', 'p22')
prop_sd_pooled <- c( wchron = .1
, w2 = .5
, importRate = .01
, p12 = .01
, p22 = .05
)
prop0 <- function(theta, theta_clade, iter, focus_clade = 5e3 )
{
vn_clade <- rownames(theta_clade )
np_clade <- length( vn_clade )
.theta <- theta
.theta_clade <- theta_clade
if ( (iter %% 2 == 1) & (iter > focus_clade)){
l <- 1 + (iter %% np_pooled )
n <- pnames_pooled[ l]
.theta[n] <- rnorm( 1, .theta[n], prop_sd_pooled[n] )
} else{ # update a clade var in order
k <- 1 + floor( iter/2 )
l <- 1 + (k %% np_clade )
n <- vn_clade[l]
.theta_clade[n,] <- theta_clade[n, ] + rnorm( ncol(theta_clade), 0, prop_sd_clade[ n ] )
}
list(.theta , .theta_clade )
}
## OBJFUN
of.mcmc0 <- function( xtheta, xtheta_clade){
lndp <- ln.dprior0( xtheta, xtheta_clade )
if (is.infinite( lndp)) return(lndp)
.theta<- theta
.theta[names(xtheta)] <- unname( xtheta )
.theta[ c('wb', 'wc')] <- unname( xtheta['wchron'] )
ll <- sum( unlist( mclapply( 1:NCLADES, function(k) {
bdt <- bdts[[k]]
.theta[pnames_clade] <- unname( xtheta_clade[pnames_clade,k] ) # set beta I0 and N for this clade
if ( !.parm.limits( .theta)) return(-Inf)
.parms <- as.list( .theta )
.parms$ssbeta.fun <- ssbeta.fun
.x0 <- x0
.x0[-m] <- .x0[-m] * .theta['I0']
suppressWarnings( { ll <- colik.pik(bdt, .parms, dm, .x0, 1980 , maxHeight=MAXHEIGHT
, integrationMethod = 'lsoda'
, res = 12*MAXHEIGHT) })
print(c( date(), ll ) )
ll
}, mc.cores = mc_cores ) ) )
print('#####' )
print( xtheta )
print( xtheta_clade )
print(c( date(), ll ) )
print('#####' )
if (is.na( ll)) return(-Inf)
lnd <- ll + lndp
}
## MCMC
MAXHEIGHT <- 15
LOGPO <- rep( NA, ITER)
THETA_POOLED <- matrix( NA, nrow=ITER, ncol = length(pnames_pooled ))
THETA_CLADE <- array( NA, dim = c( length(pnames_clade), NCLADES, ITER) )
.LOGPO <- rep( NA, ITER)
.THETA_POOLED <- matrix( NA, nrow=ITER, ncol = length(pnames_pooled ))
.THETA_CLADE <- array( NA, dim = c( length(pnames_clade), NCLADES, ITER) )
## START CONDITIONS
i_rand_start <- sample(1:ncol(starts), size=NCLADES)
theta_pooled <- rowMeans( starts[ pnames_pooled, i_rand_start ] )
theta_clade <- starts[ pnames_clade, i_rand_start ]
logpo <- of.mcmc0( theta_pooled, theta_clade )
for (iter in 1:ITER){
cat('*****************\n')
print( paste( "ITER" , iter ))
prop <- prop0( theta_pooled, theta_clade, iter )
.theta_pooled <- prop[[1]]
.theta_clade <- prop[[2]]
.logpo <- of.mcmc0( .theta_pooled, .theta_clade )
.LOGPO[iter] <- .logpo
.THETA_POOLED[iter,] <- .theta_pooled
.THETA_CLADE[,,iter] <- .theta_clade
if ( runif(1) < exp( .logpo - logpo ) ){
logpo <- .logpo
theta_pooled <- .theta_pooled
theta_clade <- .theta_clade
}
LOGPO[iter] <- logpo
THETA_POOLED[iter,] <- theta_pooled
THETA_CLADE[,,iter] <- theta_clade
}
## SAVE
ofn <- paste0( 'f5.4.mcmc0/', Sys.getpid(), '.rds' )
saveRDS( list( logpo = LOGPO, clade = THETA_CLADE, pooled = THETA_POOLED ) , ofn)