Evaluating and plotting a 3d log-likelihood function using R - r

My objective is to evaluate and to plot the following log-likelihood function with two parameters:
set.seed(123)
N = 100
x = 4*rnorm(N)
y = 0.8*x + 2*rnorm(N);
LogL <- function(param,x,y,N){
beta = param[1]
sigma2 = param[2]
LogLikelihood = -N/2*log(2*pi*sigma2) - sum( (y - beta*x)^2 / (2*sigma2) ) }
I've tried using 'outer' in order to use 'wireframe' as in the following thread:
How can I plot 3D function in r?
but without success:
param1 <- seq(-2, 2, length= 30)
param2 <- seq(0.1, 4, length= 30)
values1 <- matrix(c(param1,param2),30)
z <- outer(values1, x=x,y=y,N=N, LogL)
How can I use 'outer' properly in this case? Is there any other alternative to evaluate and to plot the function 'LogL'?

The first question is that outer() needs to have the two arguments to LogL() as separate vectors, which means rewriting your function to use the two arguments instead of one length 2 argument that you unpack. In addition, (from ?outer)
Each will be extended by rep to length the products of the lengths of X and Y before FUN is called.
so the function needs to deal with ALL the possible values of X and Y in a single call. There's probably a better way to do this, but in the interest of simplicity, I used a for() loop to loop over the different values of X and Y.
set.seed(123)
N = 100
x = 4*rnorm(N)
y = 0.8*x + 2*rnorm(N);
LogL <- function(param1, param2, x, y, N){
beta = param1
sigma2 = param2
LogLikelihood = vector("numeric",length(beta))
for (i in 1:length(beta)){
LogLikelihood[i] = -N/2*log(2*pi*sigma2[i]) - sum( (y - beta[i]*x)^2 / (2*sigma2[i]) )
}
return(LogLikelihood)
}
param1 <- seq(-2, 2, length= 30)
names(param1) <- param1
param2 <- seq(0.1, 4, length= 30)
names(param2) <- param2
z <- outer(X = param1, Y = param2, FUN = LogL, N = N, x = x, y = y)
require(lattice)
#> Loading required package: lattice
wireframe(z, drape=T, col.regions=rainbow(100))
# test it out
LogL(param1[3], param2[3], x = x, y = y, N = N)
#> [1] -11768.37
z[3,3]
#> [1] -11768.37
and that works.
Created on 2018-03-13 by the reprex package (v0.2.0).

Related

Time varying parameter-matrix in deSolve R

I am struggling with this for so long. I have a logistic growth function where the growth parameter
r is a matrix. The model is constructed in a way that I have as an output two N the N1 and N2.
I would like to be able to change the r parameter over time. When time < 50 I would like
r = r1 where
r1=matrix(c(
2,3),
nrow=1, ncol=2
When time >= 50 I would like r=r2 where
r2=matrix(c(
1,2),
nrow=1, ncol=2
Here is my function. Any help is highly appreciated.
rm(list = ls())
library(deSolve)
model <- function(time, y, params) {
with(as.list(c(y,params)),{
N = y[paste("N",1:2, sep = "")]
dN <- r*N*(1-N/K)
return(list(c(dN)))
})
}
r=matrix(c(
4,5),
nrow=1, ncol=2)
K=100
params <- list(r,K)
y<- c(N1=0.1, N2=0.2)
times <- seq(0,100,1)
out <- ode(y, times, model, params)
plot(out)
I would like ideally something like this but it does not work
model <- function(time, y, params) {
with(as.list(c(y,params)),{
N = y[paste("N",1:2, sep = "")]
r = ifelse(times < 10, matrix(c(1,3),nrow=1, ncol=2),
ifelse(times > 10, matrix(c(1,4),nrow=1, ncol=2), matrix(c(1,2),nrow=1, ncol=2)))
print(r)
dN <- r*N*(1-N/K)
return(list(c(dN)))
})
}
Thank you for your time.
Here a generic approach that uses an extended version of the approx function. Note also some further simplifications of the model function and the additional plot of the parameter values.
Edit changed according to the suggestion of Lewis Carter to make the parameter change at t=3, so that the effect can be seen.
library(simecol) # contains approxTime, a vector version of approx
model <- function(time, N, params) {
r <- approxTime(params$signal, time, rule = 2, f=0, method="constant")[-1]
K <- params$K
dN <- r*N*(1-N/K)
return(list(c(dN), r))
}
signal <- matrix(
# time, r[1, 2],
c( 0, 2, 3,
3, 1, 2,
100, 1, 2), ncol=3, byrow=TRUE
)
## test of the interpolation
approxTime(signal, c(1, 2.9, 3, 100), rule = 2, f=0, method="constant")
params <- list(signal = signal, K = 100)
y <- c(N1=0.1, N2=0.2)
times <- seq(0, 10, 0.1)
out <- ode(y, times, model, params)
plot(out)
For a small number of state variables like in the example, separate signals with approxfun from package stats will look less generic but may be slighlty faster.
As a further improvement, one may consider to replace the "hard" transitions with a more smooth one. This can then directly be formulated as a function without the need of approx, approxfun or approxTime.
Edit 2:
Package simecol imports deSolve, and we need only a small function from it. So instead of loading simecol it is also possible to include the approxTime function explicitly in the code. The conversion from data frame to matrix improves performance, but a matrix is preferred anyway in such cases.
approxTime <- function(x, xout, ...) {
if (is.data.frame(x)) {x <- as.matrix(x); wasdf <- TRUE} else wasdf <- FALSE
if (!is.matrix(x)) stop("x must be a matrix or data frame")
m <- ncol(x)
y <- matrix(0, nrow=length(xout), ncol=m)
y[,1] <- xout
for (i in 2:m) {
y[,i] <- as.vector(approx(x[,1], x[,i], xout, ...)$y)
}
if (wasdf) y <- as.data.frame(y)
names(y) <- dimnames(x)[[2]]
y
}
If you want to pass a matrix parameter you should pass a list of parameters and you can modify it inside the model when your time limit is exceeded (in the example below you don't even have to pass the r matrix to the model function)
library(deSolve)
model <- function(time, y, params) {
with(as.list(c(y,params)),{
if(time < 3) r = matrix(c(2,3), nrow = 1, ncol = 2)
else r = matrix(c(1,3), nrow = 1, ncol = 2)
N = y[paste("N",1:2, sep = "")]
dN <- r*N*(1-N/K)
return(list(c(dN)))
})
}
y <- c(N1=0.1, N2=0.2)
params <- list(r = matrix(c(0,0), nrow = 1, ncol = 2), K=100)
times <- seq(0,10,0.1)
out <- ode(y, times, model, params)
plot(out)
You can see examples of this for instance with Delay Differential Equations ?dede

How can I get derivative value in R?

I want to get the derivative value from the function below when x = 2. Is there way to keep the form of the function and also get derivative value with out any additional package?
f <- function(x)
return(x^3)
For example, I have tried below but they didn't work.
x=2
deriv(~f, "x")
x=2
deriv(~body(f),"x")
x=2
D(expression(f),"x")
You can use deriv, however, one caveat is that you can only use expressions/calls.
derivative = deriv(~ x^3, "x")
x <- 2
eval(derivative )
With a named expression:
f = expression(x^3)
dx2x <- D(f,"x")
and the rest is the same.
See this link for the documentation:
https://www.rdocumentation.org/packages/Deriv/versions/3.8.2/topics/Deriv
This would be approximation
foo = function(x, delta = 1e-5, n = 3){
x = seq(from = x - delta, to = x + delta, length.out = max(2, n))
y = x^3
mean(diff(y)/diff(x))
}
foo(2)
#[1] 12

How to loop all values in a dataframe as the start value in maxLik

I'm doing Maximum Likelihood Estimation using maxLik, which requires specifying starting values. Instead of specifying a single value, is there any way that allows me to use all the values from a matrix as the start value?
My current code of maxLik is:
f12 <- function(param){
alpha <- param[1]
rho <- param[2]
lambda <- param[3]
u <- 0.5*(p12$v_50_1)^alpha + 0.5*lambda*(p12$v_50_2)^alpha
p <- 1/(1 + exp(-rho*u))
f <- sum(p12$gamble*log(p) + (1-p12$gamble)*log(1-p))}
ml <- maxLik(f12, start = c(alpha = 1, rho=2, lambda = 1), method = "NM")
I create a dataframe with the upper and lower bounds of potential start values:
st <- expand.grid(alpha = seq(0, 2, len = 100),rho = seq(0, 1, len = 100),lambda = seq(0,2, length(100))
There are 3 parameters in my function, and my goal is to loop all the values in the above dataframe st and select the best vector of start values after running the model from a variety of starting parameters.
Thanks!
Consider Map (wrapper to mapply) to pass the st columns elementwise through your methods. Here, Map will return a list of maxLik objects, specifically inherited maxim class objects containing a list of other components. The number of items in this list will be equal to rows of st.
Notice input parameters, a, r, and l being passed into start argument of maxLik() and no longer hard-coded integers. And f12 is left untouched.
maxLik_run <- function(a, r, l) {
tryCatch({
f12 <- function(param){
alpha <- param[1]
rho <- param[2]
lambda <- param[3]
u <- 0.5*(p12$v_50_1)^alpha + 0.5*lambda*(p12$v_50_2)^alpha
p <- 1/(1 + exp(-rho*u))
f <- sum(p12$gamble*log(p) + (1-p12$gamble)*log(1-p))
}
return(maxLik(f12, start = c(alpha = a, rho = r, lambda = l), method = "NM"))
}, error = function(e) return(NA))
}
st <- expand.grid(alpha = seq(0, 2, len = 100),
rho = seq(0, 1, len = 100),
lambda = seq(0, 2, length(100)))
maxLik_list <- Map(maxLik_run, st$alpha, st$rho, st$lambda)
And to answer the question --best vector of start values after running the model from a variety of starting parameters-- requires a particular definition of "best". Once you define this, you can use Filter() on your returned list of objects to select the one or more element that yields this "best".
Below is a demonstration to find the highest value across each maximum likelihood's maximum. Use estimate if needed. Do note, this returned list can have more than one if the highest value is shared by other list items:
highest_value <- max(sapply(maxLik_list, function(item) item$maximum))
maxLik_item_list <- Filter(function(i) i$maximum == highest_value, maxLik_list)
What you are doing in your logLik function is that you are calculating alpha,lambda,rho whereas your data already has them.Those are the lines with u,p and f12(that is also your function name!). Also it is possible to calculate log likelihood for one row as your log likelihood function has single indices. So you run the code using apply like this
#create a function to find mle estimate for first row
maxlike <- function(a) {
f12 <- function(param){
alpha <- param[1]
rho <- param[2]
lambda <- param[3]
#u <- 0.5*(p12$v_50_1)^alpha + 0.5*lambda*(p12$v_50_2)^alpha
#p <- 1/(1 + exp(-rho*u))
#f12 <- sum(p12$gamble*log(p) + (1-p12$gamble)*log(1-p))
}
ml <- maxLik(f12, start = c(alpha = 1, rho=2, lambda = 1), method = "NM")
}
#then using apply with data = st, 2 means rows and your mle function
mle <- apply(st,2,maxlike)
mle

Multi-data likelihood function and mle2 function from bbmle package in R

I have written a custom likelihood function that fits a multi-data model that integrates mark-recapture and telemetry data (sensu Royle et al. 2013 Methods in Ecology and Evolution). The likelihood function is designed to be flexible in terms of whether and how many covariates are specified for different linear models in different likelihood components which is determined by values supplied as function arguments (i.e., data matrices "detcovs" and "dencovs" in my code). The likelihood function works when I directly supply it to optimization functions (e.g., optim or nlm), but does not play nice with the mle2 function in the bbmle package. My problem is that I continually run into the following error: "some named arguments in 'start' are not arguments to the specified log-likelihood function". This is my first attempt at writing custom likelihood functions so I'm sure there are general coding conventions of which I'm unaware that make such tasks much more efficient and amendable to the mle2 function. Below is my likelihood function, code creating the staring value objects, and code calling the mle2 function. Any advice how to solve the error problem and general comments on writing cleaner functions is welcome. Many thanks in advance.
Edit: As requested, I have simplified the likelihood function and provided code to simulate reproducible data to which the model can be fit. Included in the simulation code are 2 custom functions and use of the raster function from the raster package. Hopefully, I have sufficiently simplified everything to enable others to troubleshoot. Again, many thanks for your help!
Jared
Likelihood function:
CSCR.RSF.intlik2.EXAMPLE <- function(alpha0,sigma,alphas=NULL,betas=NULL,n0,yscr=NULL,K=NULL,X=X,trapcovs=NULL,Gden=NULL,Gdet=NULL,ytel=NULL,stel=NULL,
dencovs=NULL,detcovs=NULL){
#
# this version of the code handles a covariate on log(Density). This is starting value 5
#
# start = vector of starting values
# yscr = nind x ntraps encounter matrix
# K = number of occasions
# X = trap locations
# Gden = matrix with grid cell coordinates for density raster
# Gdet = matrix with gride cell coordinates for RSF raster
# dencovs = all covariate values for all nGden pixels in density raster
# trapcovs = covariate value at trap locations
# detcovs = all covariate values for all nGrsf pixels in RSF raster
# ytel = nguys x nGdet matrix of telemetry fixes in each nGdet pixels
# stel = home range center of telemetered individuals, IF you wish to estimate it. Not necessary
# alphas = starting values for RSF/detfn coefficients excluding sigma and intercept
# alpha0 = starting values for RSF/detfn intercept
# sigma = starting value for RSF/detfn sigma
# betas = starting values for density function coefficients
# n0 = starting value for number of undetected individuals on log scale
#
n0 = exp(n0)
nGden = nrow(Gden)
D = e2dist(X,Gden)
nGdet <- nrow(Gdet)
alphas = alphas
loglam = alpha0 -(1/(2*sigma*sigma))*D*D + as.vector(trapcovs%*%alphas) # ztrap recycled over nG
psi = exp(as.vector(dencovs%*%betas))
psi = psi/sum(psi)
probcap = 1-exp(-exp(loglam))
#probcap = (exp(theta0)/(1+exp(theta0)))*exp(-theta1*D*D)
Pm = matrix(NA,nrow=nrow(probcap),ncol=ncol(probcap))
ymat = yscr
ymat = rbind(yscr,rep(0,ncol(yscr)))
lik.marg = rep(NA,nrow(ymat))
for(i in 1:nrow(ymat)){
Pm[1:length(Pm)] = (dbinom(rep(ymat[i,],nGden),rep(K,nGden),probcap[1:length(Pm)],log=TRUE))
lik.cond = exp(colSums(Pm))
lik.marg[i] = sum( lik.cond*psi )
}
nv = c(rep(1,length(lik.marg)-1),n0)
part1 = lgamma(nrow(yscr)+n0+1) - lgamma(n0+1)
part2 = sum(nv*log(lik.marg))
out = -1*(part1+ part2)
lam = t(exp(a0 - (1/(2*sigma*sigma))*t(D2)+ as.vector(detcovs%*%alphas)))# recycle zall over all ytel guys
# lam is now nGdet x nG!
denom = rowSums(lam)
probs = lam/denom # each column is the probs for a guy at column [j]
tel.loglik = -1*sum( ytel*log(probs) )
out = out + tel.loglik
out
}
Data simulation code:
library(raster)
library(bbmle)
e2dist <- function (x, y){
i <- sort(rep(1:nrow(y), nrow(x)))
dvec <- sqrt((x[, 1] - y[i, 1])^2 + (x[, 2] - y[i, 2])^2)
matrix(dvec, nrow = nrow(x), ncol = nrow(y), byrow = F)
}
spcov <- function(R) {
v <- sqrt(nrow(R))
D <- as.matrix(dist(R))
V <- exp(-D/2)
cov1 <- t(chol(V)) %*% rnorm(nrow(R))
Rd <- as.data.frame(R)
colnames(Rd) <- c("x", "y")
Rd$C <- as.numeric((cov1 - mean(cov1)) / sd(cov1))
return(Rd)
}
set.seed(1234)
co <- seq(0.3, 0.7, length=5)
X <- cbind(rep(co, each=5),
rep(co, times=5))
B <- 10
co <- seq(0, 1, length=B)
Z <- cbind(rep(co, each=B), rep(co, times=B))
dencovs <- cbind(spcov(Z),spcov(Z)[,3]) # ordered as reading raster image from left to right, bottom to top
dimnames(dencovs)[[2]][3:4] <- c("dencov1","dencov2")
denr.list <- vector("list",2)
for(i in 1:2){
denr.list[[i]] <- raster(
list(x=seq(0,1,length=10),
y=seq(0,1,length=10),
z=t(matrix(dencovs[,i+2],10,10,byrow=TRUE)))
)
}
B <- 20
co <- seq(0, 1, length=B)
Z <- cbind(rep(co, each=B), rep(co, times=B))
detcovs <- cbind(spcov(Z),spcov(Z)[,3]) # ordered as reading raster image from left to right, bottom to top
dimnames(detcovs)[[2]][3:4] <- c("detcov1","detcov2")
detcov.raster.list <- vector("list",2)
trapcovs <- matrix(0,J,2)
for(i in 1:2){
detr.list[[i]] <- raster(
list(x=seq(0,1,length=20),
y=seq(0,1,length=20),
z=t(matrix(detcovs[,i+2],20,20,byrow=TRUE)))
)
trapcovs[,i] <- extract(detr.list[[i]],X)
}
alpha0 <- -3
sigma <- 0.15
alphas <- c(1,-1)
beta0 <- 3
betas <- c(-1,1)
pixelArea <- (dencovs$y[2] - dencovs$y[1])^2
mu <- exp(beta0 + as.matrix(dencovs[,3:4])%*%betas)*pixelArea
EN <- sum(mu)
N <- rpois(1, EN)
pi <- mu/sum(mu)
s <- dencovs[sample(1:nrow(dencovs), size=N, replace=TRUE, prob=pi),1:2]
J <- nrow(X)
K <- 10
yc <- d <- p <- matrix(NA, N, J)
D <- e2dist(s,X)
loglam <- t(alpha0 - t((1/(2*sigma*sigma))*D*D) + as.vector(trapcovs%*%alphas))
p <- 1-exp(-exp(loglam))
for(i in 1:N) {
for(j in 1:J) {
yc[i,j] <- rbinom(1, K, p[i,j])
}
}
detected <- apply(yc>0, 1, any)
yscr <- yc[detected,]
ntel <- 5
nfixes <- 100
poss.tel <- which(s[,1]>0.2 & s[,1]<0.8 & s[,2]>0.2 & s[,2]<0.8)
stel.id <- sample(poss.tel,ntel)
stel <- s[stel.id,]
ytel <- matrix(NA,ntel,nrow(detcovs))
d <- e2dist(stel,detcovs[,1:2])
lam <- t(exp(1 - t((1/(2*sigma*sigma))*d*d) + as.vector(as.matrix(detcovs[,3:4])%*%alphas)))
for(i in 1:ntel){
ytel[i,] <- rmultinom(1,nfixes,lam[i,]/sum(lam[i,]))
}
Specify starting values and call mle2 function:
start1 <- list(alpha0=alpha0,sigma=sigma,alphas=alphas,betas=betas,n0=log(N-nrow(yscr)))
parnames(CSCR.RSF.intlik2.EXAMPLE) <- names(start)
out1 <- mle2(CSCR.RSF.intlik2.EXAMPLE,start=start1,method="SANN",optimizer="optim",
data=list(yscr=yscr,K=K,X=X,trapcovs=trapcovs,Gden=dencovs[,1:2],Gdet=detcovs[,1:2],
ytel=ytel,stel=stel,dencovs=as.matrix(dencovs[,3:4]),detcovs=as.matrix(detcovs[,3:4]))
)

Error supplying appropriate start parameters to nlsLM

I was playing around with the nlsLM function, from the minpack.lm library, and encountered some behaviour that I don't understand.
Given that the following function produces output when I supply a numeric vector 'b' as input I wanted to use this function to fit a nonlinear model to my data.
volEquation <- function(DBH, PHt, b){
b[1] * DBH^b[2] * PHt^b[3]
}
However I have become stuck when it comes to correctly specifying the initial parameter values. R code follows:
library(minpack.lm)
n <- 20
x <- seq(12, 60, length.out = n)
y <- seq(22, 45, length.out = n)
z <- x^2 * y ^ 3 + rnorm(n, 0, 0.1)
Data <- data.frame(DBH = x, PHt = y, TVT = z)
nlsFormula <- "TVT ~ volEquation(DBH, PHt, b)"
nlsInitial <- list(b = c(0.5, 2.25, 3.25))
nlsLMOutput <- nlsLM(formula = nlsFormula, data = Data, start = nlsInitial)
nlsOutput <- nls(formula = nlsFormula, data = Data, start = nlsInitial
nls was successful at fitting the data while nlsLM gave me this error message,
Error in rownames<-(*tmp*, value = "b") :
length of 'dimnames' [1] not equal to array extent
Can anyone provide insight as to why this problem occurs in the nlsLM function? I've tried sifting through the nlsLM code but I still don't understand what's going on.
Try separating your parameters
volEquation <- function(DBH, PHt, x,y,z){
x * DBH^y * PHt^z
}
nlsFormula <- "TVT ~ volEquation(DBH, PHt, x, y, z)"
nlsInitial <- c(x=5e-3, y=2, z=1)
nlsOutput <- nlsLM(formula = nlsFormula, data = Data, start = nlsInitial, control=nls.lm.control(maxiter=100))

Resources