fun = function(lambda) {-y %*% log(a %*% lambda) + lambda %*% apply(a,2,sum)}
in Ma开发者_运维技巧ths notation.
Here lambda is a 2*1 matrix, a is a 5*2 matrix, y is a 1*5 matrix.
So i would like to plot fun(lambda) from (1:10, 1:10), Use persp() (or other command).
You're almost there. You just need to create a grid of values (1:10, 1:10) over which to evaluate your function. I've used dummy data to illustrate.
set.seed(123)
lambda<-as.matrix(runif(2),nrow=2)
a<-matrix(runif(10),ncol=2)
y<-runif(5)
fun = function(lambda) {-y %*% log(a %*% lambda) + lambda %*% apply(a,2,sum)}
#Create grid
lambda.grid<-expand.grid(1:10,1:10)
names(lambda.grid)<-c("lambda_1","lambda_2")
z<-apply(lambda.grid,1,fun) #Evaluate function for each row
#Change z into matrix form for persp plots
z.mat<-matrix(z,ncol=10,byrow=FALSE)
persp(x=1:10,y=1:10,z.mat,xlab="lambda_1",ylab="lambda_2",zlab="f(lambda_1,lambda_2)")
HTH
精彩评论