对于constraOptim中的原子向量，R $运算符无效

Question

I have a question here in how to implement constrOptim :

The example from documentation goes as this:

fQP <- function(b) {-sum(c(0,5,0)*b)+0.5*sum(b*b)}
Amat       <- matrix(c(-4,-3,0,2,1,0,0,-2,1), 3, 3)
bvec       <- c(-8, 2, 0)
constrOptim(c(2,-1,-1), fQP, NULL, ui = t(Amat), ci = bvec)

However, when I just change the code to :

fQP <- function(b) {-sum(c(0,5,0)*b)+0.5*sum(b*b)+c}
Amat       <- matrix(c(-4,-3,0,2,1,0,0,-2,1), 3, 3)
bvec       <- c(-8, 2, 0)
constrOptim(c(2,-1,-1), fQP, NULL, ui = t(Amat), ci = bvec,c=5)

An error occurs as:

Error: $ operator is invalid for atomic vectors

Could anyone help me out?

Answer 1

The error is stemming from the fact that the argument c=5 is matching the control argument to the constrOptim function, so this is the same as calling:

constrOptim(c(2,-1,-1), fQP, NULL, ui = t(Amat), ci = bvec,control=5)
# Error: $ operator is invalid for atomic vectors

The control parameter expects to be passed a list but is being passed 5 instead, so when it tries to access elements it raises this error.

To make this setup work, you need to explicitly state that c is an argument to your function (and further change it to a name that won't conflict with the built-in c function, which you also use in that function):

fQP <- function(b, c2) {-sum(c(0,5,0)*b)+0.5*sum(b*b)+c2}
Amat       <- matrix(c(-4,-3,0,2,1,0,0,-2,1), 3, 3)
bvec       <- c(-8, 2, 0)
constrOptim(c(2,-1,-1), fQP, NULL, ui = t(Amat), ci = bvec,c2=5)
# $par
# [1] 0.4762222 1.0475556 2.0951112
# 
# $value
# [1] 2.619048
# 
# $counts
# function gradient 
#      506       NA 
# 
# $convergence
# [1] 0
# 
# $message
# NULL
# 
# $outer.iterations
# [1] 3
# 
# $barrier.value
# [1] -0.0006243968