Function Handles in Julia

Question

What is the standard way to define a callback function, or a function handle in Julia?

Suppose I define

function myFun(a, b, c, d)
 a - 3* b - c * d # The return value
end

My goal is to fix b = 1, c = 2, d = 3, and pass myFun as a function of a. Something like:

newFun2(x) = myFun(x, 1 ,2, 3)
myReceiver(myFun2)

The problem is when I pass the function handle that defines an objective in the way above to a solver (NLOpt) to minimize the objective, I receive an error like myFun2 has no methods matching my usage .. — user25004, Apr 28 '15 at 17:16
So the problem should be with the requirements of the solver. :-) I just wanted to double check the function handle format in Julia, since I was not sure about it. Thank you Matt B. — user25004, Apr 28 '15 at 18:06

score 4 · Accepted Answer · answered Apr 30 '15 at 03:35

According to the documentation, the objective function for NLOpt should be of the form:

function f(x::Vector, grad::Vector):
    if length(grad) > 0:
        ...set grad to gradient, in-place...
    return ...value of f(x)...
end

Thus, the code would need to look something like:

function myFun(a, b, c, d, grad)
    a - 3* b - c * d # The return value
end

newFun2(x, grad) = myFun(x, 1 , 2, 3, grad)

myFun() would have to compute the values of the grad vector as well as returning the objective function value, if it was to work successfully with optimization algorithms that use the derivative information that myFun() is required to write to grad.

Function Handles in Julia

1 Answers1