Avoid repeating computations and dealing with constant parameters in Julia, optim

Question

I am using Optim.jl in julia. This page provides some tips for writing codes.

Avoiding repeating computation, I want to optimize a cost function with providing a gradient. If there is no constant parameter in the cost function, the code below works.

using LinearAlgebra
using Optim

function cost!(F, G, x)
    if G !== nothing
        G .= x
    end
    if F !== nothing
        return dot(x, x) / 2
    end
end
sol = optimize(Optim.only_fg!(cost!), randn(5), LBFGS()) 

But what if the cost function has a constant parameter C?

using LinearAlgebra
using Optim

function cost!(F, G, x, C)
    if G !== nothing
        G .= C * x
    end
    if F !== nothing
        return C * dot(x, x) / 2
    end
end
sol = optimize(Optim.only_fg!(cost!), randn(5), LBFGS()) 

I think I have to combine the tip that writes res = optimize(b -> sqerror(b, x, y), [0.0, 0.0]) as described on that page, but I do not know how to do that.

Answer 1

I understood. What I wanted to do is probably below:

C = 1
sol = optimize(Optim.only_fg!( (F,G,x) -> cost!(F,G,x,C) ), randn(5), LBFGS())

I did not understand the arguments of anonymous functions.