Errors using PM and MIPM mutations

[mmiller-max]

Hi, thanks for the package!

I'm running into a couple of errors with PM and MIPM

using Evolutionary
f(x) = x[1]^2 + 1
x0 = [0]
res = Evolutionary.optimize(f, x0, GA(selection=ranklinear(1), mutation=PM([-1.0], [3.0])))

ERROR: AssertionError: Need to set p_int for integer variables

I think this is because this line needs to pass a dummy value for p_float before p:

Evolutionary.jl/src/mutations.jl

Line 189 in 8555f34

return mipmmutation(lower, upper, p)

When I try with MIPM I get this

julia> res = Evolutionary.optimize(f, x0, GA(selection=ranklinear(1), mutation=MIPM([-1.0], [3.0])))
ERROR: InexactError: Int64(-0.9692105629432188)

Which I think is because this line will result in a float always:

Evolutionary.jl/src/mutations.jl

Line 209 in 8555f34

S = u .^ (1 ./ P) # random var following power distribution

The second problem goes away when the initial condition is a vector of floats - should it always be this or is the intention for it to work with a vector of ints?

Cheers

[wildart]

It's a bug, there is no truncation implemented for integer values.