Problem in expressing the nonlinear objective

Specific Domains Optimization (Mathematical)
,
1

Hi, I’m trying to express the NLobjective of a problem which is something like this : sum((A*v_j -v_exp)^2) (A is a matrix)

I tried different forms and I got error in each case

#@NLobjective(model,Min,sum((dot(module_flux[j,:],v[indx_of_P2])-measured_flux_of_modules[j,3])^2 for j in 1:size(measured_flux_of_modules,1)))

@NLobjective(model,Min,sum((module_flux[1,:]'*v[indx_of_P2]-measured_flux_of_modules[2,3])^2 for j in 1:size(measured_flux_of_modules,1 )))

this is the error for each of the above code:

image
image1154×38 19.6 KB

and for this one :
@NLobjective(model,Min,sum((module_flux*v[indx_of_P2]-measured_flux_of_modules[:,3]).^2 ))

image
image1027×41 2.51 KB

I think I’m doing what the last error says in the other two versions of code, but as you see that error come up

2

You need to write all nonlinear expressions out as scalar expressions.

so something like:

@NLobjective(
    model,
    Min,
    sum(
        (
            sum(module_flux[1, i] * v[indx_of_P2[i]] for i in 1:size(module_flux, 2)) -
            measured_flux_of_modules[2, 3]
        )^2 for j in 1:size(measured_flux_of_modules, 1),
    ),
)

You can’t use linear algebra operations like module_flux[1,:]'*v[indx_of_P2].

1 Like
3

Thanks @odow, it worked
Could you please explain why linear algebra operations can’t be used here ?

4

Nonliner has limited syntax support. We hope to fix it in a future release.

https://jump.dev/JuMP.jl/stable/manual/nlp/#Syntax-notes

1 Like