@Carl Love 

Thanks Carl.

But it seems that : test and `test` do not mean exactly the same thing here:
 

Download hum.mw

@Preben Alsholm 

Thanks, I understand now.

@vv 

I don't understand your answer.
I know the difference between sum and add, but using sum in the definition of S works perfectly well in Maple 2015 (the version I used in my comment) and Maple 2021 (I checked it right now).

So why did you make this answer? Does S := (N, x) -> sum(...) no longer work in Maple 2023?

@acer 

Great, than you very much

@Carl Love 

"It's obvious (to me at least) ..."
I guess I have to get rid of my by crystal ball and buy a new one :-)

@acer 

Thanks, it's clear now

@acer 

I vote up... even if I do not understand what really happens.
In

Int(unapply(eval(op(1,ig),y=Y),x), rhs(op(2,ig)))

what makes Maple recognize that the integration has to be done with respect of x and not Y?

Why does the following lines generate an error?

F := Y->evalf(Int(unapply(eval(op(1,ig),y=Y),x), x=rhs(op(2,ig)))):
CodeTools:-Usage(plot(F, 0 .. 0.25, size = [400, 300]));

@Scot Gould 

F...g great!
I vote up.

@nm 

Thanks, I will try to build a simple example that presents the issue.

@acer 

Sleep on it.

Thanks for the answer.

@nm

Unfortunately, I'm not ready to provide a simple example at the moment (that would require pruning a fairly large code to extract the significant part).
What I suggest is that you delay ( :-) ) my question by deleting it and coming back to the site later with more solid material.
Do you mind if I delete it?

Use the big green up-arrow in the menu bar to upload your worksheet.

[moderator: You can also upload and attach one or more data files. You can put them in a .zip file if the Mapleprimes file-manager doesn't accept their filename extension.]

@Thomas Dean 

Why do you talk about Maplet and where the message in your initial question comes from?
I don't see any Maplet command  in your code.

There is no Maple restriction.

But having already encountered this problem in my professional activities I can tell you that since Windows10 such a limitation is imposed by the OS (and defined by the value [that you can change] of variable MAX_PATH).
Neither Mac OSX not Linux impose a limitation to the lenth of a path.

First point: replace your last line by this one

obj := add((stress[i]-true_stress[i])^2, i = 1 .. 10)

Next it would be usefull you provode ranges of variation for the seven parameters (G1,G2,G3,tau1,tau2,tau3, strain) for a crude attempt to minimize obj lead to non numeric values.