Hello Emmanuel,

The Limit: Check button performs a check of the nominal system based on the current criterion. As such, OpticStudio is building a Merit Function based on the criterion you choose and reviewing the Merit Function value for the system without any perturbations. 

However, there are several reasons this value may be different than what the Limit: Check gives you. You may have a different sampling rate; you may have included boundary operands in your file that was not included in the tolerance check; you may be using a different method to compute MTF. When it comes to optimization, the Merit Function you build will probably be much more fleshed-out than the one we use for tolerancing. It will likely have more data and more targets to be met. 

I think the easiest way to understand what OpticStudio is doing is to run a single Monte Carlo trial and save it. If you want to exactly replicate what OpticStudio is doing for the nominal, you could run this Monte Carlo file when all the tolerances are zero. In the Monte Carlo file, you will see the built merit function, which is your criterion. The value reported when clicking on Check is just the mean of the all GMTA operands. You can then try it for either option of the COMP.

Hope this helps.

Best regards,

Moj