-
Notifications
You must be signed in to change notification settings - Fork 18
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Inconsistencies in problem classification strings #46
Comments
I am very happy to correct any mistakes you find (thank you). The second classification is a mistake, it should be 4-4. This is because we view the problems from an optimization perspective, the second number is the number of (non-simple-bound) constraints (inconsistent |
I appreciate the quick feedback. If the number of constraints in the classification should coincide with the number of residuals (given by the number of constraints in the problem), then we have the following discrepancies:
There are some cases where the classification number of constraints is not zero, but another value entirely. This may (?) be due to the conversion of linear constraints into bounds in the original problems, which I think does not make sense for NE problems, but I could be mistaken... |
Thank you very much, I will update as soon as I get a free few hours. |
Updates now on https://bitbucket.org/optrove/sif/src/master/ |
We were discussing the classification of the problems in the CUTest suite and found some inconsistencies specifically regarding how constraints are reported in the classification of nonlinear equation (NE) problems.
As an example, the
JUDGENE
problem is classified asNOR2-MN-2-20
, i.e., having 20 constraints. This is reflected in the dimension obtained when instantiating the problem. Of course, these correspond to the residuals to be minimized.Conversely, the
HATFLDBNE
problem is classified asNOR2-AN-4-0
, i.e., having no constraints despite having a residual vector with 4 entries.There are several other instances exhibiting these discrepancies (I wrote a script to detect these kinds of problems).
The question in this regard is which convention should be adopted (zero or the number of residuals, I would prefer the latter) and stick to that convention consistently.
The text was updated successfully, but these errors were encountered: