Hello,
I am working on a large-scale model and it has gotten big enough that singularities are causing some convergence issues. It’s an MCP, and the log file comes back with 1,465 singularities. I have 28 zero columns.
I have tried using the option proximal_perturbation to help convergence (doesn’t work for all cases), but I would like to make the model more robust by ridding it of these singularities. The model contains non-linear terms.
I know what the mathematical meaning for this is, but I have no clue how to improve the situation in GAMS.
Can somebody please provide some practical advice?
Thank you very much.
–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.
As an update, I am setting the levels of some variables to some vary small value but not zero to initiate the runs, and it seems to be taking care of most singular constraints.
On Tuesday, 3 December 2013 15:44:16 UTC+3, Doaa Fileli wrote:
Hello,
I am working on a large-scale model and it has gotten big enough that singularities are causing some convergence issues. It’s an MCP, and the log file comes back with 1,465 singularities. I have 28 zero columns.
I have tried using the option proximal_perturbation to help convergence (doesn’t work for all cases), but I would like to make the model more robust by ridding it of these singularities. The model contains non-linear terms.
I know what the mathematical meaning for this is, but I have no clue how to improve the situation in GAMS.
Can somebody please provide some practical advice?
Thank you very much.
–
To unsubscribe from this group and stop receiving emails from it, send an email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.
Doaa: Check the model for ill-behaved functions like sqrt(), 1/x,
exponentials and logarithms. Make sure that everything going into
these functions is adequately bounded. Remember it is probably good
advice to rewrite 1/x=y as 1=y.x
Regards!
Claudio
On Wed, Dec 4, 2013 at 5:36 AM, Doaa Fileli wrote:
As an update, I am setting the levels of some variables to some vary small
value but not zero to initiate the runs, and it seems to be taking care of
most singular constraints.
On Tuesday, 3 December 2013 15:44:16 UTC+3, Doaa Fileli wrote:
Hello,
I am working on a large-scale model and it has gotten big enough that
singularities are causing some convergence issues. It’s an MCP, and the log
file comes back with 1,465 singularities. I have 28 zero columns.
I have tried using the option proximal_perturbation to help convergence
(doesn’t work for all cases), but I would like to make the model more
robust by ridding it of these singularities. The model contains non-linear
terms.
I know what the mathematical meaning for this is, but I have no clue how
to improve the situation in GAMS.
Can somebody please provide some practical advice?
Thank you very much.
–
“gamsworld” group.
To unsubscribe from this group and stop receiving emails from it, send an
email to gamsworld+unsubscribe@googlegroups.com.
To post to this group, send email to gamsworld@googlegroups.com.
Visit this group at http://groups.google.com/group/gamsworld.
For more options, visit https://groups.google.com/groups/opt_out.
\