SAS Optimization, and SAS Simulation Studio

turn on suggestions

Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

Topic Options

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

09-02-2010 06:04 PM

hi all,

I´am having some problems with a network design model, because the runs are taking proximately 24 hours. the model has 167 suppliers nodes, 38 warehouses, 22 plants and 1 client.

The strange thing is that firs i ran a model with all de routes available and it takes just 4 hours, and then i closed some routes (1102) and prohibited flow across this arcs (by not generating de flow variables) and with this new configuration the run takes the mentioned 24 hours. does this make sense ?. I used to think that a model with a fewer number of decision variables takes less time to run.

Does any ones has some advices to make my model run faster.

Thanks

LEQ

I´am having some problems with a network design model, because the runs are taking proximately 24 hours. the model has 167 suppliers nodes, 38 warehouses, 22 plants and 1 client.

The strange thing is that firs i ran a model with all de routes available and it takes just 4 hours, and then i closed some routes (1102) and prohibited flow across this arcs (by not generating de flow variables) and with this new configuration the run takes the mentioned 24 hours. does this make sense ?. I used to think that a model with a fewer number of decision variables takes less time to run.

Does any ones has some advices to make my model run faster.

Thanks

LEQ

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to deleted_user

09-03-2010 04:07 PM

Hello Luis,

I assume your network design problem formulation has some binary variables. Predicting how long it takes to solve a mixed integer linear programming problem sometimes can be really tricky. It does happen that a restricted problem is harder to solve than a larger version of the same problem.

A very much simplified explanation is the following: Restricting the problem means reducing the number of solutions. That means on the one hand, we have a smaller search space to look in. On the other hand it might mean that it is harder to find any feasible solution or it becomes harder to move from one feasible solution to the next.

In general I can say that larger problems are typically harder to solve, but there is plenty of very small problems that are really hard.

For linear problems (LP, i.e. no integer variables) there is a much more clear relation between size and time to solve. But even there small problems with a lot of degeneracy or difficult numerical properties might be much harder than much larger problems.

In terms of helping you to solve you instance faster, OPTMODEL allows you to set a lot of solver parameters that might speed up things for you. You can enter a technical support request here:

http://support.sas.com/ctx/supportform/createForm

and submit your data, I'll be glad to take a look at it and suggest some parameters that work well for your instance.

Philipp

I assume your network design problem formulation has some binary variables. Predicting how long it takes to solve a mixed integer linear programming problem sometimes can be really tricky. It does happen that a restricted problem is harder to solve than a larger version of the same problem.

A very much simplified explanation is the following: Restricting the problem means reducing the number of solutions. That means on the one hand, we have a smaller search space to look in. On the other hand it might mean that it is harder to find any feasible solution or it becomes harder to move from one feasible solution to the next.

In general I can say that larger problems are typically harder to solve, but there is plenty of very small problems that are really hard.

For linear problems (LP, i.e. no integer variables) there is a much more clear relation between size and time to solve. But even there small problems with a lot of degeneracy or difficult numerical properties might be much harder than much larger problems.

In terms of helping you to solve you instance faster, OPTMODEL allows you to set a lot of solver parameters that might speed up things for you. You can enter a technical support request here:

http://support.sas.com/ctx/supportform/createForm

and submit your data, I'll be glad to take a look at it and suggest some parameters that work well for your instance.

Philipp

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Philipp_SAS

09-07-2010 09:58 AM

Thanks Philipp,

I am going to enter a technical support as you recommended

I am going to enter a technical support as you recommended

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to deleted_user

09-20-2010 02:32 PM

Let me know as soon as you submitted the request and mention this thread so that your data is sent to me.

Thanks

Philipp

Thanks

Philipp