search

ampl / mp

An open-source library for mathematical programming
https://mp.ampl.com
Other
227 stars 43 forks source link

Xpress mip:basis=1 output #222

Open 4er4er4er opened 1 year ago

4er4er4er commented 1 year ago

There are some issues with Xpress output when mip:basis=1 and tech:outlev=1. The attached example MDXoptimalmarkets.zip produces a listing that ends as follows:

Setup time = 2.929898s
Solution time = 224.029981s
Output time = 1805.603075s
XPRESS 9.0.3 (41.01.04): optimal solution; objective 196040.6327
37882 simplex iterations
38 barrier iterations
Fixed MIP for mip:basis: 562267 simplex iterationsabsmipgap=2755.82, relmipgap=0.0140574

*******
_solve_elapsed_time = 2033.33
_ampl_elapsed_time = 843.172
*******
  1. Fixed MIP for mip:basis: 562267 simplex iterations appears in the middle of AMPL's solve_message.
  2. The Output time = 1805.603075s is much larger than what one sees when using mip:basis=0. Possibly the time to solve the fixed MIP is being included in the output time rather than the solution time.
  3. When tech:outlev=1 is specified, it would be valuable for diagnostic purposes to see the log lines for the solver's progress on the fixed MIP. (In this particular example, the user wants to understand why the fixed MIP is taking so long to solve.)

The xpress_options settings are as given by the user who provided this example. Also, to avoid showing the model, the user wrote an MPS file and then converted it to a generic AMPL model file and data file. As a result, after include mdx_issue.run is invoked, there is a long delay without any output while the lengthy data file is read.