computer usage when solving using AMPL/CPLEX

635 views
Skip to first unread message

Luke Li

unread,
Apr 18, 2012, 12:45:52 AM4/18/12
to am...@googlegroups.com
Dear all,

I have constructed a MIP and try to solve it one my 64bit windows computer.
It has a Xeon C...@2.8GHz and 12G RAM.  
 
First I load the mode, and then read the data from office access.
I also checked all data (set, parameter) read into AMPL, and displayed them, which were all right.

Then, when I key in "solve;" command, the RAM usage start to increase, so does the CPU usage.
After 30s, the RAM usage reaches about 11.8G and stays like that, but the CPU usage drops.

The screen shows as below:

AMPL: include model.run;
AMPL: option solver cplexamp;
AMPL: solve;

I waited for about 30mins, but there is no "AMPL" prompt in the next line, then I killed the process.
Does anyone know what is happened? or how can I check if AMPL is working or any error in my model which resulted in this?

Any help would be great.

Regards
Luke


Is there any possible way to check if AMPL is working or some error has occur? 



Robert Fourer

unread,
Apr 20, 2012, 1:14:17 AM4/20/12
to am...@googlegroups.com

MIPs are a fundamentally intractable class of problems, and so it is not surprising that – when the problem is too large or the formulation too weak -- CPLEX may not be able to find a provably optimal solution within the time and memory you have available.  To see what is happening in the CPLEX run after you type "solve", try adding

 

   option cplex_options 'mipdisplay 2';

 

If the "nodes left" keep increasing and the "gap" remains large, then CPLEX is having trouble making progress.

 

Bob Fourer

4...@ampl.com

 

 

From: am...@googlegroups.com [mailto:am...@googlegroups.com]

On Behalf Of Luke Li
Sent: Tuesday, April 17, 2012 11:46 PM
To: am...@googlegroups.com
Subject: [AMPL 5676] computer usage when solving using AMPL/CPLEX

Luke Li

unread,
Apr 22, 2012, 12:32:05 AM4/22/12
to am...@googlegroups.com, 4...@ampl.com
Hi Bob,

Thanks for the advice.
I tried to use student version, just see how many variable in my problem.
Here is what i got from sw.exe:
#################################################################
ampl: option solver cplex;
ampl: option cplex_options 'mipdisplay 2';
ampl: solve;
Error executing "solve" command:

Too much memory used -- 1642201120 bytes; couldn't get 268435456 more.
Highest address used = 4294967295.

sw: 
####################################################################
What does this feedback mean?
 
Also, regarding to the formulation, is there any paper or book about it?
What kind formulation is considered weak? 

Regards
Luke

Robert Fourer

unread,
Apr 23, 2012, 10:31:35 AM4/23/12
to am...@googlegroups.com

The message you're seeing indicates that AMPL needs more memory to generate this problem than it can get from your computer.  The student version is a 32-bit binary and so it cannot take advantage of more than 2^32 = 4294967296 bytes of memory, even if there is more memory on your computer.  (Other versions of AMPL are also available in 64-bit binaries that can don't have any practical memory limitation.)

 

For suggestions on troubleshooting problems with insufficient memory, see question 3.6 at www.ampl.com/FAQ.  Also for CPLEX you should consult the discussion of Common Difficulties beginning on page 74 of www.ampl.com/BOOKLETS/amplcplex122userguide.pdf.

 

Most often a "weak" formulation is one that gives poor bounds when integrality is relaxed, or that has a lot of different but equivalent solutions ("symmetry"), or that defines too many zero-one variables.  If you are going to be solving really hard integer programs then it is worth studying some texts and papers on the methods used, but I do not have a recommendation for one reference that surveys everything that is currently implemented in programs like CPLEX.

On Behalf Of Luke Li
Sent: Saturday, April 21, 2012 11:32 PM
To: am...@googlegroups.com
Cc: 4...@ampl.com
Subject: Re: [AMPL 5698] computer usage when solving using AMPL/CPLEX

Luke Li

unread,
Apr 25, 2012, 5:12:49 AM4/25/12
to am...@googlegroups.com
Hi Bob,

Thank you very much for the help.

Regards
Luke
Reply all
Reply to author
Forward
0 new messages