Re: [Help-glpk] GLPSOL solver running problem

[Reginald Beardsley]

Nigel's advice is quite sound.  I was rather amused by the problem size.  I 
thought I was pushing the envelope quite a bit with some of my problems.  But 
I'm doing baby problems by comparison.

That said, if glpsol is issuing the "killed" message, it should be more 
explicit about the reason.  If the OS is issuing the message, then gpsol should 
be trapping the error.  My copy of GLPK is out of date, so I'll need to update 
to look into it further.  I've not been using it lately and have slipped behind.

Likely Heinrich or Andrew will supply a more definitive answer.  If you're 
running on Linux or other Unix style system the syscall tracing program (truss 
on Solaris, strace on Linux as I recall) will provide some information.

The man page will give the full details but general usage is:

truss -o truss.out glpsol  [arguments to glpsol]

you can then look at truss.out to see what it was doing just before it died.  
I'd suspect a call to sbreak(2) failing.

Reg

--------------------------------------------
On Sat, 2/21/15, Nigel Galloway <address@hidden> wrote:

 Subject: Re: [Help-glpk] GLPSOL solver running problem
 To: "Le Van Duc" <address@hidden>, address@hidden
 Date: Saturday, February 21, 2015, 7:14 AM
 
 Do
 you have 30GB of memory available? With no idea what you
 have put into all these rows and columns of the constraint
 matrix it is impossible to be specific, but the model is at
 least (definitely I would say at least) 5 times larger than
 anything I would even dream of attempting on my 8GB
 desktop.
 
  
 You are going to have to find some way of starting
 smaller, I would say about 1% of its current size, then
 scale up identifying the resources you are using and
 predicting the maximum you can possibly hope to build.
 
  
 Have you tried using GLPK's API to understand what
 the internal data structures are and possibly finding a way
 of using the API to build this model more efficiently then
 MATHPROG can?
 
  
 --
 
 Nigel Galloway
 
 address@hidden
 
  
 
  
  
 On Sat, Feb 21, 2015, at
 12:23 AM, Le Van Duc wrote:
 
 Dear all, 
 
  
 I have implemented a mixed linear integer programming
 model using GNU MathProg language and run program using
 GLPSOL solver. While the program is running, the result has
 been shown as:  
 
  
 ...$ glpsol --model
 controlCoverageWithConnectivity2.mod --data
 dataInput.dat
 
 GLPSOL: GLPK LP/MIP Solver, v4.45
 
 Parameter(s) specified in the command
 line:
 
  --model
 controlCoverageWithConnectivity2.mod --data
 dataInput.dat
 
 Reading model section from
 controlCoverageWithConnectivity2.mod...
 
 110 lines were read
 
 Reading data section from
 dataInput.dat...
 
 12507527 lines were read
 
 Generating totalInt...
 
 Generating c1...
 
 Generating c2...
 
 Generating c3...
 
 Generating c4...
 
 Generating c5...
 
 Generating c6...
 
 Generating c7...
 
 Generating c8...
 
 Model has been successfully
 generated
 
 GLPK Integer Optimizer, v4.45
 
 12505004 rows, 6252500 columns, 25458088
 non-zeros
 
 2500 integer variables, all of which are
 binary
 
 Preprocessing...
 
 Killed
 
  
 I do not know why the process is suddenly killed as
 above and the program is stopped. Please let me know what is
 problem and how to solve this. 
 
  
 Thank you so much! 
 
 
 
 
 
 
 _______________________________________________
 
 Help-glpk mailing list
 
 address@hidden
 
 https://lists.gnu.org/mailman/listinfo/help-glpk
 
  
 
 -- 
 http://www.fastmail.com - The way an email service should be
 
 
 -----Inline Attachment Follows-----
 
 _______________________________________________
 Help-glpk mailing list
 address@hidden
 https://lists.gnu.org/mailman/listinfo/help-glpk