0

I installed the GCG and built the soft link with hMETIS in the scipoptsuite/gcg directory:

ubuntu18:~/Documents/Software/scipoptsuite-6.0.0/gcg$ ln -s /home/yang/Documents/Software/scipoptsuite-6.0.0/hmetis-2.0pre1/Linux-x86_64/hmetis2.0pre1 hmetis

When I run the Example test of /check/instances/cs/TEST0055.lp, there are some differences compared to logs of http://gcg.or.rwth-aachen.de/doc/EXAMPLE.html which use the same TEST0055.lp document:

Presolving Time: 0.01
start creating seeedpool for current problem 
created seeedpool for current problem, n detectors: 25 
 Consclassifier "nonzeros" yields a classification with 2  different constraint classes 
 Consclassifier "constypes" yields a classification with 2 different constraint classes 
 Consclassifier "constypes according to miplib" yields a classification with 2 different constraint classes 
 Consclassifier "constypes according to miplib" is not considered since it offers the same structure as "constypes" consclassifier
 Varclassifier "vartypes" yields a classification with 2 different variable classes
 Varclassifier "varobjvals" yields a classification with 2 different variable classes
 Varclassifier "varobjvalsigns" yields a classification with 2 different variable classes
 Varclassifier "varobjvalsigns" is not considered since it offers the same structure as "varobjvals"
Begin of detection round 0 of 1 total rounds 
Start to propagate seeed with id 1 (0 of 1 in round 0) 
 in dec_consclass: there are 2 different constraint classes   
  the current constraint classifier "nonzeros" consists of 2 different classes   
  the current constraint classifier "constypes" consists of 2 different classes   
 dec_consclass found 6 new seeeds 
dec_densemasterconss found 1 new seeed 
sh: 1: zsh: not found
[src/dec_hrgpartition.cpp:314] ERROR: Calling hmetis unsuccessful! See the above error message for more details.
[src/dec_hrgpartition.cpp:315] ERROR: Call was zsh -c "hmetis gcg-r-1.metis.1l488e 20 -seed 1 -ptype rb -ufactor 5.000000 > /dev/null"
sh: 1: zsh: not found
[src/dec_hrgpartition.cpp:314] ERROR: Calling hmetis unsuccessful! See the above error message for more details.
[src/dec_hrgpartition.cpp:315] ERROR: Call was zsh -c "hmetis gcg-r-1.metis.1l488e 10 -seed 1 -ptype rb -ufactor 5.000000 > /dev/null"
sh: 1: zsh: not found
[src/dec_hrgpartition.cpp:314] ERROR: Calling hmetis unsuccessful! See the above error message for more details.
[src/dec_hrgpartition.cpp:315] ERROR: Call was zsh -c "hmetis gcg-r-1.metis.1l488e 29 -seed 1 -ptype rb -ufactor 5.000000 > /dev/null"
Detecting Arrowhead structure: 20 10 29 done, 0 seeeds found.
Start finishing of partial decomposition 1.

The objective value is same as the example on GCG's web. But solutions are different. Why do these errors appear? Is there anything wrong with the GCG or SCIP software? Another special issue is that: the number of Solving Nodes is only '1' on my test, however this number is '82' on the http://gcg.or.rwth-aachen.de/doc/EXAMPLE.html example. I also run the instance of 'bpp/N1C1W4_M.BPP.lp', and above errors happen also.

Begin of detection round 0 of 1 total rounds 
Start to propagate seeed with id 39 (0 of 1 in round 0) 
 in dec_consclass: there are 1 different constraint classes   
  the current constraint classifier "nonzeros" consists of 2 different classes   
 dec_consclass found 3 new seeeds 
dec_densemasterconss found 1 new seeed 
sh: 1: zsh: not found
[src/dec_hrgpartition.cpp:314] ERROR: Calling hmetis unsuccessful! See the above error message for more details.
[src/dec_hrgpartition.cpp:315] ERROR: Call was zsh -c "hmetis gcg-r-39.metis.wDKr6U 50 -seed 1 -ptype rb -ufactor 5.000000 > /dev/null"
sh: 1: zsh: not found
[src/dec_hrgpartition.cpp:314] ERROR: Calling hmetis unsuccessful! See the above error message for more details.
[src/dec_hrgpartition.cpp:315] ERROR: Call was zsh -c "hmetis gcg-r-39.metis.wDKr6U 51 -seed 1 -ptype rb -ufactor 5.000000 > /dev/null"
Detecting Arrowhead structure: 50 51 done, 0 seeeds found.

And it is strange that the number of Solving Nodes is still 1.

SCIP Status        : problem is solved [optimal solution found]
Solving Time (sec) : 0.72
Solving Nodes      : 1
Primal Bound       : +4.10000000000000e+01 (3 solutions)
Dual Bound         : +4.10000000000000e+01
Gap                : 0.00 %
Qidong
  • 11
  • 2
  • Without knowing the details here, I assume that you have to install `zsh` first. Try `sudo apt-get install zsh` and rerun the example. – Gregor Dec 19 '18 at 15:21
  • 1
    As Gregor says, you need zsh for the hmetis call. Additionally, the example explicitly states 'The exact performance varies amongst different architectures, operating systems, and so on. Do not be worried if your installation needs more or less time or nodes to solve.'. This is even more true as the example was prepared using GCG 1.0 and you are now using the newest 3.0 version. Thus, a different solver performance is to be expected. – Gerald Dec 19 '18 at 15:46

0 Answers0