The OPTMILP Option Tuner

Example 14.1 Tuning the Default Set of Options for a Single Problem

This example demonstrates how to tune the default set of tuning options for a single problem. The problem is the air05 problem from the MIPLIB 2003 problem set, which is introduced in the section Getting Started: The OPTMILP Option Tuner. The SAS data set that defines the problem (in MPS format) is named air05.

Because you are using only one problem to perform option tuning, you do not need to create a PROBLEMS= data set. Because you are tuning the default set of options, you do not need to create an OPTIONVALUES= data set. The following statements call the OPTMILP option tuner and determine the stopping criterion by specifying the MAXCONFIGS= option instead of the MAXTIME= option:

proc optmilp data=a.air05 maxtime=300;
   tuner maxconfigs=20 printfreq=2 tunerout=out;
   performance nthreads=4;
run;

title "Tuner Output";
proc print data=out(obs=10);
run;

The output data set is shown in Output 14.1.1.

Output 14.1.1: Single Problem with Default Tuning Options: Output

Tuner Output

Obs	RANK	PROBLEM	OBJSENSE	PRESOLVER	PROBE	CONFLICTSEARCH	NODESEL	VARSEL	HEURISTICS	CUTSTRATEGY	CUTGOMORY	CUTMILIFTED	CUTZEROHALF	STATUS	SOLUTION_STATUS	OBJECTIVE	NODES	SOLUTION_TIME
1	0	AIR05	MIN	-1	-1	-1	-1	-1	-1	-1	-1	-1	-1	OK	OPTIMAL	26374	190	61.17
2	1	AIR05	MIN	3	-1	-1	2	-1	2	2	-1	0	0	OK	OPTIMAL	26374	175	51.97
3	2	AIR05	MIN	3	-1	-1	2	-1	2	1	-1	-1	0	OK	OPTIMAL	26374	203	52.92
4	3	AIR05	MIN	3	-1	-1	2	-1	2	1	-1	0	0	OK	OPTIMAL	26374	203	54.79
5	4	AIR05	MIN	3	-1	0	2	-1	2	1	-1	0	0	OK	OPTIMAL	26374	307	55.18
6	5	AIR05	MIN	3	0	-1	2	-1	2	1	-1	0	0	OK	OPTIMAL	26374	203	55.38
7	6	AIR05	MIN	3	-1	-1	2	-1	3	1	-1	0	0	OK	OPTIMAL	26374	189	58.10
8	7	AIR05	MIN	3	-1	-1	-1	-1	2	1	-1	0	0	OK	OPTIMAL	26374	227	60.34
9	8	AIR05	MIN	3	-1	-1	1	-1	2	1	-1	0	0	OK	OPTIMAL	26374	275	69.41
10	9	AIR05	MIN	0	-1	-1	0	-1	0	-1	-1	-1	0	OK	OPTIMAL	26374	105	70.12