The OPTLP Procedure

Example 12.7 Finding an Irreducible Infeasible Set

This example demonstrates the use of the IIS= option to locate an irreducible infeasible set. Suppose you want to solve a linear program that has the following simple formulation:

$\begin{array}{rllllllcrc} \mbox{min} & & x_1 & + & x_2 & + & x_3 & & & (\mbox{cost})\\ \mbox{ subject to }& & x_1 & + & x_2 & & & \geq & 10 & (\mbox{con1})\\ & & x_1 & & & + & x_3 & \leq & 4 & (\mbox{con2})\\ & 4 \leq & & & x_2 & + & x_3 & \leq & 5 & (\mbox{con3})\\ & & & & & x_1, & x_2 & \geq & 0 & \\ & & & & 0 & \leq & x_3 & \leq & 3 & \\ \end{array}$

The corresponding MPS-format SAS data set is as follows:


/* infeasible */
data exiis;
   input field1 $ field2 $ field3 $ field4 field5 $ field6;
datalines;
NAME     .        .        .        .        .
ROWS     .        .        .        .        .
 N       cost     .        .        .        .
 G       con1     .        .        .        .
 L       con2     .        .        .        .
 G       con3     .        .        .        .
COLUMNS  .        .        .        .        .
.        x1       cost     1        con1     1
.        x1       con2     1        .        .
.        x2       cost     1        con1     1
.        x2       con3     1        .        .
.        x3       cost     1        con2     1
.        x3       con3     1        .        .
RHS      .        .        .        .        .
.        rhs      con1     10       con2     4
.        rhs      con3     4        .        .
RANGES   .        .        .        .        .
.        r1       con3     1        .        .
BOUNDS   .        .        .        .        .
UP       b1       x3       3        .        .
ENDATA   .        .        .        .        .
;

It is easy to verify that the following three constraints (or rows) and one variable (or column) bound form an IIS for this problem.

$\begin{array}{lllllcrc} x_1 & + & x_2 & & & \geq & 10 & (\mbox{con1})\\ x_1 & & & + & x_3 & \leq & 4 & (\mbox{con2})\\ & & x_2 & + & x_3 & \leq & 5 & (\mbox{con3})\\ & & & & x_3 & \geq & 0 & \\ \end{array}$

You can use the IIS= ON option to detect this IIS by using the following statements:

proc optlp data=exiis
   iis=on
   primalout=iis_vars
   dualout=iis_cons
   logfreq=1;
run;

The OPTLP procedure outputs the detected IIS to the data sets specified by the PRIMALOUT= and DUALOUT= options, then stops. The notes shown in Output 12.7.1 are printed to the log.

Output 12.7.1: The IIS= Option: Log

NOTE: The OPTLP procedure is executing in single-machine mode.

NOTE: The problem has 3 variables (0 free, 0 fixed).

NOTE: The problem has 3 constraints (1 LE, 0 EQ, 1 GE, 1 range).

NOTE: The problem has 6 constraint coefficients.

NOTE: The IIS option is enabled.

Objective Entering Leaving

Phase Iteration Value Time Variable Variable

P 1 1 1.400000E+01 0 x2 con3 (S)

P 1 2 5.000000E+00 0 x1 con2 (S)

P 1 3 1.000000E+00 0

NOTE: The IIS option found the problem to be infeasible.

NOTE: Applying the IIS sensitivity filter.

NOTE: The sensitivity filter removed 1 constraints and 3 variable bounds.

NOTE: Applying the IIS deletion filter.

NOTE: Processing constraints.

Processed Removed Time

0 0 0

1 0 0

2 0 0

3 0 0

NOTE: Processing variable bounds.

Processed Removed Time

0 0 0

1 0 0

2 0 0

3 0 0

NOTE: The deletion filter removed 0 constraints and 0 variable bounds.

NOTE: The IIS option found the problem to be infeasible.

NOTE: The IIS option found an irreducible infeasible set with 1 variables and 3

constraints.

NOTE: The IIS solve time is 0.00 seconds.

NOTE: The data set WORK.IIS_VARS has 3 observations and 10 variables.

NOTE: The data set WORK.IIS_CONS has 3 observations and 10 variables.

The data sets iis_cons and iis_vars are shown in Output 12.7.2.

Output 12.7.2: Identify Rows and Columns in the IIS

Constraints in the IIS

Obs	Objective Function ID	RHS ID	Constraint Name	Constraint Type	Constraint RHS	Constraint Lower Bound	Constraint Upper Bound	Dual Variable Value	Constraint Status	Constraint Activity
1	cost	rhs	con1	G	10	.	.	.	I_L	.
2	cost	rhs	con2	L	4	.	.	.	I_U	.
3	cost	rhs	con3	R	.	4	5	.	I_U	.

Variables in the IIS

Obs	Objective Function ID	RHS ID	Variable Name	Variable Type	Objective Coefficient	Upper Bound	Variable Value	Variable Status	Reduced Cost
1	cost	rhs	x1	N	1	1.7977E308	.		.
2	cost	rhs	x2	N	1	1.7977E308	.		.
3	cost	rhs	x3	D	1	3	.	I_L	.

The constraint $x_2 + x_3 \leq 5$ , which is an element of the IIS, is created by the RANGES section. The original constraint is con3, a " $\geq$ " constraint with an RHS value of 4. If you choose to remove the constraint $x_2 + x_3\leq 5$ , you can accomplish this by removing con3 from the RANGES section in the MPS-format SAS data set exiis. Since con3 is the only observation in the section, the identifier observation can also be removed. The modified LP problem is specified in the following SAS statements:

 /* dropping con3, feasible */
data exiisf;
   input field1 $ field2 $ field3 $ field4 field5 $ field6;
datalines;
NAME     .        .        .        .        .
ROWS     .        .        .        .        .
 N       cost     .        .        .        .
 G       con1     .        .        .        .
 L       con2     .        .        .        .
 G       con3     .        .        .        .
COLUMNS  .        .        .        .        .
.        x1       cost     1        con1     1
.        x1       con2     1        .        .
.        x2       cost     1        con1     1
.        x2       con3     1        .        .
.        x3       cost     1        con2     1
.        x3       con3     1        .        .
RHS      .        .        .        .        .
.        rhs      con1     10       con2     4
.        rhs      con3     4        .        .
BOUNDS   .        .        .        .        .
UP       b1       x3       3        .        .
ENDATA   .        .        .        .        .
;

Since one element of the IIS has been removed, the modified LP problem should no longer contain the infeasible set. Due to the size of this problem, there should be no additional irreducible infeasible sets. You can confirm this by submitting the following SAS statements:

proc optlp data=exiisf
   pout=po
   iis=on;
run;

The notes shown in Output 12.7.3 are printed to the log.

Output 12.7.3: The IIS= Option: Log

NOTE: The OPTLP procedure is executing in single-machine mode.

NOTE: The problem has 3 variables (0 free, 0 fixed).

NOTE: The problem has 3 constraints (1 LE, 0 EQ, 2 GE, 0 range).

NOTE: The problem has 6 constraint coefficients.

NOTE: The IIS option is enabled.

Objective

Phase Iteration Value Time

P 1 1 1.400000E+01 0

P 1 3 0.000000E+00 0

NOTE: The IIS option found the problem to be feasible.

NOTE: The IIS solve time is 0.00 seconds.

NOTE: The data set WORK.EXSS has 8 observations and 3 variables.

NOTE: The data set WORK.PO has 3 observations and 10 variables.

The solution summary and the primal solution are displayed in Output 12.7.4.

Output 12.7.4: Infeasibility Removed

Solution Summary

Obs	Label1	cValue1	nValue1
1	Solver	LP	.
2	Algorithm	Primal Simplex	.
3	Objective Function	cost	.
4	Solution Status	Feasible	.
5			.
6	Iterations	3	3.000000
7	Presolve Time	0.00	0
8	Solution Time	0.00	0

Primal Solution

Obs	Objective Function ID	RHS ID	Variable Name	Variable Type	Objective Coefficient	Upper Bound	Variable Value	Reduced Cost
1	cost	rhs	x1	N	1	1.7977E308	.	.
2	cost	rhs	x2	N	1	1.7977E308	.	.
3	cost	rhs	x3	D	1	3	.	.