PROC OPTMODEL: Default is dual or primal simplex?

[JMKUO]

Question 2 of 2.1 of the Optimization portion of the Text/TimeSeries/Experiments/Optimization course says the default algorithm for linear programming is dual simplex, while the "Three-Dimensional Example" video states that the default in PROC OPTMODEL is primal simplex. Which is it? Am I conflating two different parts of the program, or is this contradictory?

[WarrenKuhfeld]

It might depend. I believe if you run an example, it should tell you which method it is using.

https://support.sas.com/documentation/onlinedoc/or/143/optmodel.pdf

MODEL chooses a solver that depends on the problem type. Table 5.8 lists the default solver for each problem type.1

Table 5.8 Default Solvers and Algorithms in PROC OPTMODEL

Problem Solver Algorithm Constraint programming CLP Constraint propagation and backtracking search

Linear programming LP Dual simplex

Mixed integer linear programming MILP Branch-and-cut

General nonlinear programming NLP Interior point NLP

Quadratic programming QP Interior point QP

[JMKUO]

Thanks! - So, if asked on certification, for LP in OPTMODEL, the algorithm default is indeed dual simplex, correct?

[WarrenKuhfeld]

I can't verify t anything about the procedure, test or documentation.  All I can do is point you to the documentation.  Perhaps someone more knowledgeable will look at your question and respond.

[TheresaStemler · Accepted solution]

The dual simplex algorithm is the default algorithm for all linear programming problems in Proc OptModel.