Benchmarks for Optimization Software

by Hans Mittelmann (mittelmann@asu.edu)

---

This service temporarily halted due to a total lack of direct (NSF) or indirect (NSF) support.

---

The same holds for all 28 optimization solvers we had made available for use through NEOS.

---

Sorry for any inconvenience this may cause.

Suggestions for possible sources of support are welcome.

---

Category I: Several codes, one computer

Note that on top of many benchmarks a link to logfiles is given!

---

LINEAR PROGRAMMING

• Benchmark of commercial LP solvers (5-1-2003)( previous)
• Benchmark of free interior point LP solvers (5-27-2003)
• Benchmark of some simplex-based LP solvers (4-3-2003)

• Supplementary Concorde TSP Benchmarks (3-20-2000)

MIXED INTEGER LINEAR PROGRAMMING

• MILP Benchmark - commercial codes (5-31-2003)( previous)
• MILP Benchmark - free codes (9-26-2003)( previous)

QUADRATIC PROGRAMMING

• Benchmark of commercial and other QP solvers (6-9-2003) ( previous)
• Indefinite-QP Benchmark (6-14-2003)

MIXED INTEGER NONLINEAR PROGRAMMING

• MIQP Benchmark (6-4-2003)
• MINLP Benchmark (small problems)(11-26-2002)
• Some medium-sized MINLP Problems (11-7-2002)

QUADRATICALLY CONSTRAINED QUADRATIC PROGRAMMING

• AMPL-QCQP Benchmark (6-14-2003)

SEMIDEFINITE/SQL PROGRAMMING

• Several SDP codes on problems from SDPLIB (10-31-2003)
• SQL problems from the 7th DIMACS Challenge (8-8-2002)
• Newer SDP/SOCP-codes on the 7th DIMACS Challenge problems(10-31-2003)
• Several SDP codes on sparse and other SDP problems (10-31-2003)
• SOCP (second-order cone programming) Benchmark (5-15-2003)

LARGE-SCALE NONLINEAR SYSTEM BENCHMARK

• Large-scale nonlinear system benchmark (6-15-2003)

GEOMETRIC PROGRAMMING

• Large-Scale Geometric Programming Benchmark (6-16-2003)

BOUND-CONSTRAINED NONLINEAR PROGRAMMING

• Large-Scale Bound-Constrained NLP Benchmark (6-16-2003)

NONLINEAR PROGRAMMING

• AMPL-NLP Benchmark, IPOPT, KNITRO, LOQO, SNOPT & PENNON (6-15-2003)
• Nonsmooth NLP Benchmark (6-16-2003)
• DONLP2 on the COPS problem set (12-6-2000)
• HQP, LANCELOT, MINOS, and SNOPT on the CUTE Testset (12-1-1997)
• COBYLA on the small CUTE Testset (1-24-1997)
• COPILOT and LANCELOT on the CUTE/HS* problems (10-1-1997)

  The testenvironment comprises all of Schittkowski, Hock&Schittkowski plus additional cases

• Results for DONLP2 on testenvironment (1-6-1998)
• Results for DONLP2_D on testenvironment (1-10-1998)
• Results for NPSOL on testenvironment (11-26-1996)
• Results for FFSQP on testenvironment (1-16-1997)
• Results for NLPQL on testenvironment (3-10-1997)
• Summary of NLP benchmarks on testenvironment (3-19-1997)
  • Link to testenvironment&DONLP2(f77)
  • Link to generic testenvironment(f77)

---

Category II: One code, several platforms

---

LINEAR PROGRAMMING

• LP problems with MOSEK on Pentium and SunBlade (10-12-2002)

---

NONLINEAR PROGRAMMING

• Runs of TESTENVIRON&DONLP2 on different architectures(4-3-2003)
  • output-files for these runs

---

Benchmarks performed by others