Dear Professor Chinneck, Dear John,
thanks a lot for introducing this important call for discussion, a new thread, as well as for advancements in OR-MS!
My name is Gerhard-Wilhelm Weber from Poznan University of Technology (Poznan, Poland).
Please allow me to provide 3 little ideas in order to (hopefully) support your call and incentive.
1. Among the further experts in LP and related convex programming, there are our colleagues Professor Tibor Illés (tibor.illes@uni-corvinus.hu) and Professor Tamas Terlaky (tat208@lehigh.edu), who might be able to provide further publicly available repositories of infeasible LP models and further suitable infeasible LP models.
2. Related to item 1., and if such contacts were continued, a joint special issue of the respected Journal of Optimization Theory and Applications (editor in chief: Prof. Tamas Terlaky) might be considered on the subject of infeasible LP models and their repositories. This could also lead to further publicity and progress.
3. Perhaps a joint view from LP along with semi-infinite (SI-) LP and infinite (programming) (I-) LP onto infeasible models could lead to some additional and mutual learning. Among the experts and colleagues in SILP there are, e.g., Professor Marco Antonio López (Marco.Antonio@ua.es) and Professor Miguel Angel Goberna (mgoberna@ua.es).
Cordial thanks to you again,
and also to INFORMS which makes possible and maintains this valuable discussion platform.
Kind regards,
best wishes,
Willi (Gerhard-Wilhelm Weber)
------------------------------
Gerhard-Wilhelm Weber
Professor
Poznan University of Technology
Poznan
------------------------------
Original Message:
Sent: 09-05-2024 11:16
From: John Chinneck
Subject: Appeal for Difficult infeasible Linear Programs
An updated repository of infeasible linear programs is needed for testing algorithms that detect and isolate infeasibility. Many of the infeasible models in the current repositories are small and can be analyzed very quickly by modern solvers, often resulting in very small isolations. There is a need for larger and more difficult infeasible LP models to test the limits of the analysis algorithms. These may be take longer to analyze, or may have several infeasibilities, or may result in large isolations.
There are two main existing repositories of infeasible LPs:
- the netlib repository, dating to 1993
https://www.netlib.org/lp/infeas/index.html
- a 2024 Github repository of newer models:
https://github.com/JChinneck/InfeasibleLPs
Two requests:
1. If you are aware of other publicly available repositories of infeasible LP models, please let me know.
2. If you have suitable infeasible LP models to contribute to a public repository, please contribute them to me in MPS format (or another convertible format) at chinneck@sce.carleton.ca
Thanks for your help!
------------------------------
John Chinneck
Distinguished Research Professor Emeritus
Carleton University
Ottawa ON
------------------------------