On the importance of programming practices to simplify the implementation of OR algorithms

EasyChair Preprint no. 2529

Abstract

This is the talk that starts the session "On the best programming practices in OR" I recall from the description of  this session:

>The designer of C++ explained that a motivation for creating the new
>language regards the possibility of alleviating the suffering associated
>with certain tasks. "I swore never again to attack a problem with tools as
>unsuitable as I had suffered while designing and implementing the simulator
>[or a random task]." The new language was able to bring relief to millions of
>people. Keeping things in perspective (and in their proper proportions), this
>session is animated in the same spirit.  For example, those who have suffered
>(during their PhD) using unsuitable tools/languages for their problems are
>invited to present their experiences and, above all their solutions.  

These principles can be turned in a more general sense. The scientific world (in OR) finds itself in the grip of a negative phenomenon that I try to describe as follows.  Many theoretical science mega-stars think that the programming tasks belong to the 'small things' that are needed to produce numerical results.  Such theoretical scientists who have never seen an algorithm in execution may have no deep insight into how algorithms work in in real life; e.g., many of them may have ambitious speeches focusing on the "big ideas", without realizing that the dynamics and the efficiency of an algorithm can easily change if one modifies a "small" parameter. On the other hand, many down-to-earth programmers who see algorithms running every day don't always have the conceptual tools and the good analysis grids to dismantle certain traps of the system, which prevents them from creating a very accurate global vision of their field. As long as this fracture persists, I believe that the vitality of the scientific world will remain broken.

Keyphrases: beaux logiciels, C++ wrapper, Cutting Planes Implementation

@Booklet{EasyChair:2529,
  author = {Daniel Porumbel},
  title = {On the importance of programming practices to simplify the implementation of OR algorithms},
  howpublished = {EasyChair Preprint no. 2529},

  year = {EasyChair, 2020}}