a polyhedron() method for Linear Programs

[nathanncohen]

As the title says, this patch implements a .polyhedron() method which returns the Polyhedron object defined by the LP.

And some doc, while we are at it.

Nathann

CC: @dimpase @dcoudert @nthiery @vbraun

Component: linear programming

Author: Nathann Cohen

Reviewer: Dmitrii Pasechnik

Merged: sage-5.9.beta0

Issue created by migration from https://trac.sagemath.org/ticket/14238

[dimpase]

comment:2

I can think of two not always equal things this returns

• the original polyhedron described by the input
• the way it is represented by the backend invoked.

Please clarify in the docs. Also, please add tests for unbounded and empty cases...

[nathanncohen]

comment:3

• the original polyhedron described by the input
• the way it is represented by the backend invoked.

I don't see what you have in mind. The LP variables are internally numbered from 0 to n-1, that is also the case with the variables of a Polyhedron. With this, there is only ony way to define the constraints, isn't it ?

Nathann

[dimpase]

comment:4

Replying to @nathanncohen:

• the original polyhedron described by the input
• the way it is represented by the backend invoked.

I don't see what you have in mind. The LP variables are internally numbered from 0 to n-1, that is also the case with the variables of a Polyhedron. With this, there is only ony way to define the constraints, isn't it ?

actually, you yourself pointed out to me, a while ago, a case where a backend (GUROBI?) does some nontrivial rewriting of a constraint, if I recall right, of the form a<=x_i<=b, resulting in adding a new variable, or something like this.

Dima

[nathanncohen]

comment:5

actually, you yourself pointed out to me, a while ago, a case where a backend (GUROBI?) does some nontrivial rewriting of a constraint, if I recall right, of the form a<=x_i<=b, resulting in adding a new variable, or something like this.

Arggggggggg... Right. I had totally forgotten about that. I will add a warning. I hate these hacks :-/

Nathann

[nathanncohen]

comment:6

Done ! Thanks for noticing that !

Nathann

[jdemeyer]

comment:8

The patch needs a proper commit message.

[jdemeyer]

Reviewer: Dmitrii Pasechnik

[jdemeyer]

comment:9

Attachment: trac_14238.patch.gz

[jdemeyer]

Merged: sage-5.9.beta0