Improvements to submodules

[tscrim]

We implement the lift map as a coercion map from a submodule and extend the submodule to work over modules with infinite bases.

CC: @jhpalmieri @nthiery @darijgr @avirmaux @fchapoton @mkoeppe

Component: categories

Keywords: sd109

Author: Travis Scrimshaw

Branch: d8a1c50

Reviewer: Simon Brandhorst

Issue created by migration from https://trac.sagemath.org/ticket/19448

[tscrim]

Branch: public/categories/improve_submodules-19448

[tscrim]

comment:1

One thing that might be slightly controversial is that I construct an explicit ordering of elements used after the echelonization. However, this is needed to be stored in the submodule because if the ordering of the ambient changes (which it can by the set_order method), this could break the triangularity of the lift map. This also allowed me to construct submodules of infinite-dimensional modules.

---

New commits:

98dc756

Improvements to submodules.

[tscrim]

Commit: 98dc756

[sagetrac-git]

Changed commit from 98dc756 to 8006b8e

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

8006b8e

Merge branch 'public/categories/improve_submodules-19448' of trac.sagemath.org:sage into public/categories/improve_submodules-19448

[cheuberg]

comment:4

merge conflict

[sagetrac-git]

Changed commit from 8006b8e to d4838df

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

147583b	Merge branch 'public/categories/improve_submodules-19448' of git://trac.sagemath.org/sage into public/categories/improve_submodules-19448
d4838df	Removing cmp for key.

[tscrim]

comment:6

Actually, the more I think about it, the more I am for fixing the ordering of the submodule basis as the submodule is defined by a matrix, where there is an implicit ordering of the basis already in there.

[simonbrandhorst]

comment:7

failing doctest

[sagetrac-git]

Changed commit from d4838df to 1267467

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

09f8caa	Merge branch 'public/categories/improve_submodules-19448' of git://trac.sagemath.org/sage into public/categories/improve_submodules-19448
1267467	Fixing failures and allowing echelon_form to handle oo-dim modules.

[tscrim]

comment:9

This should now pass all tests. I also extended the echelon_form to handle the case when the ambient module is infinite dimensional.

[mkoeppe]

Changed keywords from none to sd109

[sagetrac-git]

Changed commit from 1267467 to d8a1c50

[sagetrac-git]

Branch pushed to git repo; I updated commit sha1. New commits:

d8a1c50

Fixing doctests and making `_vector_` and from_vector methods more consistent.

[tscrim]

comment:12

So doctests were failing elsewhere because _support_order was also an attribute used by the submodules. So I refactored the method to now be called _compute_support_order. I also found some other issues with to/from_vector not being consistent in their parameters. Doctests that were failing now pass.

[simonbrandhorst]

comment:13

looks reasonable to me.

[simonbrandhorst]

Reviewer: Simon Brandhorst

[vbraun]

Changed branch from public/categories/improve_submodules-19448 to d8a1c50

[mkoeppe]

comment:15

Follow-up discussion in https://groups.google.com/g/sage-devel/c/z1k8IT3ocR4/m/UfkhlDweBwAJ

[mkoeppe]

Changed commit from d8a1c50 to none