plotting sparse matrices converts the matrix to a dense matrix

[jasongrout]

Plotting big sparse matrices doesn't even work since it automatically converts the matrix to a dense matrix, instead of calling the spy() function.

CC: @rbeezer @mwhansen @williamstein @sagetrac-wcauchois @robertwb

Component: linear algebra

Author: Jason Grout

Reviewer: David Joyner

Merged: Sage 4.1.1.rc0

Issue created by migration from https://trac.sagemath.org/ticket/6554

[jasongrout]

comment:1

Attachment: trac-6554-sparse-matrix-plot.patch.gz

This took a very long time before, if it was even possible.

        sage: A=random_matrix(ZZ,100000,density=.00001,sparse=True)
        sage: matrix_plot(A,marker=',')

[jasongrout]

comment:2

(To those I'm adding as CC): if you have time, could you review this ticket? This is a simple change that makes plotting sparse matrices possible. Currently, it is very, very slow or is not even really possible to plot large sparse matrices because Sage immediately converts the matrix to a dense matrix.

[jasongrout]

comment:3

It would be really great if this was reviewed in time for the Monday deadline for 4.1.1. This is a simple change that makes plotting sparse matrices possible. Currently, it is very, very slow or is not even really possible to plot large sparse matrices because Sage immediately converts the matrix to a dense matrix.

[wdjoyner]

comment:4

This installs fine (amd64 ubuntu 9.04, sage 4.1.1.alpha0) and I'm running tests now. However, why is it that

sage: B = random_matrix(ZZ, 10, 20, density=.4, sparse=True, x = 10)
sage: matrix_plot(B, cmap='hsv').show(axes=False)

returns a ble-and white scatterplot, but

sage: C = random_matrix(ZZ, 10, 20, x = 10)
sage: matrix_plot(C, cmap='hsv').show(axes=False)

returns a multi-colored plot? The docstring indicates that the colors plotted
indicate the relative difference in sizes between the matrix entries. This seems
to be incorrect, unless I am missing something, in the sparse case. Should a
comment to this effect be added to the docstring?

[jasongrout]

Attachment: trac-6554-matrix-plot-docs.patch.gz

apply on top of previous patch

[jasongrout]

comment:5

Good point; the docstring should be updated. I've attached a small patch that updates the docstring. Can you review this docstring change?

[wdjoyner]

comment:6

Yes, looks good and passes sage -testall (intel macbook, OS 10.4.11) except for

        sage -t  "devel/sage/sage/parallel/decorate.py"
        sage -t  "devel/sage/sage/symbolic/expression.pyx"

They seem unrelated. As far as I am concerned, this gets a positive review.

[sagetrac-mvngu]

comment:7

Merged both patches.

[sagetrac-mvngu]

Reviewer: David Joyner

[sagetrac-mvngu]

Merged: Sage 4.1.1.rc0