Project

General

Profile

Feature #826

Sparse matrices

Added by John Abbott over 8 years ago. Updated over 8 years ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
Data Structures
Target version:
CoCoALib-1.0
Start date:
26 Nov 2015
Due date:
% Done:

0%

Estimated time:
Spent time:
0.50 h

Description

Mario now thinks he wants sparse matrices.

We need a clear description of the requirements.

History

#1 Updated by John Abbott over 8 years ago

I had a quick look on Wikipedia for information about sparse matrices. There they distinguished between representations which are convenient for input, and representations which are convenient for computing (presumably matrix*matrix products).

The main representations for computing are "compressed row storage" and "compressed column storage"; one is just the transpose of the other. Here is a quick summary of "compressed row storage". The repr comprises 3 lists/arrays/vectors:
  • VAL: array of non-zero values in the matrix
  • COL: array of the col indexes of those values
  • ROW1: array of first indexes into VAL, k-th entry is index of 1st entry in VAL which comes from row k

Also available in: Atom PDF