Scott Raney <[EMAIL PROTECTED]>wrote: >I second the motion: we are already planning to add some features in >this area for the next release and a list of the most useful matrix >operations to have built in would be very handy for planning purposes. This is what I'd like to see for matrices (for matrix, I assume you are referring to 2 dimensional associative array, or will you be creating a new matrix object? 1) a quick method of creating a matrix in memory out of a comma-separated-values (or whatever token the user wants to use) from a field, variable, or data file 2) a quick method of exporting the values from a matrix to a token-separated-values field, variable, or data file - both of these should allow the user to specify whether or not the first row of data contains the column labels and the first items of each row contain the row labels...if labels are specified they should be used in the associative array as myArray[r,c] otherwise some standard value for [r,c] should be used. 3) basic matrix algebra: matrix addition, subtraction, multiplication, division...division can be accompished by finding the inverse of the divisor matrix and then using it in matrix multiplication 4) matrix transposition 5) matrix transformation via square root, fourth root, log (y + 1), natural log, and whatever else the rest of the MetaCard community find useful --- these a feel are probably the MOST important, other things I'd like and could readily use are: ----- 6) matrix sorting and ranking (column-wise, row-wise, and array-wise); my matrix ranking routines are TOO slow and have been major bottlenecks in some of my analyses. 7) normalize a matrix 8) triangularize a matrix, and the inverse: make it symmetrical My jonesLib.mc stack has handlers for all of this if an algorithm or example is needed, however, I'd really like to see these optimized in the MetaCard engine to handle LARGE data sets and allow FAST responses. I've been working with some 2 dimensional arrays that were 300 rows by 300 columns (some multivariate dissimilarity data), and could usually get the job done, but often some of my handlers took a LONG time. Dave Jones Archives: http://www.mail-archive.com/metacard@lists.runrev.com/ Info: http://www.xworlds.com/metacard/mailinglist.htm Please send bug reports to <[EMAIL PROTECTED]>, not this list.
