Path: news.mathworks.com!not-for-mail
From: "Arvind Iyer" <aiyer@ict.usc.edu>
Newsgroups: comp.soft-sys.matlab
Subject: Using gradient command when explicit function description is not available
Date: Tue, 3 Mar 2009 20:53:03 +0000 (UTC)
Organization: Institute of Creative Technologies
Lines: 7
Message-ID: <gok5bf$iua$1@fred.mathworks.com>
Reply-To: "Arvind Iyer" <aiyer@ict.usc.edu>
NNTP-Posting-Host: webapp-02-blr.mathworks.com
Content-Type: text/plain; charset="ISO-8859-1"
Content-Transfer-Encoding: 8bit
X-Trace: fred.mathworks.com 1236113583 19402 172.30.248.37 (3 Mar 2009 20:53:03 GMT)
X-Complaints-To: news@mathworks.com
NNTP-Posting-Date: Tue, 3 Mar 2009 20:53:03 +0000 (UTC)
X-Newsreader: MATLAB Central Newsreader 1412562
Xref: news.mathworks.com comp.soft-sys.matlab:522228


I require to calculate the gradient of a function of n variables.

However my function is not available explicitly. I just have a table of values and the values of the dependent variable are not equally spaced. In particular, I have a function of 100 variables and I have 10000 data points.

My problem can be stated in general terms as follows: Given a finite number of data points of a function of n-variables, how can I calculate the gradient at any arbitrary point in this (n+1)-dimensional space?

How can the 'gradient' command be adapted to such problems? Are there concerns about the computational speed of the gradient command for such problems? Are there more suitable variants available in the file exchange?