calling all MatLab gurus...

[Ether]

I have an M by N matrix A and an M by 1 column vector b.

I want to find the N by 1 column vector x which minimizes sum(abs(A*x-b)).

What is the recommended way to do this in MatLab?

[Jared Russell]

I'd use linprog and solve the same linear program as in the Python thread.

[Ether]

Attached is an XLS spreadsheet with a 256-by-64 A matrix and a 256-by-1 b column vector.

Using your favorite tool or language, find the 64-by-1 column vector x which minimizes sum(abs(A*x-b)). Show your work.

Any takers?

[Ether]

Originally Posted by Ether

Attached is an XLS spreadsheet with a 256-by-64 A matrix and a 256-by-1 b column vector.

Using your favorite tool or language, find the 64-by-1 column vector x which minimizes sum(abs(A*x-b)). Show your work.

Any takers?

If there's anyone out there in CD land working on the above, please use the attached alternate b vector. It's more effective at revealing possible flaws in the solution.

[The Doctor]

Why?

[Ether]

Why what?

[Ether]

Originally Posted by Ether

I have an M by N matrix A and an M by 1 column vector b.

I want to find the N by 1 column vector x which minimizes sum(abs(A*x-b)).

What is the recommended way to do this in MatLab?

Turns out, this is a very interesting problem.

See discussion in this thread:

http://www.chiefdelphi.com/forums/sh...48&postcount=6