[SciPy-User] Optimisation of a rough function

[Charles R Harris]

On Sat, Apr 2, 2011 at 3:25 PM, David Baddeley
<david_baddeley at yahoo.com.au>wrote:

> Thanks everyone! Implicit filtering definitely sounds like it might be a
> step in
> the right direction.
>
> To address some of the previous comments, the goal function is normally
> quite
> costly, making the brute force type of method somewhat impractical. To give
> you
> a bit more insight, I'm transforming a set of (x,y) points, calculating a
> triangularisation of the transformed point set, and calculating the product
> of
> all the edge lengths in that triangularisation. I'm trying to find
> parameters
> for the transformation which minimise this product. The goal surface I
> plotted
> was for a simplified case - 100 points and a simple linear transformation
> (ie 2
> parameters). In normal applications I'll be looking at ~100k points, and a
> non-linear transformation (~10 parameters).
>
>
Map all the points to zero. Problem solved ;)

Chuck
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20110402/d296e9a9/attachment.html>