[SciPy-user] optimization advice needed

[Gael Varoquaux]

On Sun, Jan 27, 2008 at 06:27:40AM -0500, Neal Becker wrote:
> The problem is to minimize the mean-square-error between a sequence of noisy
> observations and a model.
> [...]
> 1) The problem is not to optimize the estimates (a',b') one time - it is
> more of an optimal control problem.  (a,b) are slowly varying, and we want
> to continuously refine the estimates.

This sounds exactly what a Bayesian estimator is good at. I can give you
a good explanation of Bayesian estimation because of lack of time, and
the info on the net is definitely sparse (try wikipedia, but the article
is hard to read). It is good to know that if your problem is linear, and
your noise is Gaussian, you can use a Kalman filter
(http://www.scipy.org/Cookbook/KalmanFiltering).

HTH,

Gaël