Optimize sampling times

I want to fit a nonlinear function f(t;x) to set of data at time
points t[1-10] by varying the parameters x. I know the data will be
corrupted by Gaussian noise and I want my estimate of one of the
fitted parameters x[1] to have the minimum variance. I am free to
choose the time points.

Is there a clever solution to this problem? Right now I'm using brute
force simulations to minimize variance of x[1] w.r.t the time-points.
However I'm getting the estimate of the variance from n=10^3 trials
and it's subject to error so the optimization tends to terminate at a
different solution each time I run the program.

Is there a cleverer way than increasing n?
.

Relevant Pages

  • Re: Optimize sampling times
    ... corrupted by Gaussian noise and I want my estimate of one of the ... fitted parameters xto have the minimum variance. ... Is there a clever solution to this problem? ... force simulations to minimize variance of xw.r.t the time-points. ...
    (sci.math.num-analysis)
  • Re: Design for Extension
    ... > B) How to encapsulate that quality. ... > Varying necessarily means, "Varying with time," so unless you're ... The variance can occur across several axes. ... I would have no idea where to put object boundaries. ...
    (comp.lang.java.programmer)