I have a computationally expensive fuction of three variables
F(p1,p2,p3) which I need to minimize.

This problem is solved with three-dimensional search algorithm in the
cube with the parties p1 [min_p1.. max_p1]-p2 [min_p2.. max_p2]-p3
[min_p3.. max_p3]. A step of search on p1 - dp1, on p2 - dp2, p3 -
dp3.

How to compute in this case the confidence interval of the values
received by search.

If to take this interval as the half of step of digitization (for
example),  it will be constant value - independent of quantity of
measurements.

thanks in advice,
Konst
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
.                  http://jse.stat.ncsu.edu/                    .
=================================================================

Reply via email to