Re: Combining Conditional Probabilities

V wrote:

Thanks for clarifying this.  It seems finding the bounds
will be much to complicated since I need to make a program
that will come up with approximations efficiently.



a linear objective function subject to linear and
quadratic constraints.



Now I'll need to think of some way of filling in the extra
degrees of freedom with naive assumptions.  If all fails I
will have to resort to:  P(x|a,b,c) = (P(x|a,b) + P(x|b,c))/2
which would be very unfortunate.


I think finding the extrema of the nonlinear function subject to linear constrainnts would be easier, but still perhaps not very helpful. You really have very little information whence to work. Your proposed approximation has no theoretical grounding whatsoever.

--
Stephen J. Herschkorn                        sjherschko@xxxxxxxxxxxx
Math Tutor in Central New Jersey and Manhattan

.