forum.alglib.net

I've been trying to read the documentation, and while it lists each of the functions, I haven't been able to find a procedure to actually apply one to my current problem. I'm also at a bit of a disadvantage because I'm an embedded systems programmer who hasn't done much of this math since college 45 years ago. Alglib is a class of tool that I haven't yet used, and all of the examples seem to assume a familiarity that I don't have.

I need some help on the procedure to get from what we have, to a functional curve fit. I expect that there is some very basic item that I'm missing.

What we have is:

a one dimensional C++ vector of several thousand points of experimental data.

A structure, could be an array) describing the conditions, things like temperature, pressure, etc.
A structure of the values that we need to fit

Several functions using the above two structures to simulate what we expect to get, and I can combine them into a single function that calculates intermediate vectors, of which we can look at an individual point. Since there are multiple fitting values involved, it seems that for every point in the data, this would require calculating entire vectors every time, which makes the process very slow, too slow by several orders of magnitude. I did manage to get something to build and run, but the first pass (totally ignoring any optimization) takes on the order of half an hour, when we need it to be on the order of half a second.

I need some help with the procedure for taking what I've got and feeding it to alglib to get a fit to the data.

Thanks