Origin Version (Select Help-->About Origin): 7.5 Pro
Operating System: WinXP home SP2
Hi,
I am trying to "fit" a graph [x,D(x)] (uneven data point x spacing) with a simple function f(x) (a low-order polynomial), where the rule for fitting is NOT the point-by-poit least square rule, but:
For a known D(x) find f(x) such that
1. at two predefined points xa, xb : f(xa)=D(xa) and f(xb)=D(xb) and
2. f(x) globally minimizes Int[D(x)-f(x), x, xmin, xmax].
The rationale is that f(x) and D(x) are histograms with equal total number of events. D(x) has been created from a largely unknown f(x) by a certain physical process, which conserves this number. Condition 2. above is not enough to get a unique result, that's why we use the boundary condition 1. that for two points the histogram values of f(x) and D(x) are the same.
I need to automate the process of getting D(x)->f(x). Is there any elegant (without programing the whole algorithm) way to ask Origin to do it for me?
Thanks,
Daniel |
Special case of fitting by integral minimization
Posted - 09/23/2005 : 7:01:02 PM
