Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
NCTM or The Math Forum.



Re: Approximating GCF
Posted:
Jul 17, 2014 9:22 PM


On Fri, 18 Jul 2014 00:47:58 +0000, James Waldby wrote:... > Let's say you are given n values y_1 ... y_n. By knowledge about some > process, you know that each y_i is some random noise d_i plus r * x_i, > where r, x_i, and d_i are unknowns except that all the x_i are integers. > You want to compute r. > > There may be a way of writing the problem as a mixed integer linear > programming problem, or it might be analyzed as a clustering problem, > and there might be a continuedfractions method. A Fourier analysis > might have a spike at frequency 1/r or a multiple.
Ignore the bit about Fourier analysis.
> Also consider the following brute force approach: [snipped] The snipped approach is like using a matched filter with r = u/j. See <http://en.wikipedia.org/wiki/Matched_filter> re matched filters.
 jiw



