Return to the homepage
Search the complete Wavelet Digest database
Help about the Wavelet Digest mailing list
About the Wavelet Digest
The Wavelet Digest
> Volume 3, Issue 12
Question: Matrix Compression

Previous :: Next

Author 
Message 
wvl@ippgarching.mpg.de (Wolfgang von der Linden) Guest

Posted: Mon Dec 02, 2002 1:17 pm Subject: Question: Matrix Compression




Question: Matrix Compression
Dear Colleagues,
is there anybody around, who knows how to compress
a matrix of the form
M_{ij} = e^{a * ((x_ix_j)^2 + (y_iy_j)^2)}
using discrete wavelet transform.
Where (x_i,y_i) are point on a 2d square lattice,
e.g.
x_i = mod(i,256)
y_i = integer(i/256)
The approach, given in Numerical Recipes, gives
merely a compression by 50% if one discards matrix elements
of size 10^{6}, which is worse than the original
matrix and in disagreement to W.Press's statement
that the number of matrix elements, larger than eps, is
10*N*Log_{10}(eps)
Thanks
Wolfgang von der Linden
My email address reads (wvl@ibmop5.ippgarching.mpg.de) 





All times are GMT + 1 Hour

Page 1 of 1 
