Constructing full matrix like a sparse one

Question

Ife 2023-8-3

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/2004242-constructing-full-matrix-like-a-sparse-one

评论： Ife 2023-8-3

I was wondering if there was a way to construct a full matrix like you would a sparse matrix.

I'm working with single photon data so my images are histograms of counts that I read from the bits in my data stream. Due to how sparse works I've found that it is faster to do my histogramming using sparse with 4 inputs, S = sparse(i,j,v,m,n), as it sums values that have the same indicies but this is innefficent as depending on how I want the data read it results in arrays with very low sparsity. I was wondering if there was an equivalent way to do this immediately into a full array.

Here's the relevant part of my code

% number of exposures for macro time
frameNos = 0:macroFrames:NFrames;
fBins = uint32(discretize(f,frameNos));
nExps = NFrames/macroFrames;
% generates sparse array Y*X*B x F
bin1 = y+(x-1).*NPixHeight+(b-1).*NPixHeight.*NPixWidth;
bin2 = fBins;
SpCounts = sparse(bin1,bin2,1,NPixHeight*NPixWidth*NBins,nExps);

Generally macroFrames is large so SpCounts is not sparse and therefore very memory inefficient. I imagine appending full() would fix my memory problem but I'm curious if going straight to a full array would be faster?