nnmf function usage problem

hi all, i am trying to use nnmf function. i have a 3000x4000 sized variable named resim.it has all nonnegative values. i just wrote: [a,b]=nnmf(resim,100);

it gives an error just like this: Error using \ Linear algebra is not supported for integer data types. To compute elementwise LDIVIDE, use LDIVIDE (.\) instead.

Error in nnmf>nnmf1 (line 296) h = max(0, w0\a);

Error in nnmf/loopBody (line 258) [whtry{3},whtry{4},whtry{1}] = ...

Error in internal.stats.parallel.smartForReduce (line 128) reduce = loopbody(iter, S);

Error in nnmf (line 180) whbest = internal.stats.parallel.smartForReduce(...

do you have any idea what the problem is? thank you.

Answer 1

[Tom Lane]

Most Statistics Toolbox functions aren't written to operate on integer data types. Try nnmf(double(resim),100) to convert to double precision. You may be able to use single() instead, if double() needs too much memory.

[ihsan]

thank you very much it worked, as you said it took many time for double but good for single.