vectorization - how to vectorise an xor operation in matlab -

vectorization - how to vectorise an xor operation in matlab -

i have run next code in profiler in matlab , quite essential me vectorise code sense unnecessary loop.

i have 2 matrices g , source_data. every column in g determine rows need pick source_data , xor them together.

i creating g , source_data using next piece of code

for i=1:10 source_data(i,:)=rand(1,20)<.8; end i=1:15 g(:,i)=rand(10,1)<.9; end

i performing xor operation using loop below:

z=1; while(i<=15) j=1:10 if(g(j,i)==1) intersum(z+1,:)=xor(intersum(z,:), source_data(j,:)); z=z+1; end end c(i,:)=intersum(z,:); i=i+1; end

is there way vectorise code ? time lag acceptable little matrix big matrices code quite in efficient.

thanks,

bhavya

assuming that:

i starts @ 1 intersum starts @ zeros

here's vectorized form of code produces exact same result original:

function c = version_a() source_data = rand(10,20)<.8; g = rand(10,15)<.9; intersum = zeros(1, size(source_data,2)); z = 1; = 1; while <= 15 j=1:10 if(g(j,i)==1) intersum(z+1,:)=xor(intersum(z,:), source_data(j,:)); z=z+1; end end c(i,:)=intersum(z,:); i=i+1; end ret = c; end function c = version_b() source_data = rand(10,20)<.8; % can initialize in single phone call g = rand(10,15)<.9; % same here c = zeros(size(g,2),size(source_data,2)); c(1,:) = mod(sum(source_data(g(:,1),:)),2); = 2:15 c(i,:) = mod(c(i-1,:) + sum(source_data(g(:,i),:)),2); end end

to check timing of both versions used test function:

function ret = xor_test() ret = 0; seed = 123456789; laps = 10000; tic = 1:laps randstream.getdefaultstream.reset(seed); = version_a(); end toc tic = 1:laps randstream.getdefaultstream.reset(seed); b = version_b(); end toc ret = ret + sum(sum(b ~= a)); end

and got next timings on machine:

elapsed time 13.537738 seconds. elapsed time 2.302747 seconds. ans = 0

now why changed way...

a xor operation on array of logicals pretty much checking parity of sum (treating true values 1). furhtermore, intersum beingness used accumulator, there's who's values ends in c skip altogether. taking rows g(j,i) 1 can done logical indexing.

and finally, if don't proposed version, i'd recommend preallocating your c , intersum vectors (in case you're not doing already). has made lot of difference me in past.

matlab vectorization xor