Finding the index of elements of a vector in a mesh quite fast

Hello friends,

I would like to find the index of vector of numbers A being closest in a mesh. I know how to do this but I hope you have a better idea to speed up the calculations.

An example: Consider a mesh of 10^6 regularly spaced numbers in the interval [0 10] and 10^4 random points A in this interval and we wish to find their corresponding index in our mesh (in the sense of being closest).

mesh=linspace(0,10,10^6);

A=0+(10-0).*rand(1,10000);

N=1000;

tic;

for n=1:N

ind=interp1(A,1:length(A),mesh,'nearest');

end

t=toc;

t/N

ans =

0.016923

So, on average it takes around 0.016923 seconds. I think that there should be a very smart way to do the same task with a much reduced computational time.

Any idea is greatly appreciated!

Thanks in advance,

Babak

[Torsten]

Shouldn't it be

ind=interp1(mesh,1:length(mesh),A,'nearest');

instead of

ind=interp1(A,1:length(A),mesh,'nearest');

?

[Mohammad Shojaei Arani]

Yes, of course. I was not carefull

Answer 1

[Torsten]

mesh=linspace(0,10,10^6);
A=0+(10-0).*rand(1,10000);
N=1000;
tic;
for n=1:N
   y = discretize(A,mesh);
   idx = A-mesh(y) >= mesh(y+1)-A;
   y(idx) = y(idx) + 1;
end
t=toc;
t/N

ans = 0.0024

[Mohammad Shojaei Arani]

Great man!

I am happy with this!