matlab says busy but isn't doing anything while running a script

Here's my code:

%Reading in data into two matrices from the knnsearch function
%to be used in determining the smoothing length to give a particle
function filename = neighborsearch(tracer_x,tracer_y)
    IDX(1:4999, 1:4000, 1:10) = 0;
    D(1:4999, 1:4000, 1:10) = 0;

      %The IDX and Distance matrices are initialized
      %
          for i=1:4999 %looping through all frames
              [IDX(i,:,:), D(i,:,:)] = knnsearch([(tracer_x(i,:))',(tracer_y(i,:))'], [(tracer_x(i,:))', (tracer_y(i,:))'], 'k', 10);
          end

end

I'm trying to output two matrices with particle id's and distances to nearest neighbors (1-10 in this, but 1-15 would be preferred). I changed it to n_max = 10 because when I tried to execute the script with 15 neighbors I got a "OUT OF MEMORY" error; So when I try to run it with 10, matlab says busy at the bottom but nothing happens, nothing is output or saved into the workspace, nada. Have I simply written the code wrong or is this pushing it on my available memory? I'll include the memory output just in case. I'm working on Windows 64-bit, Matlab 64-bit R2015a.

Thanks!

>> memory
Maximum possible array:        4433 MB (4.648e+09 bytes) *
Memory available for all arrays:        4433 MB (4.648e+09 bytes) *
Memory used by MATLAB:        1755 MB (1.840e+09 bytes)
Physical Memory (RAM):        4000 MB (4.194e+09 bytes)

*  Limited by System Memory (physical + swap file) available.

[Adam]

How do you know it isn't doing anything?

Without looking too deeply at your code I imagine it is processing, but you are giving it a large amount to do so it takes a long time. Neighbourhood searches are often slow, especially if not programmed with any optimisations.

[Carrie Elliott]

Adam, I thought the same but it finally gave the input prompt >> again without the matrices added to the workspace.

[Adam]

Well they wouldn't be unless you have a breakpoint set as that is a function. The matrices will go out of scope when it completes and only 'filename' would be returned, except that doesn't appear to be assigned in the function so unless you have only posted a small part of the function it should crash at that point.

[Carrie Elliott]

No, this isn't a portion, this is all of it - so what am I missing?

[Image Analyst]

Like Adam said, you never assigned anything for filename so if you ever got to the point where the function exited, it would throw an error because you never assigned filename as the function declaration

function filename = neighborsearch(tracer_x,tracer_y)

requires. Also, if we're to run this, you must tell us what values you supplied for tracer_x and tracer_y. How else can we reproduce what you did?????

[Carrie Elliott]

My bad... They're two 4999x4000 matrices so I didn't include them because they were huge ; it's a good chunk of data, like ~200 MB so I'll have to throw it on my dropbox account.

Tracer X, Tracer Y

[Image Analyst]

How long does it go before you decide it's not doing anything with this large chunk of data? If you're doing knnsearch, it probably does take a long time. Chances are you can do it on a subset of the data and get results just as good.

[Star Strider]

One thing I do for long simulations is to put in an fprintf call as the last statement before the end of the loop:

fprintf('\tLoop: %4d at %4d-%02d-%02d %02d:%02d:%06.3f\n', i, datevec(now))

It will tell you where it is in the loop, about how long each loop is taking, and probably won’t add significant overhead. (I usually add an etime call, but then I like details.)

Answer 1

[Kelly Kearney]

You need to assign the proper output variables. My guess is you're calling the function by typing

neighborsearch(tracer_x,tracer_y)

at the command line. Because you don't request any output variables, the function doesn't throw an error, and runs perfectly fine. On the other hand, if you had typed

filename = neighborsearch(tracer_x,tracer_y)

as the function syntax implied you may have wanted, it would throw the error everyone is talking about above.

But to get back to your original problem, I think your function is working fine, but after it completes its calculations, it throws them away because you didn't tell it to do anything with the new matrices. You need to make sure to return the output you actually want, which I believe are the IDX and D variables.

Change the function itself to:

function [IDX, D] = neighborsearch(tracer_x,tracer_y)

IDX(1:4999, 1:4000, 1:10) = 0;
D(1:4999, 1:4000, 1:10) = 0;

for i=1:4999 
    [IDX(i,:,:), D(i,:,:)] = knnsearch(...
              [(tracer_x(i,:))',(tracer_y(i,:))'], ...
              [(tracer_x(i,:))', (tracer_y(i,:))'], 'k', 10);
end

and then call it as:

[IDX, D] = neighborsearch(tracer_x, tracer_y);