circshift error in loop

hi, my code is

for k = 1:last_frame-1
    ca=circshift(im,[k,0]);
    da  = im - ca;
    clear ca
    da = da(da(:,4)==0 & da(:,3)==k,:);
    da = da(1:k:end,:);
    msdx(k) = sum(da(:,1).*da(:,1));
    msdy(k) = sum(da(:,2).*da(:,2));
    nmsd(k) = length(da(:,1));
end

here im is 2D array, last_frame is a integer number

??? Error using ==> circshift at 37 Invalid shift type: must be a finite, nonsparse, real integer vector.

what's wrong with this code?

[Matt Fig]

Are you sure that is where the error is occuring? I can see nothing wrong with your use of CIRCSHIFT.

Answer 1

[Matt J]

Just a guess. last_frame was accidentally given a value like NaN or something like that. Hence, k also has a bad value when it is passed to CIRCSHIFT.

The way to find out is to use DBSTOP, or breakpoints.

[Qiuyang]

I both printed out last_frame and k, they are 500 and 1 respectively. What else could be the reason?

[Matt J]

Well, use DBSTOP anyway. When it stops inside circshift again, check the value of the variable p. Then use DBUP to see what value is being passed to circshift as p.