CODEWITHSUNDEEP
matlabgradientnumerical-methods
shamalaia
shamalaia

Reputation: 2347

cannot reproduce error with gradient

In computing the gradient of variable

[dINTRHOdx,dINTrhody,~] = gradient(INTrho, DELTAx, DELTAy, depth);

I get an error

Index exceeds matrix dimensions.

Error in gradient (line 67) g(2:n-1,:) = (f(3:n,:)-f(1:n-2,:)) ./ (h(3:n) - h(1:n-2));

However, the inputs have coherent dimensions:

size(INTrho)
size(DELTAx)
size(DELTAy)
size(depth)

ans =

    1080         149          52

ans =

    1080           1

ans =

149 1

ans =

52     1

and if I try the following

[dx,dy,~]=gradient(rand(5,5,3),1:5,1:5,1:3)

(gradient computation of variable with coherent dimensions), I get no errors.

Note also that I do not have some variable called gradient:

which gradient

/Applications/MATLAB_R2017a.app/toolbox/matlab/datafun/gradient.m

What could it be the reason of my error?

EDIT: Partial answer

If I make the dimensions exactly the same

DELTAx=repmat(DELTAx, 1,numel(DELTAy),numel(depth));
DELTAy=repmat(DELTAy, size(DELTAx,1),1,numel(depth));
ddepth=repmat(depth,  size(DELTAx,1),size(DELTAy,2),1);

then gradient works

[dINTRHOdx,dINTrhody,~] = gradient(INTrho, DELTAx, DELTAy, ddepth);

But then, why does [dx,dy,~]=gradient(rand(5,5,3),1:5,1:5,1:3) work?

Upvotes: 0

Views: 67

Answers (1)

J. Mel
J. Mel

Reputation: 91

Your problem is the order of your inputs. The reason [dx,dy,~]=gradient(rand(5,5,3),1:5,1:5,1:3) works is because the first two inputs are both of length 5. A more clear example to see how your input order matters is the following.

[dx,dy,~]=gradient(rand(5,6,3),1:6,1:5,1:3)

If you change the order of your original inputs to the following it should work.

[dINTRHOdx,dINTrhody,~] = gradient(INTrho, DELTAy, DELTAx, depth);

Upvotes: 1

Related Questions