Use interp1 to interpolate a matrix row-wise (2024)

Thank you for your reply but sadly it did not help. In my case, both the x and the y are matrices and I want to run interp1 row for row. In that example, x is a vector. I tried to transpose my y-matrix as it is done in the example but that made no difference. I still get an error saying that the index exceeds the matrix dimension.

Then please post a small running example, which reproduces your problem. I cannot follow your descriptions exactly enough to understand the problem, when you explain it in text form.

The documentation for interp1() is explicit in that x [the first input to interp1()] has to be a vector. I assumed that you had the same x values for each row of your y matrix, and that is what my example does.

If you do not have that, I'm not sure you can do this other than via a for loop. (A quick web search on the keywords suggests that it is not possible.)

I was hoping to avoid yet another loop because it's already quite a few nested loops and the interpolation is one of the most time consuming in the code. Guess I'll have to accept the extra loop.

I do not dare to try another method since I have some stability issues. It's actually not really an interpolation but more of an extrapolation. It's part of a CFD problem where I iterate through lots of cells lots of times so stability is important, even more important than speed. My supervisor has tried many different kinds of methods and he has found interp1 with the flags 'linear' and 'extrap' to be the most stable.

Thanks for all your help.

That's how I solved it but it's not the answer to my question. I would like a way to do exactly that but without the for-loop.

What's wrong with the for loop? It is probably faster, and clearer, than the alternatives.

The values of x change for each row of y.

Ahh.

Then just use a for-loop!

Note that new_c is just the y-intercepts of the line defined by the first 2 points.