How do I make my code faster. It now takes 2 hrs

Hi, I have a code in which the I have to find the pressure at each point. Hence at each point of the square/rectangular grid has to be found out. this requires the element wise multiplication of an array of scalar values with a 2x2 matrix. Because of this requirement i need to use nested loops and hence it takes a lot of time to compute. Is there an easire way?

[Birdman]

Can you share your code?

[Stephen23]

Use a 3D array? Without your code we have no idea what you are actually doing, or what the required output is, so we are left to guessing and speculating. Upload your code if you want more useful help than just guesses.

[Chirag Anand]

Hi, I attach the m file as the code. Its a function so I also attach the main file. It is not pretty and not well commented

[Chirag Anand]

anybody has any ideas about how to make it faster?

[Jos (10584)]

The most important step: comment the code extensively!

(you will thank me for this a year from now)

[Adam]

doc profile

is the first place to start. Never start trying to optimise code without knowing exactly which parts of the code are actually slow and what the maximum effect you could get from optimising a given part is!

Answer 1

[William Smith]

1) Not clear from your code, but the first step is always to vectorize. Matlab is good at matrix operations (it's not called Mat lab for nothing). You appear to have loops looping through each element of your matrices.

2) Once you've done that, try things like parfor() in the parallel computation toolbox. But only once you've thoroughly vectorized your code.

[Chirag Anand]

Hi, I have tried vectorizing and it does cut down a lot. but i have had to change it because I now need to find a matrix for each value in the grid. So each value has to be multplied with the matrix and the pressure at each point needs to be calculated