Question

To begin, this problem is easily solvable with a for-loop. However, I’m trying to force/teach myself to think vector-wise to take advantage of what Matlab does best.

Simplified, here is the problem explanation:

• I have a vector with data in it.
• I have a 2xN array of start/stop indices that represent ranges of interesting data in the vector.
• I want to perform calculations on each of those ranges, resulting in a number (N results, corresponding to each start/stop range.)

In code, here’s a pseudoexample of what I’d like to have at the end:

A = 1:10000;
startIndicies = [5 100 1000];
stopIndicies = [10 200 5000];
...
calculatedResults = [func(A(5:10)) func(A(100:200)) func(A(1000:5000))]

The length of A, and of the start/stop index array is variable.

Like I said, I can easily solve this with a for loop. However since could be used with a large data set, I’d like to know if there’s a good solution without a for loop.

Answer 1

Here is one possible solution, although, I won’t call it a fully vectorized solution, rather a one liner one.

 out = cellfun(@(i,j) fun(A(i:j)), num2cell(startIndicies), num2cell(stopIndicies) );

or, if you plan to have homogeneous outputs,

 out = arrayfun(@(i,j) fun(A(i:j)), startIndicies, stopIndicies);