h2g2 The Hitchhiker's Guide to the Galaxy: Earth Edition

I have a matlab question that nobody I've asked seems to know the answer to but ought to be obvious, perhaps the author of this entry can help?

Matlab promotes the use of vectorized code, that is if you wanted to square every element of a vector x you'd write

x = x.^2;

and not

for i=1:length(x)
x(i)=x(i)^2
end

My question is - what do you do if you have a function f whose definition you can't alter which behaves differently if passed a scalar or a vector, and you want to apply f to each element of a vector? Suppose (for the sake of argument) that abs(x) computes the absolute value of x if x is a scalar, but the length of the vector x if x is a vector. Now suppose I wanted to apply abs(x) to each scalar value in a vector x, how do I do it? At the moment, I'm stuck with doing

for i=1:length(x)
x(i) = abs(x(i));
end

but what I want is something like

x = abs(x as vector of scalars)

or something like that. Any way of doing it? In Mathematica I would write x = abs /@ x;

How about defining a function so that it can only take scalar values and not vector values?

Hmmm - I've never come across that problem before. You could try:

x = (abs(x())'

Using the colon operator to look at each scalar in the vector x should get abs to calculate a column of output values, which is then turned into the correct output using the ' operator. It works for abs, but I'm not sure if it'll work for your problem.

Alex

Ok, so it turned my reply into a smiley

Here's my post, but with the offending () replaced with {}

x = (abs(x{:}))'