Asked by Knut
on 18 Sep 2012

This is a post about "how to write neat, efficient, readable MATLAB code". I get the job done by doing work-arounds, but I would rather not do those.

Assume that a, b are two vectors of unknown orientation, and we want to compute something similar to the dot-product.

It might look something like this:

function y = myfunc(a,b)

//make sure that vectors are oriented the same way

y = a(:) .* b(:);

But what if I need to calculate y for elements 2:N? We might do something like this, but I think it clutters the code:

function y = myfunc(a,b)

a = a(2:end)

b = b(2:end)

y = a(:) .* b(:);

I am wishing for a functional form of the colon operator that might look something like this:

function y = myfunc(a,b)

y = colon(a(2:end)) .* colon(b(2:end));

Answer by Daniel Shub
on 18 Sep 2012

I thought that squeeze would do what you want, but apparently it doesn't "work" on row vectors. What would be nice is if squeeze had a row flag. Barring that, you can easily create your own colon function. From

type squeeze

it should be obvious what you need to change.

Sign in to comment.

Answer by Royi Avital
on 10 Oct 2017

Steven Lord
on 10 Oct 2017

We did, a very long time ago. It's called reshape.

A = magic(5);

B = reshape(A, [], 1);

% or to avoid the need to implicitly compute the "missing" size

% inside reshape

C = reshape(A, numel(A), 1);

Jan
on 10 Oct 2017

@Royi: Do you have an example where the mentioned methods are not applicable or "ugly"?

Royi Avital
on 15 Jan 2019

@Jan, I find something like vec(A) to be much more elegant.

Sign in to comment.

Answer by Jan
on 10 Oct 2017

What's wrong with:

y = x(:)

or:

reshape(x, [], 1)

There is a functional form of the x(:) operator:

subsref(x, struct('type', '()', 'subs', {{':'}}))

Walter Roberson
on 17 Oct 2018

Sometimes it is easiest to define an auxillary function, such as

column = @(M) M(:);

after which you can

column(a(2:end)) .* column(b(2:end))

Sign in to comment.

Opportunities for recent engineering grads.

Apply Today
## 0 Comments

Sign in to comment.