How can I loop over a series of vectors - X0, X1, X2, X3... etc - and evaluate a function on each vector?

Hi,

I've tried digging around for a way to do this, but the nearest question to mine was unanswered, so I'm hoping that there's someone who knows how to do this.

So I have 1000 text files containing two columns of data, but different numbers of rows - the (x,y) values for a given data set.

I've managed to import these files into my workspace, where they're stored as vectors and given names from X1 to X1000.

I've created a function - called parameter(x) - that can evaluate one of these vectors and give me a number. I'd like to collect all these numbers together in a vector.

Rather than evaluate parameter 1000 times by hand,I thought if I could loop over the vectors in a script and evaluate parameter(Xi), I could put this result in the ith column of a vector, i.e. bvector[i] = parameter(Xi).

The problem arises when I try to evaluate parameter(Xi). I've been using sprintf to create the vector name and then evaluate:

bvector = [1:1000];   
for i = 1:1000
  name = sprintf('X%d',i)
  j = i+1;
  bvector(j) = parameter(name);
end

I get an error saying that parameter can't be evaluated, and I realised that this is because (e.g) X12 just a name, and has no link to the vector X12.

I've tried varies ways of getting it to do what I want, like using eval(), but I'm a bit new at this, and I'm not sure how to proceed. Any ideas?

Answer 1

[Walter Roberson]

http://matlab.wikia.com/wiki/FAQ#How_can_I_create_variables_A1.2C_A2.2C....2CA10_in_a_loop.3F

[Rehan Deen]

Thanks. I'd seen this a couple of times around, and I applied it to something different before, but I managed to use this to get it to work.

Basically, I used eval(load) to import the files into a workspace, which I then saved, and loaded the workspace onto a structure S. So vector X1 gets called into the loop by:

input = S.(sprintf('X%d',i))

And the rest follows.

Answer 2

[Jan]

Do not use EVAL.

"X1" to "X1000" is a very bad idea, a cell is much more efficient, easier to maintain and expand: "X{1}" to "X{1000}".

Answer 3

[Daniel Shub]

First lets generate some data with problematic naming:

N = 10;
x1 = rand(1, 1);
x2 = rand(2, 1);
x3 = rand(3, 1);
x4 = rand(4, 1);
x5 = rand(5, 1);
x6 = rand(6, 1);
x7 = rand(7, 1);
x8 = rand(8, 1);
x9 = rand(9, 1);
x10 = rand(10, 1);

Now lets work around the naming problem

temp = cellstr(arrayfun(@(n)sprintf('x%d', n), 1:N, 'UniformOutput',false));
save('myx.mat', temp{:});
clear(temp{:});
x = load('myx');

Now you no longer have x1, x2, ..., xN in the workspace, but rather just structure x with fields x1, x2, ..., xN. You can now get what you want in a for loop:

for ii = N:-1:1
  bvector(ii) = parameter(x.(['x', num2str(ii)]));
end