Writing a loop "for"

Hey, i have a problem with writing a loop, which will be able to do this:

 phi(1)=a_locs(1)/((a_odd(1)));
 phi(2)=a_locs(2)/((a_odd(1)));
 phi(3)=a_locs(3)/((a_odd(1)));
 phi(4)=a_locs(4)/((a_odd(1)));
 phi(5)=(a_locs(5)-a_odd(1))/((a_odd(2)));
 phi(6)=(a_locs(6)-a_odd(1))/((a_odd(2)));
 phi(7)=(a_locs(7)-a_odd(1))/((a_odd(2)));
 phi(8)=(a_locs(8)-a_odd(1))/((a_odd(2)));
 phi(9)=(a_locs(10)-(a_odd(1)+a_odd(2)))/((a_odd(3)));
 phi(10)=(a_locs(11)-(a_odd(1)+a_odd(2)))/((a_odd(3)));

Someone can help me?

[James Tursa]

Why do you want a loop? Is your real problem actually bigger than this? Is everything shown variables, or are some of them functions?

[Michael Monka]

All are variables, and i have to evaluate a large number (450) of equations, which looks as i write before, and i thought it is possible to write (one or two?) loops which will be do that thing...?

[James Tursa]

So what are the sizes of the variables involved, and can you be more explicit in the pattern of the calculations for these larger sizes?

[Image Analyst]

The thing I can't figure out is why a_locs(9) is totally missing from any of the right hand side expressions. And there are not enough equations to figure out which elements need to be "skipped" in the other equations, and what equations the "skipping" would happen at.

[Stephen23]

It seems like this might be able to be written using vectorized code, thus avoiding any loops... but it would be nice to know the general algorithm here.

[Michael Monka]

Okay, so the size of variable which i have to obtain (which is phi) is 450. The pattern of the calculations goes like:

 phi(1)=a_locs(1)/((a_odd(1)));
 phi(2)=a_locs(2)/((a_odd(1)));
 phi(3)=a_locs(3)/((a_odd(1)));
 phi(4)=a_locs(4)/((a_odd(1)));

 phi(5)=(a_locs(5)-a_odd(1))/((a_odd(2)));
 phi(6)=(a_locs(6)-a_odd(1))/((a_odd(2)));
 phi(7)=(a_locs(7)-a_odd(1))/((a_odd(2)));
 phi(8)=(a_locs(8)-a_odd(1))/((a_odd(2)));

 phi(9)=(a_locs(9)-(a_odd(1)+a_odd(2)))/((a_odd(3)));
 phi(10)=(a_locs(10)-(a_odd(1)+a_odd(2)))/((a_odd(3)));
 phi(11)=(a_locs(11)-(a_odd(1)+a_odd(2)))/((a_odd(3)));
 phi(12)=(a_locs(12)-(a_odd(1)+a_odd(2)))/((a_odd(3)));

 phi(13)=(a_locs(13)-(a_odd(1)+a_odd(2)+a_odd(3)))/((a_odd(4)));
 phi(14)=(a_locs(14)-(a_odd(1)+a_odd(2)+a_odd(3)))/((a_odd(4)));
 phi(15)=(a_locs(15)-(a_odd(1)+a_odd(2)+a_odd(3)))/((a_odd(4)));
 phi(16)=(a_locs(16)-(a_odd(1)+a_odd(2)+a_odd(3)))/((a_odd(4)));

 phi(17)=(a_locs(17)-(a_odd(1)+a_odd(2)+a_odd(3)+a_odd(4)))/((a_odd(5)));
 phi(18)=(a_locs(18)-(a_odd(1)+a_odd(2)+a_odd(3)+a_odd(4)))/((a_odd(5)));
 phi(19)=(a_locs(19)-(a_odd(1)+a_odd(2)+a_odd(3)+a_odd(4)))/((a_odd(5)));
 phi(20)=(a_locs(20)-(a_odd(1)+a_odd(2)+a_odd(3)+a_odd(4)))/((a_odd(5)));

 phi(21)=...

[Jan]

@Michael: It looks like phi consists of blocks of the length 4. But 450 is not evenly divisable by 4?

Answer 1

[Jan]

With a loop:

phi = zeros(1, 452);  % Instead of 450?
s   = 0;
m   = 1;
for k = 1:4:452
  phi(k:k+4) = (a_locs(k:k+4) - s) / a_odd(m);
  s = s + a_odd(m);
  m = m + 1;
end

And vectorized - assuming that a_odd is a row vector:

phi = reshape(a_locs, 4, 113);
phi = bsxfun(@minus, phi, [0, cumsum(a_odd(1:112)]);
phi = bsxfun(@rdivide, phi, a_odd(1:113));
phi = reshape(phi, 1, 452);