Combined value of different arrays with constraints

Hi! I have a problem regarding computing combined values from arrays that not have the same size. For a given timeseries, t, I want my for loop to compute a value if some constraints are fulfilled.

I have a vector p_dh which contains t samples where only t-n are valid and useful. Valid elements of p_dh is given by vector A1.

Another vector A contains values where other constraints are accounted for. To compute the value, given that current element of p_dh is valid(t~= any element of A1), my plan was to use a for loop comprising som if and elseif statements. This pseudo code may be useful for understanding the problem:

for t=1:length(p_dh),
  if t == any element of A1,
      value(t) = nan;
  elseif t == any element of A,
      value(t) = p_dh(t) - p_c(t) - constant_a;
  else
      value(t) = nan;
  end
end

In the end I want a vector value of size [1:4020] which has both nan elements and valid numerical values according to the for loop and given constraints. It's worth noticing that lengths of t, A and A1 is individual. t = [1:4020], A = [1:502] and A1 = [1:453]. Can you guys help me out here?

thanks!

[Stephen23]

You write that "p_dh is valid(t~= any element of A1)", but the code seems to indicate that p_dh is valid for any(t~=A1)&any(t==A). Can you confirm this?

[Andreas Volden]

The code is correct. My text formulation may indicate otherwise, I see that. If the first statement is true(if t == any element of A1), set value to nan and increment t. If this statement is false, move to elseif statement and check t == any element of A. If this is true compute value and then increment t. And so on...

[Stephen23]

Do any of the suggested solutions work for you?

Answer 1

[Thorsten]

 result = nan(size(p_th));

 for t=1:length(p_th)
  if ismember(p_th(t), A)
    result(t) = p_th(t) - p_c(t) - constant_a;
  end
end

[Andreas Volden]

Works fine. Thank you!

Answer 2

[Stephen23]

This is a perfect example of where using vectorization can really help to make MATLAB code neater and faster, by applying some operation to the whole array at once instead of in loops:

First we create some fake data:

p_dh = 0:9;
A = 0:2:8;
A1 = 0:3:9;

then we define the logic of which elements you need:

vec = 1:numel(p_dh);
idx = ismember(vec,A) & ~ismember(vec,A1);

and then simply transfer the data from the original arrays:

out(idx) = p_dh(idx); % + p_c(idx) - constant_a
out(~idx) = nan;

[Andreas Volden]

This also works fine. Thank you!