Info

# for loop indexing in temporary arrary problem in MATLAB.

1 view (last 30 days)
hello_world on 18 May 2016
Closed: Geoff Hayes on 18 May 2016
Hello Friends,
I have the following code:
P = [1, 0, 3];
Q = [0, 4, 5];
AB = [P + Q]; % A vector
CD = [P - Q]; % A vector
EF = [P ./ Q]; % A vector
M = {'AB', 'CD', 'EF'};
for i = 1:length(M)
P1 = P(M{i}~=0); %It removes those elements of P which correspond to 0 entries in M.
t = M{i}; %Create a temporary variable.
M = t(M{i}~=0); %It removes 0 entries from M.
P = P1(~isnan(M(i))); %It removes those elements of P which correspond to NaN entries in M.
M = t(~isnan(M(i))); %It removes NaN entries from M.
if strcmp(M, 'AB')
f = f(P,M);
elseif strcmp(M, 'CD')
f = g(Q,M);
elseif strcmp(M, 'EF')
f = h(R,M);
end
end
This code is not computing P1, t, M, P values properly. For example the following line of code takes M{i} = AB for i = 1, but gives totally wrong answer.
I have tried to change {} to () and [], etc., but nothing works. I will appreciate any advice.
P1 = P(M{i}~=0);

Geoff Hayes on 18 May 2016
Your line of code has me confused
M = t(M{i}~=0); %It removes 0 entries from M.
The comment says that you are removing zero entries from M. But M has been defined to be a cell array of strings as
M = {'AB', 'CD', 'EF'};
What is the intent of the above assignment? Do you really mean for M to be a cell array of strings, or do you mean it to be a cell array of numbers?
M = {AB, CD, EF};
Also, the line of code
CD = [P*Q]
will generate an error since P and Q do not have the compatible dimensions for matrix multiplication.
Use the debugger to step through the code and you will probably get a good idea as to what is going on. Always look at the variables to verify that they are (with respect to dimension, type, etc.) what you expect them to be.
Geoff Hayes on 18 May 2016

Todd Leonhardt on 18 May 2016
I don't understand how that code could even get far enough to attempt to compute P1, etc. It should error out on the 4th line of code where you attempt to compute:
CD = P * Q;
P and Q are both 1 x 3 matrices, so you can not perform matrix multiplication on them as such. You can do any one of the following:
CD = P' * Q; % Take transpose of P, P', so it is a 3 x 1 matrix and results is a 3x3
CD = P * Q'; % Take transpose of Q, Q', so it is a 3 x 1 matrix and result is a 1x1 scalar
CD = P .* Q; % Perform element-wise multiplication so result is a 1 x 3