Optimization of code with double for-loops

I am looking to optimize the following code specifically, and more generally, I am trying to learn optimization techniques in MATLAB for a larger project. The code below is just a snippet example which I am looking to learn techniques and then apply to a larger problem. I am potentially looking at values of L being as high as with four nested loops as opposed to just the two shown below.

Any pointers (or links to decent YouTube video on MATLAB optimization) would be welcomed.

clc; clearvars;

T = 1; Ts = 1e-2; t = 0:Ts:L*T;

L = 2^5; 

start_time = clock(); 

for l = 0:L-1
    for q = 0:L-1
        flag = (t>=(l-q-1)*T) & (t<=(l-q+1)*T);
        partA = (T-abs(t-(l-q)*T));
    end
end

elapsed_time = clock();
time = elapsed_time(4)*3600+elapsed_time(5)*60+elapsed_time(6)

[Walter Roberson]

    for q = 0:L-1
        flag = (t>=(l-q-1)*T) & (t<=(l-q+1)*T);

It looks to me as if you should see discretize and possibly splitapply

[Catalytic]

@Howard Wilton You appear to have opened several previous threads on similar subjects to this and abandoned them, without accepting an answer or follow-up discussion. It is better if you not do this, so that contributors can know if the question is to be taken seriously.

[Howard Wilton]

The questions didn't really hit the answer so i thought i would cast the questions differently.

However, your advice is duly noted and i shall be careful in the future. Thanks.

[Howard Wilton]

I have taken the code a bit further and managed to dispense with the nested loops . However, I still need a 1D loop. Performance is substantially improved.

clc; clearvars;

L = 2^8; T = 1; t = 0:0.2:L*T;

start_time = clock();

l = 1:L; q = 1:L;

X = (l(:)-q(:).')*T;

for loop = 1:length(t)

t2(:,:,loop) = t(loop) - X;

end

stop_time = elapsedTime(start_time)

function time = elapsedTime(start_time)

elapsed_time = clock-start_time;

time = elapsed_time(4)*3600+elapsed_time(5)*60+elapsed_time(6);

end

Keen to figure out a way to eliminate the 1D loop and generate a matrix .

Answer 1

[Walter Roberson]

X = (l(:)-q(:).')*T;
for loop = 1:length(t)
  t2(:,:,loop) = t(loop) - X;
end

replace with

t2 = reshape(t, 1, 1, []) - X;

[Howard Wilton]

thankyou !!