How to simplify the for loop for 3D matrix?

4 views (last 30 days)
I have the following code, where I am looking to eliminate only the outer for-loop (k) and include 'k' inside the inner loop. Here, I perform inverse considering each calculation point (k=1:NFocP) in a loop, but would like to some how eliminate outer loop and perform the inverse for all points at one instance. And the other concern is that I don't want to disturb the mldivide (\) function. Is there any way to solve this issue?
clear all;
load A.mat
load s1.mat
load s2.mat
Foc = size(s1,3); % no of calculation points
Nt = size(s1,2); % length of signal
M = size(s1,1); % no of rsensors
for k=1:Foc
for p = 1:Nt
H = [s1(:,p,k) s2(:,p,k)];
tmp = H\ A(:,p);
Ainv1(k,p) = tmp(1);
Ainv2(k,p) = tmp(2);
end
end
  2 Comments
Benjamin Thompson
Benjamin Thompson on 11 Jan 2023
That would require calculating an inverse of a 3D matrix for H which is not possible. You may be able to use foreach to speed up the code by doing the outer loops in parallel.
Kalasagarreddi Kottakota
Kalasagarreddi Kottakota on 11 Jan 2023
Attached H and A1
The matrix H looks like 9 x 2 x 14 and matrix A1 is 9 x1. I need to perform inversion between every page of H with A1 using mldivide so that output looks like 2 x 1 x14 matrix. No loops. Is this possible?
clear all;
close all;
load H.mat
load A1.mat

Sign in to comment.

Accepted Answer

Bruno Luong
Bruno Luong on 11 Jan 2023
Edited: Bruno Luong on 11 Jan 2023
load('A.mat');
load('s1.mat');
load('s2.mat');
[m,n,p] = size(s1);
% s = [reshape(s1, [m,1,n,p]), reshape(s2, [m,1,n,p])];
% Ar = reshape(A,m,1,[]);
% C=pagemldivide(s,Ar);
% Ainv1=reshape(C(1,:,:),n,p).';
% Ainv2=reshape(C(2,:,:),n,p).';
s = reshape([s1;s2], [m,2,n,p]);
Ainv1 = zeros(p,n);
Ainv2 = zeros(p,n);
for i=1:n
C=pagemldivide(s(:,:,i,:),A(:,i));
Ainv1(:,i)=C(1,:);
Ainv2(:,i)=C(2,:);
end
  4 Comments
Kalasagarreddi Kottakota
Kalasagarreddi Kottakota on 11 Jan 2023
Edited: Kalasagarreddi Kottakota on 11 Jan 2023
Ok, I see now. I have 2021. Is there any possibility to replace pagemldivide with any other alternate functions to do same task?
Bruno Luong
Bruno Luong on 11 Jan 2023
You might try this that requires the FEX
https://fr.mathworks.com/matlabcentral/fileexchange/27762-small-size-linear-solver#functions_tab
[m,n,p] = size(s1);
s = reshape([s1;s2], [m,2,n,p]);
Ainv1 = zeros(p,n);
Ainv2 = zeros(p,n);
Ar = reshape(A,m,1,[]);
scs = pagemtimes(s,'ctranspose',s,'none');
scA = pagemtimes(s,'ctranspose',Ar,'none');
for i=1:n
scsi = reshape(scs(:,:,i,:),2,2,[]);
scAi = reshape(scA(:,:,i,:),2,1,[]);
C = inv2(scsi, scAi); % https://fr.mathworks.com/matlabcentral/fileexchange/27762-small-size-linear-solver#functions_tab
Ainv1(:,i)=C(1,:);
Ainv2(:,i)=C(2,:);
end

Sign in to comment.

More Answers (0)

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!