There is an error in my code

Here is the function:

function [L,U] = eluinv(A)

[~,n]=size(A);

[L,U] = lu(A)

LU = roundingfixer(L*U);

if (A == LU)

disp('LU factorization is valid')

else

disp('A mistake was made somewhere.')

end

if (rank(U) == rank(A))

disp('U is an echelon form of A')

else

disp('A mistake was made somewhere.')

end

if (rank(A) == n)

% Use rref to row reduce the matrices

invL = [L eye(n)];

invU = [U eye(n)];

invL = rref(invL);

invU = rref(invU);

% Cut off the first half of the matrix

invL = invL(:,(n+1:n*2));

invU = invU(:,(n+1:n*2));

invA = roundingfixer(invU * invL)

P = roundingfixer(inv(A))

if (invA == P)

disp('LU works for calculating the inverse for A')

else

disp('Error in calculating the inverse for A')

end

else

sprintf('A is not invertible.')

invA = [];

end

end

roundingfixer.m:

function B = roundingfixer(A)

[m,n]=size(A);

A=closetozeroroundoff(A,7)

for i=1:m

for j=1:n

A(i,j) = round(A(i,j),8);

end

end

B=A;

end

Answer 1

[Cris LaPierre]

Look at line 17: invL = [L eye(n)];

You are concatenating L and eye(n) horizontally. The error message is again telling you what the problem is. See this documentation page for more.

A = [1; 2]

A = 2×1

1 2

B = [3 4 5]

B = 1×3

3 4 5

C = [A B]

Error using horzcat
Dimensions of arrays being concatenated are not consistent.

[Tianlan Yang]

If I want it run, how can I edit that?

[Cris LaPierre]

Figure out why your sizes are not what you expected, and update your code accordingly.

[Tianlan Yang]

Hi, I still cannot figure that out. Why 3*4 matrix can run but 4*3 matrix has this error.

[Cris LaPierre]

Horizontal concatenation means the number of rows (first dimension) must be the same. This is why a 3x4 and a 3x3 can be horizontally concatenated (both have 3 rows), but a 4x3 and a 3x3 cannot (one has 4, the other has 3, so a dimension mismatch error occurs).

[Tianlan Yang]

Ok, I got it! So that means I need to change the horizontal concatenation to something else to make it run. Could you please provide another method to operate it beyond the row limitation? Thank you!

[Cris LaPierre]

I'm afraid I don't undertand what it is you are trying to do, so am not in a position to recommend a solution.

[Tianlan Yang]

Hi, I mean that is there a function that can replace horizontal concatenation (can both applied to row not equivalent situation)? Thank you so much！

[Cris LaPierre]

I am not aware of a function that will do this for you. I believe you will need to write your code in a way to either ensure the two matrices have the same number of rows.