How to operate on 2d matrices in a 3D array?

조회 수: 13 (최근 30일)
Alon Rozen
Alon Rozen 2020년 2월 11일
댓글: Alon Rozen 2020년 2월 11일
Hi,
I have a set of n 3x3 matricies orgenized in a 3D array with dimesions 3x3xn.
I want to operate on all 3x3 matricies: I want to get a vector of determinants and a vector of trace of each 3x3 matrix.
I also want to calculate a 3x3xn array which will hold the 3x3 inverse of each matix.
I can do it using a for loop, of couse, calculating in each iteration for the ith matrix.
Is there a simpler and faster way to do it?
Thanks,
Alon
  댓글 수: 1
Stephen23
Stephen23 2020년 2월 11일
편집: Stephen23 2020년 2월 11일
"Is there a simpler and faster way to do it?"
Calculating the matrix inverse is rarely required or nor efficient when doing numeric computing. Read this to know why:
If you are really interested in efficiency, then use better tools than inv, just as the documentation recommends.
The determinant also tends to be overused by beginners who do not realize how useless it is in numeric computing. Search this forum to know why the determinant is essentially useless in numeric computing.

댓글을 달려면 로그인하십시오.

채택된 답변

Bhaskar R
Bhaskar R 2020년 2월 11일
data = randi([1,10], 3,3,4); % your matrix with n = 4
data_det = arrayfun(@(i)det(data(:,:,i)), 1:size(data,3));
data_trace = arrayfun(@(i)trace(data(:,:,i)), 1:size(data,3));
data_inv = arrayfun(@(i)inv(data(:,:,i)), 1:size(data,3), 'UniformOutput', false);
  댓글 수: 6
Stephen23
Stephen23 2020년 2월 11일
"Why is that?"
Essentially because repeatedly calling a function handle that runs sme code is slower than just calling that code directly.
"If I will use only one fuction that returns all calculation results, it also will be slower then a loop?"
Probably... try it and find out.
Alon Rozen
Alon Rozen 2020년 2월 11일
Thanks Stepen!

댓글을 달려면 로그인하십시오.

추가 답변 (0개)

카테고리

Help CenterFile Exchange에서 Matrix Indexing에 대해 자세히 알아보기

제품


릴리스

R2019a

Community Treasure Hunt

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

Start Hunting!

Translated by