how to reshape an n x 1 matrix into a squre matrix using matlab?

hi experts........... greetings to all. hear a challenge question? how to reshape an n x 1 matrix into a square matrix using matlab?please let me know.i would be grateful to you.i am waiting for solution.

thanks in advance

Answer 1

[Jos (10584)]

This function will take any vector (or even a matrix) V and reshapesit into the smallest square matrix that can hold it. Note that it does not use reshape.

function M = reshapeVintosquareM (V)
% Reshapes a vector into the smallers square matrix possible. If V is too short
% the remainder of M is filled with NaNs.
N = numel(V) ;
M = NaN(ceil(sqrt(N))) ;
M(1:N) = V ;

[Jos (10584)]

Update: this approach is basically the same as WR's solution, of course.

[Walter Roberson]

Yep, there are special-cases of my code for some specific fill values such as nan:

SQ = nan(ceil(sqrt(length(vec)))); SQ(1:length(vec)) = vec;

[ramesh jallu]

thanks my friend

Answer 2

[Andreas Goser]

The answer is already in your question. RESHAPE.

x=[1 2 3 4 5 6 7 8 9];
reshape(x,3,3)

[ramesh jallu]

thanks for your reply.

i have an n x 1 matrix

[Andreas Goser]

Well 1xn / nx1, what does it matter? The code also works with reshape(x',3,3)

[Jan]

A patient answer for an easy problem. +1

[ramesh jallu]

what if n is a prime?

e.g. n=7,11,13,19....

after reshaping product of dimensions should be same.

i mean, i have an input n x 1 matrix, so n*1 should be equal to m*m,where m x m is the dimension of output square matrix and m<n.

in your example you have taken 1 x 9.we can write 1*9=3*3,so you could reshape the matrix 1 x 9 as 3 x 3 square matrix.

[Walter Roberson]

Sorry, Matlab is not able to handle that situation because Matlab uses finite precision arithmetic, and the situation you outline requires infinite precision arithmetic in order to properly be able to distinguish between indices 4.1231056256176605498214098559740770251471992253736, sqrt(17), and 4.1231056256176605498214098559740770251471992253737

[ramesh jallu]

thanks for your awesome response

Answer 3

[Walter Roberson]

Vectors which are not the square of a positive integer can be accomodated, if the unused spaces can be filled with some specific value of the same data class as the array values.

SQ = FillValue * ones(ceil(sqrt(length(vec)))); SQ(1:length(vec)) = vec;

The above will also work for vectors which would form perfect squares.

[Jan]

What an excellent service. You get ultimative solutions even for very basic questions.

I hope that Ramesh is goind to accept an answer.

[ramesh jallu]

thanks dude

Answer 4

[David Young]

The SQRT function may be useful here.

[ramesh jallu]

hi david,

sqrt won't work.it works only when n is a perfect square.what,if n is not a perfect square?

[Matt Tearle]

Then there's no way to do what you ask in the original question -reshape n-by-1 into a square matrix.

[ramesh jallu]

can we do that by adding some dummy rows and columns?

[Walter Roberson]

Adding dummy rows or columns would be contrary to the requirement that "after reshaping product of dimensions should be same".

If n is composite, it would be possible to arrange n in to a rectangle and add dummy rows and columns such the the result was a square matrix. If n is prime, then you would have to add (n-1) dummy rows (or columns) of length n in order to form a square.

I would have to think a bit to come up with an algorithm for determining the fewest total dummy rows and columns necessary to create the square matrix. I suspect that might also be the same solution as "the lowest total number of dummy elements added" but I am too tired at the moment to prove it mentally.

In the above discussion, I am assuming that n must be arranged in to a rectangle. The answer I gave above with SQ is based upon arranging n in to a square that has just a single partial strip of elements missing and filling those missing elements with a dummy value.

Answer 5

[Paulo Silva]

function [SQmatrix,flag]=vec2SQmat(vec)
%[SQmatrix,flag]=vec2SQmat(matrix)
%
%This function receives a vector and tries
%to convert it to a square matrix SQmatrix
%flag indicates the sucess (1) or failure (0)
%
flag=0; 
sq=sqrt(numel(vec));
if (round(sq)==sq)
SQmatrix=reshape(vec,sq,sq);
flag=1;
end