reshape

Reshape array

Basic Idea

When an array (say, a) is reshaped, it is first converted to a vector by indexing all of its elements linearly, i.e., a(1:numel(a)). Then, this vector is rearranged to form an array of the desired size. The number of elements after reshaping remains the same.

For example, the $3\times 5\times 3$ array below is reshaped to an $5\times 9$ array.

Usage

reshape(A, sizeVec)

• A should be an array.
• sizeVec should be a (size) vector containing real non-negative integers not equal to Inf nor NaN. It specifies the desired size of the output.
• sizeVec should be given such that the number of elements remains unchanged after shaping.

reshape(A, s1, s2, ..., sn)

• Each of s1, s2, ..., sn is either empty or is a real non-negative integer scalar not equal to Inf nor NaN.
• They define the desired size vector of the output.
• If one of s1, s2, ..., sn is an empty array, the length of the corresponding dimension will be calculated automatically to fit the number of elements of A. However, there should not be more than one [].

Examples

• The $3\times 4\times 2$ array r1 below is reshaped to the $8\times 3$ array r2. Although the number of dimensions has changed (from 3 to 2) after reshaping, the number of elements (24) remains unchanged.

Input

% 24 elemenets before reshaping
r1 = rand(3,4,2)
% 24 element after reshaping
r2 = reshape(r1,8,3)

Output

r1(:, :, 1) = 1e-1 ×
 9.7679   6.0583   6.3534   9.5830
 2.7812   2.4721   2.5032   2.5139
 3.9738   6.1257   1.0086   0.9723

r1(:, :, 2) = 1e-1 ×
 3.2380   4.9974   6.0711   7.8545
 4.8453   5.1516   6.4629   1.4396
 1.5734   9.0568   9.7506   2.7975

r2 = 1e-1 × 
 9.7679   1.0086   5.1516
 2.7812   9.5830   9.0568
 3.9738   2.5139   6.0711
 6.0583   0.9723   6.4629
 2.4721   3.2380   9.7506
 6.1257   4.8453   7.8545
 6.3534   1.5734   1.4396
 2.5032   4.9974   2.7975

• Calculating size automatically. Reshape the $3\times 4$ array r1 to a $4\times 3$ array. For r2, the length of the first dimension is calculated automatically, such that r1 and r2 have the same numbers of elements.

Input

% 12 elements in r1
r1 = rand(3,4)

% 1st dim's length of r2 calculated automatically
r2 = reshape(r1,[],3)

Output

r1 = 1e-1 × 
 6.0433   4.9525   7.5725   5.5289
 9.4648   5.7737   9.4426   8.4097
 5.8870   3.3192   3.1616   6.8046

r2 = 1e-1 × 
 6.0433   5.7737   3.1616
 9.4648   3.3192   5.5289
 5.8870   7.5725   8.4097
 4.9525   9.4426   6.8046