Note: This page has been translated by MathWorks. Please click here

To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

To view all translated materals including this page, select Japan from the country navigator on the bottom of this page.

MATLAB^{®} never creates sparse matrices automatically.
Instead, you must determine if a matrix contains a large enough percentage
of zeros to benefit from sparse techniques.

The *density* of a matrix is the number of
nonzero elements divided by the total number of matrix elements. For
matrix `M`

, this would be

nnz(M) / prod(size(M));

nnz(M) / numel(M);

Matrices with very low density are often good candidates for use of the sparse format.

You can convert a full matrix to sparse storage using the `sparse`

function
with a single argument.

S = sparse(A)

For example:

A = [ 0 0 0 5 0 2 0 0 1 3 0 0 0 0 4 0]; S = sparse(A)

produces

S = (3,1) 1 (2,2) 2 (3,2) 3 (4,3) 4 (1,4) 5

The printed output lists the nonzero elements of `S`

,
together with their row and column indices. The elements are sorted
by columns, reflecting the internal data structure.

You can convert a sparse matrix to full storage using the `full`

function,
provided the matrix order is not too large. For example ```
A
= full(S)
```

reverses the example conversion.

Converting a full matrix to sparse storage is not the most frequent way of generating sparse matrices. If the order of a matrix is small enough that full storage is possible, then conversion to sparse storage rarely offers significant savings.

You can create a sparse matrix from a list of nonzero elements
using the `sparse`

function with five arguments.

S = sparse(i,j,s,m,n)

`i`

and `j`

are vectors of
row and column indices, respectively, for the nonzero elements of
the matrix. `s`

is a vector of nonzero values whose
indices are specified by the corresponding `(i,j)`

pairs. `m`

is
the row dimension for the resulting matrix, and `n`

is
the column dimension.

The matrix `S`

of the previous example can
be generated directly with

S = sparse([3 2 3 4 1],[1 2 2 3 4],[1 2 3 4 5],4,4) S = (3,1) 1 (2,2) 2 (3,2) 3 (4,3) 4 (1,4) 5

The `sparse`

command has a number of alternate
forms. The example above uses a form that sets the maximum number
of nonzero elements in the matrix to `length(s)`

.
If desired, you can append a sixth argument that specifies a larger
maximum, allowing you to add nonzero elements later without reallocating
the sparse matrix.

The matrix representation of the second difference operator is a good example of a sparse matrix. It is a tridiagonal matrix with -2s on the diagonal and 1s on the super- and subdiagonal. There are many ways to generate it—here's one possibility.

D = sparse(1:n,1:n,-2*ones(1,n),n,n); E = sparse(2:n,1:n-1,ones(1,n-1),n,n); S = E+D+E'

For `n = 5`

, MATLAB responds with

S = (1,1) -2 (2,1) 1 (1,2) 1 (2,2) -2 (3,2) 1 (2,3) 1 (3,3) -2 (4,3) 1 (3,4) 1 (4,4) -2 (5,4) 1 (4,5) 1 (5,5) -2

Now `F = full(S)`

displays the corresponding
full matrix.

F = full(S) F = -2 1 0 0 0 1 -2 1 0 0 0 1 -2 1 0 0 0 1 -2 1 0 0 0 1 -2

Creating sparse matrices based on their diagonal elements is
a common operation, so the function `spdiags`

handles
this task. Its syntax is

S = spdiags(B,d,m,n)

To create an output matrix `S`

of size *m*-by-*n* with
elements on `p`

diagonals:

`B`

is a matrix of size`min(m,n)`

-by-*p*. The columns of`B`

are the values to populate the diagonals of`S`

.`d`

is a vector of length`p`

whose integer elements specify which diagonals of`S`

to populate.

That is, the elements in column `j`

of `B`

fill
the diagonal specified by element `j`

of `d`

.

If a column of `B`

is longer than the diagonal
it's replacing, super-diagonals are taken from the lower part of the
column of `B`

, and sub-diagonals are taken from the
upper part of the column of `B`

.

As an example, consider the matrix `B`

and
the vector `d`

.

B = [ 41 11 0 52 22 0 63 33 13 74 44 24 ]; d = [-3 0 2];

Use these matrices to create a 7-by-4 sparse matrix `A`

:

A = spdiags(B,d,7,4) A = (1,1) 11 (4,1) 41 (2,2) 22 (5,2) 52 (1,3) 13 (3,3) 33 (6,3) 63 (2,4) 24 (4,4) 44 (7,4) 74

In its full form, `A`

looks like this:

full(A) ans = 11 0 13 0 0 22 0 24 0 0 33 0 41 0 0 44 0 52 0 0 0 0 63 0 0 0 0 74

`spdiags`

can also extract diagonal elements
from a sparse matrix, or replace matrix diagonal elements with new
values. Type `help`

`spdiags`

for
details.

You can import sparse matrices from computations outside the MATLAB environment.
Use the `spconvert`

function
in conjunction with the `load`

command to import
text files containing lists of indices and nonzero elements. For example,
consider a three-column text file `T.dat`

whose first
column is a list of row indices, second column is a list of column
indices, and third column is a list of nonzero values. These statements
load `T.dat`

into MATLAB and convert it into
a sparse matrix `S`

:

load T.dat S = spconvert(T)

The `save`

and `load`

commands can also process sparse
matrices stored as binary data in MAT-files.