The DIAG_MATRIX function constructs a diagonal matrix from an input vector, or if given a matrix, then DIAG_MATRIX will extract a diagonal vector.

## Examples

Create a tridiagonal matrix and extract the diagonal using the following program:

`PRO ExDiagMatrix; Convert three input vectors to a tridiagonal matrix:diag = [1, -2, 3, -4]sub = [5, 10, 15]super = [3, 6, 9]array = DIAG_MATRIX(diag) + \$DIAG_MATRIX(super, 1) + DIAG_MATRIX(sub, -1)PRINT, 'DIAG_MATRIX array:'PRINT, array; Extract the diagonal:PRINT, 'DIAG_MATRIX diagonal:'PRINT, DIAG_MATRIX(array)END`

When this program is compiled and run, IDL prints:

`DIAG_MATRIX array:`
`1       3       0       0`
`5      -2       6       0`
`0      10       3       9`
`0       0      15      -4`
`DIAG_MATRIX diagonal:`
`1      -2       3      -4`

## Syntax

Result = DIAG_MATRIX(A [, Diag] )

## Return Value

• If given an input vector with n values, the result is an n-by-n array of the same type. The DIAG_MATRIX function may also be used to construct subdiagonal or superdiagonal arrays.
• If given an input n-by-m array, the result is a vector with MIN(n,m) elements containing the diagonal elements. The DIAG_MATRIX function may also be used to extract subdiagonals or superdiagonals.

## Arguments

### A

Either an n-element input vector to convert to a diagonal matrix, or a n-by-m input array to extract a diagonal. A may be any numeric type.

### Diag

An optional argument that specifies the subdiagonal (Diag < 0) or superdiagonal (Diag > 0) to fill or extract. The default is Diag=0 which puts or extracts the values along the diagonal. If A is a vector with the m elements, then the result is an n-by-n array, where n = m + ABS(Diag). If A is an array, then the result is a vector whose length depends upon the number of elements remaining along the subdiagonal or superdiagonal.

Tip: The Diag argument may be used to easily construct tridiagonal arrays. For example, the expression,

DIAG_MATRIX(VL,-1) + DIAG_MATRIX(V) + DIAG_MATRIX(VU,1)

will create an n-by-n array, where VL is an (n - 1)-element vector containing the subdiagonal values, V is an n-element vector containing the diagonal values, and VU is an (n - 1)-element vector containing the superdiagonal values.

None.

## Version History

 5.6 Introduced