ldl(A, lower=True, hermitian=True, overwrite_a=False, check_finite=True)
This function returns a block diagonal matrix D consisting blocks of size at most 2x2 and also a possibly permuted unit lower triangular matrix L
such that the factorization A = L D L^H
or A = L D L^T
holds. If :None:None:`lower` is False then (again possibly permuted) upper triangular matrices are returned as outer factors.
The permutation array can be used to triangularize the outer factors simply by a row shuffle, i.e., lu[perm, :]
is an upper/lower triangular matrix. This is also equivalent to multiplication with a permutation matrix P.dot(lu)
, where P
is a column-permuted identity matrix I[:, perm]
.
Depending on the value of the boolean :None:None:`lower`, only upper or lower triangular part of the input array is referenced. Hence, a triangular matrix on entry would give the same result as if the full matrix is supplied.
This function uses ?SYTRF
routines for symmetric matrices and ?HETRF
routines for Hermitian matrices from LAPACK. See for the algorithm details.
Depending on the :None:None:`lower` keyword value, only lower or upper triangular part of the input array is referenced. Moreover, this keyword also defines the structure of the outer factors of the factorization.
Square input array
This switches between the lower and upper triangular outer factors of the factorization. Lower triangular ( lower=True
) is the default.
For complex-valued arrays, this defines whether A = A.conj().T
or A = A.T
is assumed. For real-valued arrays, this switch has no effect.
Allow overwriting data in A (may enhance performance). The default is False.
Whether to check that the input matrices contain only finite numbers. Disabling may give a performance gain, but may result in problems (crashes, non-termination) if the inputs do contain infinities or NaNs.
If input array is not square.
If a complex-valued array with nonzero imaginary parts on the diagonal is given and hermitian is set to True.
The (possibly) permuted upper/lower triangular outer factor of the factorization.
The block diagonal multiplier of the factorization.
The row-permutation index array that brings lu into triangular form.
Computes the LDLt or Bunch-Kaufman factorization of a symmetric/ hermitian matrix.
Given an upper triangular array a
that represents the full symmetric array with its entries, obtain l
, 'd' and the permutation vector :None:None:`perm`:
>>> import numpy as np
... from scipy.linalg import ldl
... a = np.array([[2, -1, 3], [0, 2, 0], [0, 0, 1]])
... lu, d, perm = ldl(a, lower=0) # Use the upper part
... lu array([[ 0. , 0. , 1. ], [ 0. , 1. , -0.5], [ 1. , 1. , 1.5]])
>>> d array([[-5. , 0. , 0. ], [ 0. , 1.5, 0. ], [ 0. , 0. , 2. ]])
>>> perm array([2, 1, 0])
>>> lu[perm, :] array([[ 1. , 1. , 1.5], [ 0. , 1. , -0.5], [ 0. , 0. , 1. ]])
>>> lu.dot(d).dot(lu.T) array([[ 2., -1., 3.], [-1., 2., 0.], [ 3., 0., 1.]])See :
The following pages refer to to this document either explicitly or contain code examples using this.
scipy.linalg._decomp_ldl.ldl
Hover to see nodes names; edges to Self not shown, Caped at 50 nodes.
Using a canvas is more power efficient and can get hundred of nodes ; but does not allow hyperlinks; , arrows or text (beyond on hover)
SVG is more flexible but power hungry; and does not scale well to 50 + nodes.
All aboves nodes referred to, (or are referred from) current nodes; Edges from Self to other have been omitted (or all nodes would be connected to the central node "self" which is not useful). Nodes are colored by the library they belong to, and scaled with the number of references pointing them