get_lapack_funcs(names, arrays=(), dtype=None, ilp64=False)
Arrays are used to determine the optimal prefix of LAPACK routines.
This routine automatically chooses between Fortran/C interfaces. Fortran code is used whenever possible for arrays with column major order. In all other cases, C code is preferred.
In LAPACK, the naming convention is that all functions start with a type prefix, which depends on the type of the principal matrix. These can be one of {'s', 'd', 'c', 'z'} for the NumPy types {float32, float64, complex64, complex128} respectively, and are stored in attribute typecode
of the returned functions.
Name(s) of LAPACK functions without type prefix.
Arrays can be given to determine optimal prefix of LAPACK routines. If not given, double-precision routines will be used, otherwise the most generic type in arrays will be used.
Data-type specifier. Not used if :None:None:`arrays` is non-empty.
Whether to return ILP64 routine variant. Choosing 'preferred' returns ILP64 routine if available, and otherwise the 32-bit routine. Default: False
List containing the found function(s).
Return available LAPACK function objects from names.
Suppose we would like to use '?lange' routine which computes the selected norm of an array. We pass our array in order to get the correct 'lange' flavor.
>>> import scipy.linalg as LA
... rng = np.random.default_rng()
... a = rng.random((3,2))
... x_lange = LA.get_lapack_funcs('lange', (a,))
... x_lange.typecode 'd'
>>> x_lange = LA.get_lapack_funcs('lange',(a*1j,))
... x_lange.typecode 'z'
Several LAPACK routines work best when its internal WORK array has the optimal size (big enough for fast computation and small enough to avoid waste of memory). This size is determined also by a dedicated query to the function which is often wrapped as a standalone function and commonly denoted as ###_lwork
. Below is an example for ?sysv
>>> import scipy.linalg as LASee :
... rng = np.random.default_rng()
... a = rng.random((1000, 1000))
... b = rng.random((1000, 1)) * 1j
... # We pick up zsysv and zsysv_lwork due to b array
... xsysv, xlwork = LA.get_lapack_funcs(('sysv', 'sysv_lwork'), (a, b))
... opt_lwork, _ = xlwork(a.shape[0]) # returns a complex for 'z' prefix
... udut, ipiv, x, info = xsysv(a, b, lwork=int(opt_lwork.real))
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