
- •If a is a 2-d matrix.
- •In which case they are expanded so that the first three arguments
- •Working with sparse matrices
- •If s is symmetric, then colperm generates a permutation so that
- •Linear algebra
- •If sigma is a real or complex scalar including 0, eigs finds the
- •Is compensated so that column sums are preserved. That is, the
- •Xreginterprbf/condest
- •X and y are vectors of coordinates in the unit square at which
- •If you have a fill-reducing permutation p, you can combine it with an
- •Miscellaneous
- •In previous versions of matlab, the augmented matrix was used by
Miscellaneous
<symbfact> - Symbolic factorization analysis.
SYMBFACT Symbolic factorization analysis.
Analyzes the Cholesky factorization of A, A'*A, or A*A'.
count = SYMBFACT(A) returns row counts of R = CHOL(A)
count = SYMBFACT(A,'sym') same as SYMBFACT(A)
count = SYMBFACT(A,'col') returns row counts of R = CHOL(A'*A)
count = SYMBFACT(A,'row') returns row counts of R = CHOL(A*A')
count = SYMBFACT(A,'lo') same as SYMBFACT(A'), uses TRIL(A)
The flop count for a subsequent Cholesky factorization is sum(count.^2)
[count,h,parent,post,R] = SYMBFACT(...) returns:
h: height of the elimination tree
parent: the elimination tree itself
post: postordering of the elimination tree
R: a 0-1 matrix whose structure is that of CHOL(A) for the
symmetric case, CHOL(A'*A) for the 'col' case, or
CHOL(A*A') for the 'row' case.
SYMBFACT(A) and SYMBFACT(A,'sym') uses the upper triangular part of A
(TRIU(A)) and assumes the lower triangular part is the transpose of
the upper triangular part. SYMBFACT(A,'lo') uses TRIL(A) instead.
[count, h, parent, post, L] = SYMBFACT(A,TYPE,'lower'), where TYPE is
one of 'sym', 'col', 'row', or 'lo' returns a lower triangular symbolic
factor L = R'. This form is quicker and requires less memory.
See also chol, etree, treelayout.
Reference page in Help browser
doc symbfact
<spparms> - Set parameters for sparse matrix routines.
SPPARMS Set parameters for sparse matrix routines.
SPPARMS('key',value) sets one or more of the "tunable" parameters
used in the sparse routines, particularly sparse / and \.
SPPARMS, by itself, prints a description of the current settings.
If no input argument is present, values = SPPARMS returns a
vector whose components give the current settings.
[keys,values] = SPPARMS returns that vector, and also returns
a character matrix whose rows are the keywords for the parameters.
SPPARMS(values), with no output argument, sets all the parameters
to the values specified by the argument vector.
value = SPPARMS('key') returns the current setting of one parameter.
SPPARMS('default') sets all the parameters to their default settings.
SPPARMS('tight') sets the minimum degree ordering parameters to their
"tight" settings, which may lead to orderings with less fill-in, but
which makes the ordering functions themselves use more execution time.
The parameters with the default and "tight" values are:
keyword default tight
values(1) 'spumoni' 0
values(2) 'thr_rel' 1.1 1.0
values(3) 'thr_abs' 1.0 0.0
values(4) 'exact_d' 0 1
values(5) 'supernd' 3 1
values(6) 'rreduce' 3 1
values(7) 'wh_frac' 0.5 0.5
values(8) 'autommd' 1
values(9) 'autoamd' 1
values(10) 'piv_tol' 0.1
values(11) 'bandden' 0.5
values(12) 'umfpack' 1
values(13) 'sym_tol' 0.001
values(14) 'ldl_tol' 0.01
The meanings of the parameters are
spumoni: The Sparse Monitor Flag controls diagnostic output;
0 means none, 1 means some, 2 means too much.
thr_rel,
thr_abs: Minimum degree threshold is thr_rel*mindegree + thr_abs.
exact_d: Nonzero to use exact degrees in minimum degree,
Zero to use approximate degrees.
supernd: If > 0, MMD amalgamates supernodes every supernd stages.
rreduce: If > 0, MMD does row reduction every rreduce stages.
wh_frac: Rows with density > wh_frac are ignored in COLMMD.
autommd: Nonzero to use SYMMMD and COLMMD orderings with \ and /.
autoamd: Nonzero to use AMD or COLAMD ordering with CHOLMOD, UMFPACK, and SuiteSparseQR in \ and /.
piv_tol: Pivot tolerance used by LU-based (UMFPACK) \ and /.
bandden: Backslash uses band solver if band density is > bandden.
If bandden = 1.0, never use band solver.
If bandden = 0.0, always use band solver.
umfpack: Nonzero to use UMFPACK instead of the v4 LU-based solver
in \ and /.
sym_tol: Symmetric pivot tolerance used by UMFPACK. See LU for
more information about the role of the symmetric pivot
tolerance.
ldl_tol: Pivot tolerance used by LDL-based (MA57) \ and /.
Note:
Solving symmetric positive definite matrices within \ and /:
The CHOLMOD CHOL-based solver uses AMD.
Solving general square matrices within \ and /:
The UMFPACK LU-based solver uses either AMD or a modified COLAMD.
The v4 LU-based solver uses COLMMD.
Solving rectangular matrices within \ and /:
The SuiteSparseQR QR-based solver uses COLAMD.
All of these algorithms respond to SPPARMS('autoamd') except for the
v4 LU-based, which responds to SPPARMS('autommd').
See also amd, colamd, symamd.
Reference page in Help browser
doc spparms
<spaugment> - Form least squares augmented system.
SPAUGMENT Form least squares augmented system.
S = SPAUGMENT(A,c) creates the sparse, square, symmetric indefinite
matrix S = [c*I A; A' 0]. This matrix is related to the least
squares problem
min norm(b - A*x)
by
r = b - A*x
S * [r/c; x] = [b; 0].
The optimum value of the residual scaling factor c, involves
min(svd(A)) and norm(r), which are usually too expensive to compute.
S = SPAUGMENT(A), without a specified value of c, uses
max(max(abs(A)))/1000.