libgenua
Basic Geometry, Numerical Algorithms and Interfaces
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Groups Pages
Public Member Functions | List of all members
CraigSolver< Scalar > Class Template Reference

Detailed Description

template<typename Scalar>
class CraigSolver< Scalar >

Craig's method for under-determined systems.

CRAIG finds a solution x to the linear equation Ax = b, where A is a real matrix with m rows and n columns, and b is a real m-vector. If A is square and nonsingular, CRAIG finds the unique solution x = A(inv)b. If the system Ax = b is under-determined (i.e. there are many solutions), CRAIG finds the solution of minimum Euclidean length, namely x = A inv(A A') b. Thus, CRAIG solves the problem

     min  x'x  subject to  Ax = b.

y returns a vector satisfying A'y = x. Hence AA'y = b.

A is an m by n matrix (ideally sparse), or a function handle such that y = A(x,1) returns y = A*x (where x will be an n-vector); y = A(x,2) returns y = A'*x (where x will be an m-vector).

CRAIG uses an iterative (conjugate-gradient-like) method. For further information, see

  1. C. C. Paige and M. A. Saunders (1982a). LSQR: An algorithm for sparse linear equations and sparse least squares, ACM TOMS 8(1), 43-71.

08 Apr 2003: First craig.m derived from Fortran 77 version of craig1.for. Michael Saunders, Systems Optimization Laboratory, Dept of MS&E, Stanford University. 09 Apr 2003: Experimenting on singular systems (for inverse iteration). Separately, on full-rank Ax = b, "Acond" seems to over-estimate cond(A) drastically. 02 Oct 2006: Output y such that x = A'y (already in the f77 version). 15 Aug 2014: A can now be a matrix or a function handle, as in lsqrSOL. 28 Aug 2014: Fixed glitches found by Dominique Orban.

See Also
SolIterativeSolver, LsmrSolver

#include <craig.h>

Inheritance diagram for CraigSolver< Scalar >:
[legend]
Collaboration diagram for CraigSolver< Scalar >:
[legend]

Public Member Functions

int solve (const Operator &A, const DVector< Scalar > &b, DVector< Scalar > &x)
 minimize |x|^2 such that |Ax = b|
 
- Public Member Functions inherited from SolIterativeSolver
void tolerance (int maxiter, double atoler, double btoler, double conlim=1e8)
 set convergence tolerances and maximum number of iterations
 

Additional Inherited Members

- Static Public Member Functions inherited from SolIterativeSolver
static const char * statusMessage (int code)
 access text corresponding to error code
 
static bool success (int code)
 does exit code indicate succes (in some sense, at least)?
 
- Protected Attributes inherited from SolIterativeSolver
double m_atol = 1e-6
 convergence criteria
 
double m_btol = 1e-6
 convergence criteria
 
double m_conlim = 1e8
 condition number limit (actually, limit for the estimated condition)
 
size_t m_maxiter = 128
 permitted number of iteration
 

The documentation for this class was generated from the following file: