LCLSQ1; linear least squares with constraints. [IBM303x,360,370; FORTRAN]
LCLSQ1 solves linear least-squares fitting problems with or without linear equality or inequality constraints. Specifically, given a matrix C with p rows and n columns (p .GE. n) and a p-dimensional vector d, LCLSQ1 finds the n-dimensional vector x which minimizes the sum of squares in the residual vector Cx-d. Further, this vector x can be required to satisfy m linear equality and/or inequality constraints, e.g., a(i)x .GE. b(i), where a(i) is an n-dimensional row vector of coefficients and b(i) is a scalar constant. The role of constraints is particularly useful in controlling the nature of x for linear data-fitting applications.IBM303x,360,370; FORTRAN; OS/MVT; LCLSQ1 compiles and executes the sample problem in less than 180K bytes. The GO step requires 168K..
- Research Organization:
- Argonne National Lab., IL (USA)
- OSTI ID:
- 6358863
- Report Number(s):
- ANL/NESC-989; ON: DE83048989
- Country of Publication:
- United States
- Language:
- English
Similar Records
Solving linear inequalities in a least squares sense
Selected algorithms for the linearly constrained least squares problem: a user's guide. [Usage examples for LSEI and WNNLS, in FORTRAN]