NLPQLPOptimizer Class Reference

Wrapper class for the NLPQLP optimization library, Version 2.0. More...

Inheritance diagram for NLPQLPOptimizer:

List of all members.

Public Member Functions
	NLPQLPOptimizer (Model &model)
	standard constructor
	NLPQLPOptimizer (NoDBBaseConstructor, Model &model)
	alternate constructor
	~NLPQLPOptimizer ()
	destructor
void	find_optimum ()
	Redefines the run virtual function for the optimizer branch.
Protected Member Functions
void	derived_pre_run ()
	performs run-time set up
Private Member Functions
void	initialize ()
	Shared constructor code.
void	allocate_workspace ()
	Allocates workspace for the optimizer.
void	deallocate_workspace ()
	Releases workspace memory.
void	allocate_constraints ()
	Allocates constraint mappings.
Private Attributes
int	L
	the serial version by setting L=1.
int	numEqConstraints
	numEqConstraints : Number of equality constraints.
int	MMAX
	MMAX must be at least one and greater or equal to M.
int	N
	N : Number of optimization variables.
int	NMAX
	than N.
int	MNN2
	MNN2 : Must be equal to M+N+N+2.
double *	X
	function values should be computed simultaneously.
double *	F
	values to be computed from L iterates stored in X.
double *	G
	function values to be computed from L iterates stored in X.
double *	DF
	of F to compute DF.
double *	DG
	has to be equal to MMAX.
double *	U
	inequality constraints should be nonnegative.
double *	C
	to NMAX.
double *	D
	array D.
double	ACC
	than the accuracy by which gradients are computed.
double	ACCQP
	by NLPQLP and subsequently multiplied by 1.0D+4.
double	STPMIN
	by STPMIN**(1/L-1). If STPMIN<=0, then STPMIN=ACC is used.
int	MAXFUN
	than 50.
int	MAXIT
	gradients (e.g. 100).
int	MAX_NM
	MAX_NM=0, monotone line search is performed.
double	TOL_NM
	non-negative (e.g. 0.1).
int	IPRINT
	values are displayed during the line search.
int	MODE
	function in C and D in form of an LDL decomposition.
int	IOUT
	write-statements start with 'WRITE(IOUT,... '.
int	IFAIL
	constraint.
double *	WA
	WA(LWA) : WA is a real working array of length LWA.
int	LWA
	LWA : LWA value extracted from NLPQLP20.f.
int *	KWA
	KWA(LKWA) : The user has to provide working space for an integer array.
int	LKWA
	LKWA : LKWA should be at least N+10.
int *	ACTIVE
	ACTIVE(J)=.TRUE., J=1,...,M.
int	LACTIVE
	least 2*M+10.
int	LQL
	contains only an upper triangular factor.
int	numNlpqlConstr
	total number of constraints seen by NLPQL
SizetList	nonlinIneqConMappingIndices
	constraints used in computing the corresponding NLPQL constraints.
RealList	nonlinIneqConMappingMultipliers
	constraints to the corresponding NLPQL constraints.
RealList	nonlinIneqConMappingOffsets
	constraints to the corresponding NLPQL constraints.
SizetList	linIneqConMappingIndices
	constraints used in computing the corresponding NLPQL constraints.
RealList	linIneqConMappingMultipliers
	constraints to the corresponding NLPQL constraints.
RealList	linIneqConMappingOffsets
	constraints to the corresponding NLPQL constraints.

---

Detailed Description

Wrapper class for the NLPQLP optimization library, Version 2.0.

********************************************************************

AN IMPLEMENTATION OF A SEQUENTIAL QUADRATIC PROGRAMMING METHOD FOR SOLVING NONLINEAR OPTIMIZATION PROBLEMS BY DISTRIBUTED COMPUTING AND NON-MONOTONE LINE SEARCH

This subroutine solves the general nonlinear programming problem

minimize F(X) subject to G(J,X) = 0 , J=1,...,ME G(J,X) >= 0 , J=ME+1,...,M XL <= X <= XU

and is an extension of the code NLPQLD. NLPQLP is specifically tuned to run under distributed systems. A new input parameter L is introduced for the number of parallel computers, that is the number of function calls to be executed simultaneously. In case of L=1, NLPQLP is identical to NLPQLD. Otherwise the line search is modified to allow L parallel function calls in advance. Moreover the user has the opportunity to used distributed function calls for evaluating gradients.

The algorithm is a modification of the method of Wilson, Han, and Powell. In each iteration step, a linearly constrained quadratic programming problem is formulated by approximating the Lagrangian function quadratically and by linearizing the constraints. Subsequently, a one-dimensional line search is performed with respect to an augmented Lagrangian merit function to obtain a new iterate. Also the modified line search algorithm guarantees convergence under the same assumptions as before.

For the new version, a non-monotone line search is implemented which allows to increase the merit function in case of instabilities, for example caused by round-off errors, errors in gradient approximations, etc.

The subroutine contains the option to predetermine initial guesses for the multipliers or the Hessian of the Lagrangian function and is called by reverse communication.

---

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

• NLPQLPOptimizer.H
• NLPQLPOptimizer.C

---