==================================================
NUMERICAL ANALYSIS TITLES -- Volume 2, Number 2
	June, 1984
==================================================

[1] Ron S. Dembo and Ulrich Tulowitzki

	On the Minimization of Quadratic Functions
	Subject to Box Constraints

	SOM Working Paper Series B #71
	Yale University School of
	Organization and Management
	1983

We study efficient algorithms for very large quadratic problems
subject to box constraints an propose some new convergent methods
that permit the addition and deletion of many constraints each
time a search direction is calculated.  Numerical experiments,
on problems of up to 10,000 variables, indicate that the perfor-
mance of some of the proposed algorithms appears to be insensitive
to the number of binding constraints at the optimal solution or
to the starting point.  Also it appears as if solving the box
constrained problem is at worst marginally more expensive than
solving the same problem without constraints.

Submitted by: Ron Dembo
Obtainable from:
	Ron Dembo
	Yale University
	School of Organization and Management
	P.O. Box 1A
	New Haven, Conn.  06520

---------------

[2] Ron S. Dembo

	A Primal Truncated Newton Algorithm with
	Application to Large-Scale Nonlinear
	Network Optimization

	SOM Working Paper Series B #72
	Yale University School of
	Organization and Management
	March 1983

We describe a new, convergent, primal-feasible algorithm for
linearly constrained optimization.  It is capable of rapid
asymptotic behavior and has relatively low storage requirements.
Its application to large scale nonlinear network optimization
is discussed and computational results on problems of over
2,000 variables and 1,000 constraints are presented.  Indications
are that it could prove to be significantly better than known
methods for this class of problem.

Submitted by: Ron Dembo
Obtainable from:
	Ron Dembo
	(address above)

---------------



.