AdaPower Logged in as Guest
Ada Tools and Resources

Ada 95 Reference Manual
Ada Source Code Treasury
Bindings and Packages
Ada FAQ


Join >
Home >
Articles >
Getting Started >
Ada FAQ >
Source Treasury >
Books & Tutorials >
Packages for Reuse >
Latest Additions >
Ada Projects >
Press Releases >
Ada Audio / Video >
Home Pages >
Links >
Contact >
About >
Login >
Back
Unconstrained Minimization of a Real-Valued Function (Vinicius F. Arcaro)

NLP is a package for the solution of the unconstrained
minimization of a real-valued function f(x). It uses the Memoryless
Quasi-Newton Method (self scaling BFGS). A description of this method
can be found in the following book:

Linear and Nonlinear Programming
David G. Luenberger
Addison-Wesley, 1984, ISBN 0-201-15794-2

The NLP package can be downloaded here.

Example of package use:

-- f : real-valued function. -- x : vector of variables. -- g : gradient vector of f(x). -- n : number of variables. -- maxminor : an upper bound for the number of iterations (usually 10 * n). -- minor : number of iterations. -- maxnorm : an upper bound for the infinity norm of the gradient vector. -- optimal : boolean variable (true if success). -- maxlsi : an upper bound for the number of cubic fits of the line search. -- It must be greater that one (usually 20). -- lstol : Controls the accuracy of the line search. It must lie in the -- range 0.0 <= lstol < 1.0. Decreasing lstol tends to increase the accuracy -- of the line search (usually 0.1). -- author: Vinicius Fernando Arcaro -- e-mail: vfa@turing.unicamp.br, vfa@widesoft.com.br -- date: October, 1996 with text_io, ada.long_float_text_io, ada.numerics.long_elementary_functions, unchecked_deallocation, terminology, nlp; procedure test is use text_io, ada.long_float_text_io; optimal : boolean; n,maxminor,maxlsi,minor : integer; maxnorm,lstol,f,a : long_float; x,g : terminology.fvector_ptr; procedure free_fvector is new unchecked_deallocation(terminology.fvector,terminology.fvector_ptr); procedure energy (n : in integer; f : out long_float; x : in terminology.fvector_ptr; g : in terminology.fvector_ptr) is use ada.numerics.long_elementary_functions; begin f := (x(1) - a) ** 4 + (x(1) - a * x(2)) ** 2; g(1) := 4.0 * (x(1) - a) ** 3 + 2.0 * (x(1) - a * x(2)); g(2) := -2.0 * a * (x(1) - a * x(2)); end energy; procedure equilibrium is new nlp.minimize(energy); begin -- global variable to be used by procedure energy a := 2.0; -- parameters to be used by procedure equilibrium n := 2; maxminor := 10 * n; maxlsi := 20; maxnorm := 1.0E-04; lstol := 1.0E-01; -- dynamic memory allocation x := new terminology.fvector (1 .. n); g := new terminology.fvector (1 .. n); -- starting point x(1) := 0.0; x(2) := 3.0; equilibrium(n,maxminor,maxlsi, minor, maxnorm,lstol, f, x,g, optimal); put("x(1) = "); put(x(1)); new_line; put("x(2) = "); put(x(2)); new_line; -- dynamic memory deallocation free_fvector(x); free_fvector(g); end test;


(c) 1998-2004 All Rights Reserved David Botton