0000994: IMPLEMENTED: some NLP solvers can now recover from infinite/NAN target values

Viewing Issue Advanced Details [ Jump to Notes ]

[ View Simple ] [ Issue History ] [ Print ]

Category

Severity

Reproducibility

Date Submitted

Last Update

0000994

[ALGLIB] Optimization

feature

have not tried

2023-10-29 23:57

Reporter

SergeyB

View Status

public

Assigned To

SergeyB

Priority

normal

Resolution

implemented

Platform

Status

resolved

Projection

none

OS Version

ETA

none

Fixed in Version

4.01.0

Product Version

Target Version

4.01.0

Product Build

Summary

0000994: IMPLEMENTED: some NLP solvers can now recover from infinite/NAN target values

Description

Prior to ALGLIB 4.01, all optimizers stopped as soon as user callback returned infinite/NAN values in target/constraints vector or their Jacobian.

Now some ALGLIB solvers (SQP, DFO-LSA, 2PS) can recover from infinite values by reducing a trust radius. Upon successful recovery they continue optimization and add +800 to their completion code after return. If no successful recovery was possible (trust radius was decreased below lower bound), -8 completion code (critical error) is returned.

Steps To Reproduce

Additional Information

Programming language

Unspecified

Attached Files

Relationships

There are no notes attached to this issue.

Issue History
Date Modified	Username	Field	Change
2023-10-29 23:57	SergeyB	New Issue
2023-10-29 23:57	SergeyB	Status	new => assigned
2023-10-29 23:57	SergeyB	Assigned To	=> SergeyB
2023-10-29 23:57	SergeyB	Programming language	=> Unspecified
2023-10-29 23:57	SergeyB	Status	assigned => resolved
2023-10-29 23:57	SergeyB	Fixed in Version	=> 4.01.0
2023-10-29 23:57	SergeyB	Resolution	open => implemented

Mantis 1.1.6[^]