3
Multi-objective Scheduling for NASA‘s Future Deep Space Network Array by Mark D. Johnston Roman Barták (commentator) Roman Barták (commentator) Charles University, Czech Republic Charles University, Czech Republic [email protected] http://ktiml.mff.cuni.cz/~bartak Commentary on the paper

Multi-objective Scheduling for NASA‘s Future Deep Space Network Array by Mark D. Johnston

Embed Size (px)

DESCRIPTION

Commentary on the paper. Multi-objective Scheduling for NASA‘s Future Deep Space Network Array by Mark D. Johnston. Roman Barták (commentator) Charles University, Czech Republic [email protected] http://ktiml.mff.cuni.cz/~bartak. Problem specification. Formal specification - PowerPoint PPT Presentation

Citation preview

Multi-objective Scheduling for NASA‘s Future Deep Space Network Arrayby Mark D. Johnston

Roman Barták (commentator)Roman Barták (commentator)Charles University, Czech RepublicCharles University, Czech Republic

[email protected]://ktiml.mff.cuni.cz/~bartak

Commentary on the paper

IWPSS 2006, Roman Barták

Problem specificationProblem specification

Formal specification The problem is described at a narrative level

only (OK, it is a future real-life problem). Still, a more formal description would help to

understand technical details.

Hard constraints vs. objectives Almost everything is objective (soft

constraint), even if it looks like a hard constraint. A crisp border between hard and soft

constraints would be appropriate.

Data availability It would be nice to have some benchmark data

even for a simplified version of the problem.

IWPSS 2006, Roman Barták

Solving technologySolving technology

Single technology Only evolutionary algorithms tried to solve the

problem. Isn’t problem specification conformed to solving

technology? What if the hard constraints are changed? Is providing more solution candidates a real advantage?

Dynamicity of the problem New requests, changed weather etc. are solved

via re-scheduling. The new schedule may be far from the

published solution which could make some customers unhappy (further re-scheduling).

Use similarity of solutions as additional objective.