Contents

• List of Figures
• List of Tables
• Choices, Chances and Results
  • Choices: The Dorm Assignment Problem
    • Background
    • The Environment
  • Chances: The Simulated Annealing Solution
    • Introduction to Annealing and Simulated Annealing
    • Optimizing Dorm Assignments
  • Results: Fall 1998 - Fall 2000
  
  
• Choices: The Dorm Assignment Problem
  • The Objective Function
    • Neighborhood Landscapes
    • Constraint Costs
    • Modifying the Objective Function
  • Choices in Policy
    • Inflationary Costs
    • Multiple Assignment Runs
  • Choices in the Data
  
  
• Chances: The Simulated Annealing Solution
  • Simulated Annealing Schedules
  • Dorm Assignments: Five Cooling Schedules
  • Guaranteed Local Minimum
  • Discussion: Adaptive Schedules
  • Handling Ties
  
  
• Results: Fall 1998 - Fall 2000
  • The Artifact
    • Choosing Constraint Coefficients
    • Between Two Systems
    • Batch Management
    • Checkpoint Restart
    • Security Issues
    • Efficiency Considerations
  • Case Studies
    • Fall 1998, versions v0.01 - v0.39
    • Fall 1999, versions v0.40 - v0.85
    • Spring 2000, versions v0.90 - v1.03
    • Fall 2000, versions v1.04 - v1.15
  • Experimental Results
    • Characteristics of the Data
    • Methodology
    • Random Optimization
    • Taking Uphill Moves
    • Importance of Lateral Moves on a Mesa
    • Varying Probabilities for Better Moves
    • Reaching a Local Minimum
    • Comparing the Schemes: Scores and Attempts
  
  
• Conclusions
• Appendices
• HMS Rules for Assignment with the DAO
• The DAO Objective Function
• Bibliography
• About this document ...

download thesis