Revision d15e0c422273bfb0b837bbf4426e89a664273791 (click the page title to view the current version)

Continuous GA

Changes from d15e0c422273bfb0b837bbf4426e89a664273791 to 5f8dc9a40dfc8a0e7083d53372e3bef7b163598e

---
title: Continuous GA
categories: session
---

# Reading
# Reading and Preparation

+ Haupt and Haput 
1. Make sure to complete last week's exercises and make sure you
   understand how the GA system works.
2. The goal for this week is to expand the repertoire of representations,
   cross-over functions, and mutation functions.  In particular we consider
   floating point vectors as chromosomes.
3. Read Haupt and Haput 
    + Chapter 3 on the Continuous GA.  
      You should read this in relation to the exercises you did last week.
      How would you adapt the GA to handle floating point chromosomes?
    + Chapter 4


# Briefing

+ The Cross-Over function
+ The Mutation function
1.  What do we know about GA?

# Exercises

As we did last week, we will use the demo on
[github](https://github.com/hgeorgsch/pygax).  
You may want to check for a new release.

The exercises below are phrased more briefly than last week,
but this does not mean that you should be as brief in your solutions.
The intention is to play around and understand
*what configurations make a difference to the performance* of a GA.

## Recap

If you did not complete many exercises last week, you should
first check that you are able to

1. run the GA
2. make and optimise different cost functions
3. vary the mutation and/or cross-over function

## Discussion

What do we need to change to turn the binary GA into a Continuous GA?

Inspect your demo code and discuss if the changes made there are
appropriate.

## Initial testing

1. Test the Continuous GA using the test script and make sure it works
   on your system.
2. Test the (continuous) cost functions that you tested last week on the
   continuous GA.
3. Are there cost functions where the binary GA performs better?
   Where the Continuous GA performs better?

## Varying the GA

Consider alternative mutation and crossover functions for
the Continuous GA.  Implement a few variations and test if you
can improve the performance compared to the default.

# Midway status

# New Problems

Chapter 4 in H&H presents a number of practical problems.
Each table chooses one or two to implement, test, and briefly
present to the class.

1.  Mary had a little lamb
2.  Genetic (visual) art
3.  Word Guess 
4.  Locating an emergency response unit
5.  Antenna Array Design
6.  Evolution of Horses

Even if you are not able to implement a complete solution, you
should be able to present the problem and the representation of
chromosomes.

# Debrief