06.03.2013 Views

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

Artificial Intelligence and Soft Computing: Behavioral ... - Arteimi.info

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

The above theorem is called the fundamental theorem of GA or the<br />

Schema theorem. It is evident from the Schema theorem that for a given set<br />

of values of d(H), O(H), L, pc <strong>and</strong> pm , the population of schema H at the<br />

subsequent generations increases exponentially when fH > fav. This, in fact,<br />

directly follows from the difference equation:<br />

m H ( t +1) – m H (t) ≥ ( fH / fav –1 / K ) K mH (t) (15.9)<br />

where K = 1 – pc d (H) / (L – 1) –pm O(H). (15.10)<br />

⇒ ∆mH (t) ≥ K ( f H / fav -1/ K ) m H (t). (15.11)<br />

Replacing ∆ by (E –1), where E is the extended difference operator, we find<br />

(E –1- K1) m H (t) ≥ 0 (15.12)<br />

where K1 = K ( f H / fav -1 / K). (15.13)<br />

Since m H (t) in equation (15.12) is positive, E ≥ (1 +K1 ). Thus, the<br />

solution of (15.12) is given by<br />

m H (t) ≥ A ( 1 + K1) t (15.14)<br />

where A is a constant. Setting the boundary condition at t = 0, <strong>and</strong> substituting<br />

the value of K1 by (14.13) therein, we finally have:<br />

m H (t) ≥ mH (0) ( K fH / fav) t (15.15)<br />

Since K is a positive number, <strong>and</strong> f H / fav > 1, mH (t) grows<br />

exponentially with iterations. The process of exponential increase of mH (t)<br />

continues until some iteration r, when fH approaches fav. This is all about the<br />

proof of the schema theorem.<br />

15.4 The Markov Model for<br />

Convergence Analysis<br />

To study the convergence of the GA, let us consider an exhaustive set of<br />

population states, where ‘state’ means possible members (chromosomes) that<br />

evolve at any GA cycle. As an illustration, let us consider 2-bit chromosomes

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!