15.08.2013 Views

General Computer Science 320201 GenCS I & II Lecture ... - Kwarc

General Computer Science 320201 GenCS I & II Lecture ... - Kwarc

General Computer Science 320201 GenCS I & II Lecture ... - Kwarc

SHOW MORE
SHOW LESS

Create successful ePaper yourself

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

P.1.2.3 then there are ei ∈ Ebool, such that fi = fei and C(ei) = 2 n−1 + d. (IH)<br />

P.1.2.4 thus f = fe, where e := xn ∗ e0 + xn ∗ e1 and κ(n) = 2 · 2 n−1 + 2d + 4.<br />

c○: Michael Kohlhase 149<br />

The next proof is quite a lot of work, so we will first sketch the overall structure of the proof,<br />

before we look into the details. The main idea is to estimate a cleverly chosen quantity from<br />

above and below, to get an inequality between the lower and upper bounds (the quantity itself is<br />

irrelevant except to make the proof work).<br />

A Lower Bound for the Cost of BF with n Variables<br />

Theorem 270 κ ∈ Ω( 2 n<br />

log 2 (n) )<br />

Proof: Sketch (counting again!)<br />

P.1 the cost of a function is based on the cost of expressions.<br />

P.2 consider the set En of expressions with n variables of cost no more than κ(n).<br />

P.3 find an upper and lower bound for #(En): (Φ(n) ≤ #(En) ≤ Ψ(κ(n)))<br />

P.4 in particular: Φ(n) ≤ Ψ(κ(n))<br />

P.5 solving for κ(n) yields κ(n) ≥ Ξ(n) so κ ∈ Ω( 2 n<br />

log 2 (n) )<br />

We will expand P.3 and P.5 in the next slides<br />

c○: Michael Kohlhase 150<br />

A Lower Bound For κ(n)-Cost Expressions<br />

Definition 271 En := {e ∈ Ebool | e has n variables and C(e) ≤ κ(n)}<br />

Lemma 272 #(En) ≥ #(B n → B)<br />

Proof:<br />

P.1 For all fn ∈ B n → B we have C(fn) ≤ κ(n)<br />

P.2 C(fn) = min({C(e) | fe = fn}) choose efn with C(efn ) = C(fn)<br />

P.3 all distinct: if eg ≡ eh, then feg = feh and thus g = h.<br />

Corollary 273 #(En) ≥ 2 2n<br />

Proof: consider the n dimensional truth tables<br />

P.1 2n entries that can be either 0 or 1, so 22n possibilities<br />

so #(B n → B) = 2 2n<br />

c○: Michael Kohlhase 151<br />

An Upper Bound For κ(n)-cost Expressions<br />

P.2 Idea: Estimate the number of Ebool strings that can be formed at a given cost by looking at<br />

the length and alphabet size.<br />

80

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

Saved successfully!

Ooh no, something went wrong!