Büchi Complementation via Alternating Automata
Büchi Complementation via Alternating Automata
Büchi Complementation via Alternating Automata
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
<strong>Automata</strong> Theory Seminar<br />
<strong>Büchi</strong> <strong>Complementation</strong> <strong>via</strong><br />
<strong>Alternating</strong> <strong>Automata</strong><br />
Fabian Reiter<br />
July 16, 2012
<strong>Büchi</strong> <strong>Complementation</strong><br />
BA B<br />
2 Θ(n log n) BA B<br />
BA: <strong>Büchi</strong><br />
Automaton<br />
AA A<br />
AA A<br />
AA: <strong>Alternating</strong><br />
Automaton<br />
Expensive: If B has n states, B has 2 Θ(n log n) states in<br />
the worst case (Michel 1988, Safra 1988).<br />
Complicated: Direct approaches are rather involved.<br />
Consider indirect approach: detour over alternating automata.<br />
1 / 33
Transition Modes (1)<br />
Existential: some run is accepting<br />
q 0 q 1a q 2a q 3a q 4a q 5a · · ·<br />
q 0 q 1b q 2b q 3b q 4b q 5b · · ·<br />
q 0 q 1c q 2c q 3c q 4c q 5c · · ·<br />
q 0 q 1d q 2d q 3d q 4d q 5d · · ·<br />
q 0 q 1e q 2e q 3e q 4e q 5e · · ·<br />
Universal: every run is accepting<br />
q 0 q 1a q 2a q 3a q 4a q 5a · · ·<br />
q 0 q 1b q 2b q 3b q 4b q 5b · · ·<br />
q 0 q 1c q 2c q 3c q 4c q 5c · · ·<br />
q 0 q 1d q 2d q 3d q 4d q 5d · · ·<br />
q 0 q 1e q 2e q 3e q 4e q 5e · · ·<br />
2 / 33
Transition Modes (2)<br />
<strong>Alternating</strong>: in some set of runs every run is accepting<br />
q 0 q 1a q 2a q 3a q 4a q 5a · · ·<br />
q 0 q 1b q 2b q 3b q 4b q 5b · · ·<br />
q 0 q 1c q 2c q 3c q 4c q 5c · · ·<br />
q 0 q 1d q 2d q 3d q 4d q 5d · · ·<br />
q 0 q 1e q 2e q 3e q 4e q 5e · · ·<br />
q 0 q 1f q 2f q 3f q 4f q 5f · · ·<br />
q 0 q 1g q 2g q 3g q 4g q 5g · · ·<br />
q 0 q 1h q 2h q 3h q 4h q 5h · · ·<br />
q 0 q 1i q 2i q 3i q 4i q 5i · · ·<br />
3 / 33
Alternation and <strong>Complementation</strong><br />
Special case: A in existential mode<br />
A accepts iff ∃ run ρ : ρ fulfills acceptance condition of A<br />
A accepts iff ∀ run ρ : ¬(ρ fulfills acceptance condition of A)<br />
iff ∀ run ρ: ρ fulfills dual acceptance condition of A<br />
⇒ complementation ̂= dualization of:<br />
transition mode<br />
acceptance condition<br />
Want acceptance condition that is closed under dualization.<br />
4 / 33
Outline<br />
1 Weak <strong>Alternating</strong> Parity <strong>Automata</strong><br />
2 Infinite Parity Games<br />
3 Proof of the <strong>Complementation</strong> Theorem<br />
4 <strong>Büchi</strong> <strong>Complementation</strong> Algorithm<br />
5 / 33
Outline<br />
1 Weak <strong>Alternating</strong> Parity <strong>Automata</strong><br />
Definitions and Examples<br />
Dual Automaton<br />
2 Infinite Parity Games<br />
3 Proof of the <strong>Complementation</strong> Theorem<br />
4 <strong>Büchi</strong> <strong>Complementation</strong> Algorithm<br />
6 / 33
Preview<br />
Example ( (b ∗ a) ω )<br />
<strong>Büchi</strong> automaton B:<br />
b<br />
a<br />
a<br />
q 0 q 1<br />
b<br />
Equivalent WAPA A:<br />
a<br />
b<br />
a, b<br />
q 0<br />
2<br />
b<br />
•<br />
q 1<br />
1<br />
a<br />
q 2<br />
0<br />
7 / 33
Weak <strong>Alternating</strong> Parity Automaton<br />
☺<br />
Definition (Weak <strong>Alternating</strong> Parity Automaton)<br />
A weak alternating parity automaton (WAPA) is a tuple<br />
where<br />
Q<br />
Σ<br />
finite set of states<br />
finite alphabet<br />
δ : Q × Σ → B + (Q)<br />
q in<br />
initial state<br />
π : Q → N<br />
A := 〈Q, Σ, δ, q in , π〉<br />
parity function<br />
transition function<br />
(Thomas and Löding, ∼ 2000)<br />
B + (Q): set of all positive Boolean formulae over Q<br />
(built only from elements in Q ∪ {∧ , ∨, ⊤, ⊥})<br />
8 / 33
Transitions<br />
☺<br />
Example (a ω )<br />
a<br />
q 0<br />
2<br />
a<br />
a a<br />
q<br />
• 1<br />
•<br />
1<br />
•<br />
δ : Q × Σ → B + (Q)<br />
〈q 0 , a〉 ↦→ q 0 ∨ (q 1 ∧ q 2 )<br />
〈q 1 , a〉 ↦→ (q 0 ∧ q 1 ) ∨ (q 1 ∧ q 2 )<br />
q 2<br />
0<br />
a<br />
〈q 2 , a〉 ↦→ q 2<br />
Definition (Minimal Models)<br />
Mod ↓ (θ) ⊆ 2 Q : set of minimal models<br />
of θ ∈ B + (Q), i.e. the set of minimal<br />
subsets M ⊆ Q s.t. θ is satisfied by<br />
{<br />
true if q ∈ M<br />
q ↦→<br />
false otherwise<br />
Example<br />
Mod ↓ (q 0 ∨ (q 1 ∧ q 2 ))<br />
= {{q 0 }, {q 1 , q 2 }}<br />
9 / 33
Run Graph (1)<br />
☺<br />
Example (a ω )<br />
Accepting run:<br />
a<br />
q 0<br />
2<br />
a<br />
a a<br />
q<br />
• 1<br />
•<br />
1<br />
•<br />
q 0 , 0 q 0 , 1 q 0 , 2 q 0 , 3 q 0 , 4 q 0 , 5 · · ·<br />
Rejecting run:<br />
q 2<br />
0<br />
q 0 , 0 q 0 , 1 q 0 , 4<br />
q 1 , 2 q 1 , 3 q 1 , 4 q 1 , 5 · · ·<br />
q 2 , 2 q 2 , 3 q 2 , 4 q 2 , 5 · · ·<br />
a<br />
10 / 33
Run Graph (2)<br />
☺<br />
Definition (Run)<br />
A run of a WAPA A = 〈Q, Σ, δ, q in , π〉 on a word a 0 a 1 a 2 . . . ∈ Σ ω<br />
is a directed acyclic graph<br />
where<br />
V ⊆ Q × N<br />
R := 〈V , E〉<br />
with 〈q in , 0〉 ∈ V<br />
V contains only vertices reachable from 〈q in , 0〉.<br />
E contains only edges of the form 〈 〈p, i〉, 〈q, i + 1〉 〉 .<br />
For every vertex 〈p, i〉 ∈ V the set of successors is a minimal<br />
model of δ(p, a i )<br />
{<br />
q ∈ Q |<br />
〈<br />
〈p, i〉, 〈q, i + 1〉<br />
〉<br />
∈ E<br />
}<br />
∈ Mod↓ (δ(p, a i ))<br />
11 / 33
Acceptance<br />
☺<br />
Definition (Acceptance)<br />
Let A be a WAPA, w ∈ Σ ω and R = 〈V , E〉 a run of A on w.<br />
An infinite path ρ in R satisfies the acceptance condition of A<br />
iff the smallest occurring parity is even, i.e.<br />
min{π(q) | ∃ i ∈N:〈q, i〉 occurs in ρ} is even.<br />
R is an accepting run iff every infinite path ρ in R satisfies the<br />
acceptance condition.<br />
A accepts w iff there is some accepting run of A on w.<br />
12 / 33
Infinitely many a’s<br />
Example ( (b ∗ a) ω )<br />
a<br />
b<br />
a, b<br />
q 0<br />
2<br />
b<br />
•<br />
q 1<br />
1<br />
a<br />
q 2<br />
0<br />
Run on b ω :<br />
q 0 , 0<br />
b<br />
q 0 , 1<br />
b<br />
q 0 , 2<br />
b<br />
q 0 , 3<br />
b<br />
q 0 , 4<br />
b<br />
q 0 , 5<br />
b<br />
q 0 , 6<br />
b<br />
· · ·<br />
q 1 , 1 q 1 , 2 q 1 , 3 q 1 , 4 q 1 , 5 q 1 , 6 · · ·<br />
Run on (ba) ω :<br />
q 0 , 0<br />
b<br />
q 0 , 1<br />
a<br />
q 0 , 2<br />
b<br />
q 0 , 3<br />
a<br />
q 0 , 4<br />
b<br />
q 0 , 5<br />
a<br />
q 0 , 6<br />
b<br />
· · ·<br />
q 1 , 1 q 1 , 3 q 1 , 5 · · ·<br />
q 2 , 2 q 2 , 3 q 2 , 4 q 2 , 5 q 2 , 6 · · ·<br />
13 / 33
Dual Automaton (1)<br />
☺<br />
Definition (Dual Automaton)<br />
The dual of a WAPA A = 〈Q, Σ, δ, q in , π〉 is<br />
A := 〈Q, Σ, δ, q in , π〉<br />
where<br />
δ(q, a) is obtained from δ(q, a) by exchanging ∧ , ∨ and ⊤, ⊥<br />
π(q) := π(q) + 1<br />
for all q ∈ Q and a ∈ Σ<br />
14 / 33
Dual Automaton (2)<br />
Example ( (b ∗ a) ω )<br />
WAPA A:<br />
a<br />
b<br />
a, b<br />
δ(q 0 , a) = q 0<br />
δ(q 0 , b) = q 0 ∧q 1<br />
q 0<br />
2<br />
b<br />
•<br />
q 1<br />
1<br />
a<br />
q 2<br />
0<br />
δ(q 1 , a) = q 2<br />
δ(q 1 , b) = q 1<br />
δ(q 2 , a) = q 2<br />
δ(q 2 , b) = q 2<br />
Dual A:<br />
a, b<br />
b<br />
a, b<br />
δ(q 0 , a) = q 0<br />
δ(q 0 , b) = q 0 ∨q 1<br />
q 0<br />
3<br />
b<br />
q 1<br />
2<br />
a<br />
q 2<br />
1<br />
δ(q 1 , a) = q 2<br />
δ(q 1 , b) = q 1<br />
δ(q 2 , a) = q 2<br />
δ(q 2 , b) = q 2<br />
15 / 33
<strong>Complementation</strong> Theorem<br />
Main statement of this talk:<br />
Theorem (<strong>Complementation</strong>)<br />
The dual A of a WAPA A accepts its complement, i.e.<br />
L(A) = Σ ω \ L(A)<br />
(Thomas and Löding, ∼ 2000)<br />
16 / 33
Outline<br />
1 Weak <strong>Alternating</strong> Parity <strong>Automata</strong><br />
2 Infinite Parity Games<br />
3 Proof of the <strong>Complementation</strong> Theorem<br />
4 <strong>Büchi</strong> <strong>Complementation</strong> Algorithm<br />
17 / 33
Automaton vs. Pathfinder<br />
c<br />
b<br />
a<br />
player A<br />
find accepting run R<br />
player P<br />
find rejecting path in R<br />
18 / 33
Infinite Parity Game (1)<br />
☺<br />
Example (a ω )<br />
a<br />
q 0<br />
a a<br />
q<br />
• 1<br />
•<br />
1<br />
A: •<br />
2 a<br />
w = a ω<br />
q 2<br />
a<br />
0<br />
Game G A,w :<br />
q 0 , 0 {q 0 }, 0 q 0 , 1 {q 0 }, 1 q 0 , 2 · · ·<br />
{q 1 , q 2 }, 0<br />
{q 1 , q 2 }, 1 · · ·<br />
q 1 , 1 {q 0 , q 1 }, 1 q 1 , 2 · · ·<br />
q 2 , 1 {q 2 }, 1 q 2 , 2 · · ·<br />
19 / 33
Infinite Parity Game (2)<br />
☺<br />
Definition (Game)<br />
A game for a WAPA A = 〈Q, Σ, δ, q in , π〉 and w = a 0 a 1 a 2 . . . ∈ Σ ω<br />
is a directed graph<br />
where<br />
G A,w := 〈V A ˙∪ V P , E〉<br />
V A := Q × N (decision nodes of player A)<br />
V P := 2 Q × N (decision nodes of player P)<br />
E ⊆ (V A × V P ) ∪ (V P × V A )<br />
s.t. the only contained edges are<br />
• 〈 〈q, i〉, 〈M, i〉 〉 iff M ∈ Mod ↓ (δ(q, a i ))<br />
• 〈 〈M, i〉, 〈q, i + 1〉 〉 iff q ∈ M<br />
for q ∈ Q, M ⊆ Q, i ∈ N<br />
(Thomas and Löding, ∼ 2000)<br />
20 / 33
Playing a Game<br />
☺<br />
Definition (Play)<br />
A play γ in a game G A,w is an infinite path starting with 〈q in , 0〉.<br />
Definition (Winner)<br />
The winner of a play γ is<br />
player A iff the smallest parity of occurring V A -nodes is even<br />
player P · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · odd<br />
X ∈ {A, P}: a player,<br />
Definition (Strategy)<br />
X : its opponent<br />
A strategy f X : V X → V X<br />
for player X selects for every<br />
decision node of player X one of its successor nodes in G A,w .<br />
f X is a winning strategy iff player X wins every play γ that is<br />
played according to f X .<br />
21 / 33
Strategies<br />
Example<br />
Winning strategy for player A (so far):<br />
parities<br />
q 0 , 0 {q 0 }, 0 q 0 , 1 {q 0 }, 1 q 0 , 2 · · · q 0 ↦→ 2<br />
{q 1 , q 2 }, 1 · · ·<br />
q 1 , 1 {q 0 , q 1 }, 1 q 1 , 2 · · · q 1 ↦→ 1<br />
{q 1 , q 2 }, 0<br />
q 2 , 1 {q 2 }, 1 q 2 , 2 · · · q 2 ↦→ 0<br />
Not a winning strategy for player A:<br />
q 0 , 0 {q 0 }, 0 q 0 , 1 {q 0 }, 1 q 0 , 2 · · ·<br />
{q 1 , q 2 }, 1 · · ·<br />
q 1 , 1 {q 0 , q 1 }, 1 q 1 , 2 · · ·<br />
{q 1 , q 2 }, 0<br />
q 2 , 1 {q 2 }, 1 q 2 , 2 · · ·<br />
22 / 33
Outline<br />
1 Weak <strong>Alternating</strong> Parity <strong>Automata</strong><br />
2 Infinite Parity Games<br />
3 Proof of the <strong>Complementation</strong> Theorem<br />
Lemma 1<br />
Lemma 2<br />
Lemma 3<br />
Sublemma<br />
Putting it All Together<br />
4 <strong>Büchi</strong> <strong>Complementation</strong> Algorithm<br />
23 / 33
Lemma 1<br />
Let A be a WAPA and w ∈ Σ ω .<br />
Lemma 1<br />
Player A has a winning strategy in G A,w iff A accepts w.<br />
Explanation (oral):<br />
Player A wins every play γ<br />
played according to f A .<br />
G A,w :<br />
· · · q, i + 1<br />
There is a run graph R in which<br />
every path ρ is accepting.<br />
q, i + 1<br />
p, i {q, q ′ , q ′′ }, i q ′ , i + 1<br />
· · · q ′′ , i + 1<br />
R:<br />
p, i q ′ , i + 1<br />
q ′′ , i + 1<br />
24 / 33
Lemma 2<br />
Let A be a WAPA and w ∈ Σ ω .<br />
Lemma 2<br />
Player P has a winning strategy in G A,w iff A does not accept w.<br />
Explanation (oral):<br />
Player P wins every play γ<br />
played according to f P .<br />
G<br />
· · ·<br />
A,w :<br />
{.., q, ..}, i q, i + 1<br />
· · ·<br />
· · ·<br />
p, i {.., q ′ , ..}, i q ′ , i + 1<br />
· · ·<br />
· · ·<br />
{.., q ′′ , ..}, i q ′′ , i + 1<br />
· · ·<br />
(pointed out by Jan Leike)<br />
Every run graph R contains a<br />
rejecting path ρ.<br />
R:<br />
· · ·<br />
p, i q, i + 1<br />
· · ·<br />
· · ·<br />
R ′ : p, i q ′ , i + 1<br />
· · ·<br />
· · ·<br />
R ′′ : p, i q ′′ , i + 1<br />
· · ·<br />
25 / 33
Sublemma<br />
☺<br />
Let θ ∈ B + (Q) be a formula over Q.<br />
Sublemma<br />
S ⊆ Q is a model of θ iff for all M ∈ Mod ↓ (θ): S ∩ M ≠ ∅.<br />
Proof:<br />
W.l.o.g. θ is in DNF, i.e.<br />
∨<br />
θ =<br />
Then θ is in CNF, i.e.<br />
θ =<br />
M∈Mod ↓ (θ)<br />
∧<br />
M∈Mod ↓ (θ)<br />
∧<br />
q<br />
q∈M<br />
∨<br />
q<br />
q∈M<br />
Thus S ⊆ Q is a model of θ iff it contains at least one<br />
element from each disjunct of θ.<br />
26 / 33
Lemma 3 (1)<br />
Let A be a WAPA, A its dual and w = a 0 a 1 a 2 . . . ∈ Σ ω .<br />
Lemma 3<br />
Player A has a winning strategy in G A,w<br />
iff player P has a winning strategy in G A,w<br />
.<br />
Proof:<br />
⇒ Construct a winning strategy f P for player P in G A,w<br />
.<br />
· · ·<br />
⇐ Construct a winning strategy f A for player A in G A,w .<br />
· · ·<br />
27 / 33
Lemma 3 (2)<br />
⇒ Construct a winning strategy f P for player P in G A,w<br />
.<br />
At position 〈S, i〉∈V P<br />
in G A,w<br />
:<br />
· · · · · ·<br />
p, · · · i S, i q, · · · i + 1<br />
· · · · · ·<br />
f A : winning strategy for player A in G A,w<br />
Assume there is 〈p, i〉∈V A occurring in a<br />
play γ in G A,w played according to f A s.t.<br />
S ∈ Mod ↓ (δ(p, a i )) (otherwise don’t care).<br />
f A<br />
(<br />
〈p, i〉<br />
)<br />
=〈M, i〉 ⇒ M ∈ Mod↓ (δ(p, a i ))<br />
(sublemma)<br />
=⇒ There exists a q ∈S ∩ M.<br />
Define f P<br />
(<br />
〈S, i〉<br />
)<br />
:= 〈q, i + 1〉<br />
in G A,w :· · · · · ·<br />
p, i M, · · · i q, · · · i + 1<br />
· · · · · ·<br />
∀ γ: play in G A,w<br />
played according to f P<br />
∃ γ: play in G A,w played according to f A<br />
s.t. γ and γ contain the same V A -nodes.<br />
• Player A wins γ in G A,w .<br />
• ∀ q ∈Q : π(q) = π(q) + 1<br />
⇒ Player P wins γ in G A,w<br />
.<br />
28 / 33
Lemma 3 (3)<br />
⇐ Construct a winning strategy f A for player A in G A,w .<br />
At position 〈p, i〉∈V A<br />
in G A,w :· · · q, i + 1<br />
p, i M, · · · i q ′ , i + 1<br />
· · · q ′′ , i + 1<br />
in G A,w<br />
:<br />
· · ·<br />
S, i q, i + 1<br />
· · ·<br />
· · ·<br />
p, i S ′ , i q ′ , i + 1<br />
· · ·<br />
· · ·<br />
S ′′ , i q ′′ , i + 1<br />
· · ·M∗<br />
f P : winning strategy for player P in G A,w<br />
M ∗ := { q ∈ Q | ∃ S ∈Mod ↓ (δ(p, a i )) :<br />
f P<br />
(<br />
〈S, i〉<br />
)<br />
= 〈q, i + 1〉<br />
}<br />
(sublemma)<br />
=⇒ M ∗ is a model of δ(p, a i ).<br />
M: subset of M ∗ that is a minimal model<br />
M ⊆ M ∗ , M ∈ Mod ↓ (δ(p, a i ))<br />
Define f A<br />
(<br />
〈p, i〉<br />
)<br />
:= 〈M, i〉<br />
∀ γ: play in G A,w played according to f A<br />
∃ γ: play in G A,w<br />
played according to f P<br />
s.t. γ and γ contain the same V A -nodes.<br />
• Player P wins γ in G A,w<br />
.<br />
• ∀ q ∈Q : π(q) = π(q) − 1<br />
⇒ Player A wins γ in G A,w .<br />
29 / 33
All Three Lemmas<br />
☺<br />
Let A be a WAPA, A its dual and w ∈ Σ ω .<br />
Lemma 1<br />
Player A has a winning strategy in G A,w iff A accepts w.<br />
Lemma 2<br />
Player P has a winning strategy in G A,w iff A does not accept w.<br />
Lemma 3<br />
Player A has a winning strategy in G A,w<br />
iff player P has a winning strategy in G A,w<br />
.<br />
30 / 33
<strong>Complementation</strong> Theorem<br />
☺<br />
Theorem (<strong>Complementation</strong>)<br />
The dual A of a WAPA A accepts its complement, i.e.<br />
L(A) = Σ ω \ L(A)<br />
Proof:<br />
A accepts w<br />
(lemma 1)<br />
⇐⇒<br />
(lemma 3)<br />
⇐⇒<br />
(lemma 2)<br />
⇐⇒<br />
(Thomas and Löding, ∼ 2000)<br />
player A has a winning strategy in G A,w<br />
player P has a winning strategy in G A,w<br />
A does not accept w<br />
31 / 33
Outline<br />
1 Weak <strong>Alternating</strong> Parity <strong>Automata</strong><br />
2 Infinite Parity Games<br />
3 Proof of the <strong>Complementation</strong> Theorem<br />
4 <strong>Büchi</strong> <strong>Complementation</strong> Algorithm<br />
32 / 33
<strong>Büchi</strong> <strong>Complementation</strong> Algorithm<br />
☺<br />
BA B<br />
2Ω(n log n)<br />
BA B<br />
O(n 2 )<br />
WAPA A<br />
O(1)<br />
2 O(n)<br />
WAPA A<br />
Total complexity: 2 O(n2 )<br />
Can reach 2 O(n log n) (lower bound) by improving A → B.<br />
33 / 33
References<br />
Thomas, W. (1999)<br />
<strong>Complementation</strong> of <strong>Büchi</strong> <strong>Automata</strong> Revisited.<br />
In J. Karhumäki et al., editors, Jewels are Forever, Contributions on Th.<br />
Comp. Science in Honor of Arto Salomaa, pages 109–122, Springer.<br />
Klaedtke, F. (2002)<br />
<strong>Complementation</strong> of <strong>Büchi</strong> <strong>Automata</strong> Using Alternation.<br />
In E. Grädel et al., editors, <strong>Automata</strong>, Logics, and Infinite Games, LNCS<br />
2500, pages 61-77. Springer.<br />
Löding, C. and Thomas, W. (2000)<br />
<strong>Alternating</strong> <strong>Automata</strong> and Logics over Infinite Words.<br />
In J. van Leeuwen et al., editors, IFIP TCS 2000, LNCS 1872, pages<br />
521–535. Springer.<br />
Kupferman, O. and Vardi, M. Y. (2001)<br />
Weak <strong>Alternating</strong> <strong>Automata</strong> Are Not that Weak.<br />
In ACM Transactions on Computational Logic, volume 2, No. 3, July<br />
2001, pages 408–429.
Appendix
From BA to WAPA<br />
Given:<br />
B = 〈Q, Σ, δ, q in , F 〉: BA<br />
n = |Q|<br />
☺<br />
Construction (BA → WAPA)<br />
where<br />
A := 〈 Q ×{0, . . . , 2n} , Σ, δ ′ , 〈q<br />
} {{ }<br />
in , 2n〉, π 〉<br />
⎧∨<br />
⎪⎨ ∨<br />
δ ′ (〈p, i〉, a) := ∨<br />
π(〈p, i〉) := i<br />
⎪⎩ ∨<br />
O(n 2 )<br />
q∈δ(p,a)<br />
q∈δ(p,a)<br />
q∈δ(p,a)<br />
q∈δ(p,a)<br />
〈q, 0〉 if i = 0<br />
〈q, i〉 ∧ 〈q, i − 1〉 if i even, i > 0<br />
〈q, i〉<br />
〈q, i − 1〉<br />
if i odd, p /∈ F<br />
if i odd, p ∈ F<br />
for p ∈ Q, a ∈ Σ, i ∈ {0, . . . , 2n}<br />
(Thomas and Löding, ∼ 2000)<br />
Back
From WAPA to BA<br />
Given:<br />
A = 〈Q, Σ, δ, q in , π〉: stratified WAPA, i.e.<br />
∀p ∈Q ∀a∈Σ : δ(p, a) ∈ B +( {q ∈ Q | π(p) ≥ π(q)} )<br />
E ⊆ Q: all states with even parity<br />
Construction (WAPA → BA)<br />
where<br />
B := 〈 2<br />
} Q ×2<br />
{{ Q<br />
}<br />
, Σ, δ ′ , 〈{q in }, ∅〉, 2 Q ×{∅} 〉<br />
2 O(n) {<br />
〈M ′ , M ′ \E〉 ∣ M ′ ( ∧<br />
∈ Mod ↓<br />
δ ′ (〈M, ∅〉, a) :=<br />
δ ′ (〈M, O〉, a) :=<br />
{<br />
〈M ′ , O ′ \E〉<br />
for a ∈ Σ, M, O ⊆ Q, O ≠ ∅<br />
Back<br />
q∈M<br />
δ(q, a))}<br />
∣ M ′ (<br />
∈ Mod ↓<br />
∧q∈M δ(q, a)) ,<br />
O ′ ⊆ M ′ ,<br />
O ′ ( ∧<br />
∈ Mod ↓ q∈O<br />
δ(q, a))}<br />
(Miyano and Hayashi, 1984)<br />
☺