Title: Approximate Satisfiability and Equivalence
1Approximate Satisfiability and Equivalence
- Michel de Rougemont
- University Paris II LRI
- Joint work with E. Fischer, Technion,
- F. Magniez, LRI ,
- LICS 2006
-
2Plan
- Tester on a class K approximation of decision
problems - Equality between two strings (trees) (e2, e)
Tolerant tester , additive approximation of the
Edit Distance with Moves. - Membership is w in L ?
- Equivalence tester between two regular
properties polynomial time algorithm (Exact
Equivalence is PSPACE complete) - Generalizations regular trees, context-free
languages, infinite words, - Current research probabilistic systems.
3Motivations
- Decisions on noisy inputs (distance to a
language) - Model Checking can we approximate hard problems?
Bounded MC, Abstraction, .. - Black-Box Checking does B satisfies P ?
L
B
41. Testers on a class K
- Let F be a property on a class K of structures
U - An e -tester for F is a probabilistic algorithm
A such that - If U F, A accepts
- If U is e far from F, A rejects with high
probability - F is testable if there is a probabilistic
algorithm A such that - A is an e -tester for all e
- Time(A) is independent of nsize(U).
- Robust characterizations of polynomials, R.
Rubinfeld, M. Sudan, 1994 - Property Testing and its connection to Learning
and Approximation. O. Goldreich, S. Goldwasser,
D. Ron, 1996. -
- Tester usually implies a linear time corrector.
(e1, e2)-Tolerant Tester -
5Approximate Satisfiability and Equivalence
- Satisfiability T F
- Approximate Satisfiability T F
- Approximate Equivalence
- Image on a class K of trees
G
6 Edit Distances with Moves
- Classical Edit DistanceInsertions, Deletions,
Modifications - Edit Distance with moves dist(w,w)
- 0111000011110011001
- 0111011110000011001
- 3. Edit Distance with Moves generalizes to
Ordered Trees
7Tester for equality Block and Uniform statistics
W001010101110 length n, b.stat
consecutive subwords of length k, n/k
blocks u.stat any subwords of length k, n-k1
blocks, shingles
For k2, n/k6
8 Tester for equality
Edit distance with moves. NP-complete problem,
but approximable in constant time with additive
error. Uniform statistics ( )
W001010101110 Theorem 1. u.stat(w)-u.stat(w)
approximates dist(w,w)/n. Sample N subwords
of length k, compute Y(w) and Y(w) Lemma
(Chernoff). Y(w) approximates u.stat(w). Corollar
y. Y(w)-Y(w) approximates dist(w,w)/n. Tester
1 If Y(w)-Y(w) lte. accept, else reject.
9Theorem 1 Soundness and Robustness
- Soundness e-close strings have close statistics
- Robustness e-far strings have far statistics
- We prove
- b.stat is robust
- u.stat is sound
- u.stat is robust (harder)
-
hard
10Robustness of b.stat
Robustness of b-stat
11Soundness of u.stat
Soundness of u-stat Simple edit Move
wA.B.C.D, wA.C.B.D Hence, for e2.n
operations, Remark b.stat is not
sound. Problem robustness of u.stat ? Harder!
We need an auxiliary distribution and two key
lemmas.
12Statistics on words
Block statistics b.stat
k
Uniform statistics u.stat
k
Block Uniform statistics bu.stat
t
k-t
K
13Uniform Statistics
A
B
Lemma 2
14Block Uniform Statistics
Lemma 1
15Robustness of the uniform Statistics
Lemma 2 Lemma 1
(Probabilistic method)
- Tolerant tester
- Theorem for two words w and w large enough, the
tester - Accepts if ww with probability 1
- Accepts if w,w are e2-close with probability 2/3
- Rejects if w,w are e-far with probability 2/3
163. Testers for Membership and Equivalence
- Membership decide if
- Inclusion and Equivalence
- Equivalence tester
-
17Automata for Regular languages
A automaton with m states on S, Ak automaton
with m states on Sk. Basic property Proposit
ion Caratheodorys theorem in dimension d,
convex hull of N points can be decomposed into in
the union of convex hulls of d1 points. Large
loops can be decomposed. Small loops (less than
mA) suffice.
18Approximate Parikh mapping
Lemma For every X in H, w of size n s. t.
H is a fair representation of L
w
X .
b-stat(w)
19Example
Hstat(w) w in r is a union of
polytopes. 2 Polytopes for r.
Y(w)
Membership Tester
20Construction of H in polynomial time
Enumeration of all m loops Number of b-stat
of words of length m on Sk is less than
Some loops have same b-stat ABBA and
BBAA Construct H by matrix iteration
21Membership tester
Membership Tester for w in L (regular) 1.
Construction of the tester Precompute He 2.
Tester Compute Y(w) (approx. b.stat(w)).
Accept iff Y(w) is at distance less
than e to He Construction Time is Tester
query complexity in time complexity
in Remark 1 Time complexity of previous
testers was exponential in m. Remark 2 The same
method works for L context-free.
22Equivalence Tester for regular properties
Time polynomial in mMax(A , B )
233. Generalization Trees
(1,.)c
(1,(1,(1,.),1),.)c
T squeleton
T Ordered (extended) Tree of rank 2
W word with labels. Apply u.stat on W and
define u.stat(T).
24Infinite words
- Buchi Automata. Distance on infinite
words - Two words are e-close if
- A word is e-close to a language L if there exists
w in L s. t. W and w are e-close. - Statistics set of accumulation points of
- H compatible loops of connected components of
accepting states - Tester for Buchi Automata
- Compute HA and HB
- Reject if HA and HB are different.
- Approximate Model-Checking
25Other Logics
Equivalence of Context-free grammars is
undecidable, Approximate Equivalence in
exponential time. Consider formulas in different
Logics (LFP, m-calculus,..). Can Equivalence,
Implication be approximated on a definable class
K with a distance? Definability and
approximation can first-order definable classes
of trees testable with the Edit distance?
264. Probabilistic Systems
Probabilistic Automata Ma is a stochastic
matrix for letter a. If wa1 a1 . an then Mw
Ma1 . Man
PM Probabilistic Membership Is ut.Mw.vgt ? ?
- APM Approximate Probabilistic Membership Let P
ut.Mw.vgt ? - Decide if w satisfies P or if w is e -far
from P.
27Approximations for Probabilistic Automata
- Approximate probabilities
- Introduce e around ?
- Approximate membership
- Approximate Equivalence (Tzeng 92) is harder than
Equivalence. - Approximate distances between states
- Generalization of bisimulation
- Desharnais et al., Van Breugel-Worell
- Our approach Approximation on the input
- http//www.lri.fr/mdr/verap
28Basic Decompositions in H
s3ccc
s4aa
s3bc
s1abc
s2ba
H1
H2 Waabaaaaababcabcabcabcabcabcbc close
to W(aa)3(ba)2(abc)6 N Samples approximate
ustat(W) close to ?1.ustat((aa))
?2.ustat((ba)) ?3.ustat((abc))
29Basic Decompositions in H1
s4aa
s3
s1abc
s2ab
- For each summit s, basic loop in A, let
h(s,n)Probability to follow s after n iterations
of s - Analyze all loops mutliple of s h(s,n) rn for
n large enough. - Analyze all possible decompositions of ustat(w)
in H
30Claim
- Hypothesis all simple loops are distinct.
- Input W of length n
- Claim Upper bound for ut.Mw.v for W close to
W. - ?1.ustat((aa)) ?2.ustat((ba))
?3.ustat((abc)) indicates densities ?1, ?2, ?3
to follow aa, ba, abc on Hi. - We need to connect loops aa, ba, abc by some
inputs there are finitely many possibilities.
Let Ci the best probability.
31Tester for APM in O(1)
- Input W of length n
- Tester(W,k)
- Sample W with N(k) samples,
- Select H such that ustat is close,
- Decompose ustat on possible subpolytopes Hi with
at most d1 summits, and obtain a bound Bi, - Consider all possible links on Hi, let Ci the
optimal bound, - Let DMaxi Ci.Di
If ?lt D, Accept else Reject.
32Non determinism and Probabilistic
- Can we combine both non determinism and
probabilistic behaviors? - Stationary distributions for a given scheduler
are distributions on the states, for which there
is also a polytope representation. Classical
results exist about positional schedulers. - Problem does the projection of these
distributions on ustat vectors keep the
distances? Thesis of Mathieu Tracol. - For large scale systems, evolutionary games also
provide a statistical representation of the
states. Can we predict approximate properties of
the Equilibria?
33Conclusion
- Tolerant Tester for Equality on strings under the
Edit Distance with Moves - Additive approximation in O(1) of the EDM
- Equivalence tester for automata
- Polynomial time approximate algorithm
(PSPACE-complete) - Generalization to Buchi automata approximate
Model-Checking - Context-Free Languages exponential algorithm
(exact problem is undecidable) - Generalization to trees, infinite words
- Probabilistic systems.