SC Chapter 8 Quantifier-free Non-linear Real Arithmetic

Questions about the lecture 'Satisfiability Checking' of the RWTH Aachen Chapter 8 Quantifier-free Non-linear Real Arithmetic

Questions about the lecture 'Satisfiability Checking' of the RWTH Aachen Chapter 8 Quantifier-free Non-linear Real Arithmetic


R. B.
This flashcard set covers advanced topics in quantifier-free non-linear real arithmetic at the university level. It delves into concepts like polynomial constraints, propagation, and the Cauchy bound, focusing on the characteristics and properties of polynomials, roots, and regions. The set also explores methods like the Sturm sequence, virtual substitution, and the Newton method, which are crucial for solving and understanding non-linear arithmetic problems. This flashcard set is particularly beneficial for university students and researchers in computer science who need to grasp the intricacies of real arithmetic and its applications in algorithmic problem-solving.
Karten
63
Lernende
1
Sprache
Englisch
Kategorie
Informatik
Stufe
Universität
Erstellt / Aktualisiert
18.02.2017 / 28.02.2017

Lernkarten

What is the definition?

[univariate form.definition.CAD.QFNRA]

p = akxk + … + a1x1 + a0x0 with ak != 0

What does square free means?

[definition.CAD.QFNRA]

No square factors i.e. no repeated roots

What is the degree of a polynomial?

[definition.CAD.QFNRA]

deg(p) = max0<j<k+1(Sumi=1neij) // Max sum of exponents

When is a formula phi linear?

[definition.CAD.QFNRA]

If every p has deg(p)<2

When is a formula phi non-linear?

[definition.CAD.QFNRA]

If one p has deg(p)>1

What is the solution space?

[definition.CAD.QFNRA]

Possible assignment for variables

What is the sign of a polynomial for input a?

[definition.CAD.QFNRA]

One element of {-1,0,1}

What are the characteristics?

[region.definition.CAD.QFNRA, 4]

1. Non-empty

2. Connected

3. Subset of Rn

4. Sign-invariant if all polynomials have same sign

What does connected means?

[region.definition.CAD.QFNRA]

If for all A,B holds (R cut A) cup (R cut B) is R

What are the characteristics?

[roots.definition.CAD.QFNRA, 4]

1. Amount lies between 0 and deg(p)

2. Sign changes only here

 

3. sigma(e) is number of sign changes in a sequences

4. For each a<b #distinctRealRoots of p is sigma(a)-sigma(b)

What are the types?

[roots.definition.CAD.QFNRA, 4]

1. Rational

2. Irrational

3. Not representable by radicals

4. Interval representation with one root in (l,r)

What is the idea?

[CAD.QFNRA, 3]

1. Assume finite set P of polynomials with sign condition for each p

2. Produce decompositions of Rn into a finite number of P-sign-invariant regions (So-called CAD cells)

3. Test sample point from each CAD cells

What are the outcomes of testing the sample point?

[idea.CAD.QFNRA]

If all sign conditions satisfied for #point>0 return SAT else return UNSAT

What is the cauchy bound C?

[for R.CAD.QFNRA, 2]

1. C = 1+max(|ai|/|ak|) >= real root |e| with ak most-left coefficient

2. (-C,C) is interval containing all real roots

What is the idea?

[sturm seq.for R.CAD.QFNRA, 2]

1. Counts real roots in open interval

2. Finites sequence of polynomials of decreasing degree

How to compute?

[sturm seq.for R.CAD.QFNRA, 3]

1. p0=p

2. p1=p’

3. pi=-rem(pi-2,pi-1) with rem is remainder of polynomial division of p0 by p1

What are the properties?

[sturm seq.for R.CAD.QFNRA, 4]

1. If p(a)=0 then sgn(pi(a))=sgn(p’(a)) and sgn(p’(a)!=0

2. If pi(a)=0 then sgn(pi-1(a))=-sgn(pi+1(a))

3. pl does not change its sign

4. rem(pl-1,pl)=0

What is the procedure?

[for R.CAD.QFNRA]

Split C until each sub-interval I contains at most one real root

What are possible enhancements?

[for R.CAD.QFNRA, 2]

1. Incrementality with split regarding to new constraint

2. Collect infeasible subsets for each CAD interval one constraint not satisfying the sign condition

What are the two phases?

[for Rn.CAD.QFNRA, 2]

1. In projection phase decrease dimension step by step

2. In construction phase increase dimension step by step

What holds for delineability for P?

[for Rn.CAD.QFNRA, 3]

1. #roots of pi(a) is constant

2. #differentRoots of pi(a) is constant

3. #commonRoots of pi(a) and pj(a) is constant

What holds for CAD C?

[for Rn.CAD.QFNRA, 4]

1. Cup C = Rn

2. For all cells of C the cut is empty

3. If n=1 then every cell is maximal P-sign Invariant region

4. If n>1 then every cell of C’ is P-delineable and cell \subseteq cell’ x R is a maximal P-sign invariant region

What holds for CAD projection?

[for Rn.CAD.QFNRA, 3]

1. Mapping decreasing one dimension

2. If any region R of Rn-1 is proj(P)-sign invariant iff R is P-delineable

 

3. Usually |proj(P)| = |P|² // O(|P|2^n-1) for univariate case

Lernen