Inconsistency Proofs for ASP: The ASP-DRUPE Format

Answer Set Programming (ASP) solvers are highly-tuned and complex procedures that implicitly solve the consistency problem, i.e., deciding whether a logic program admits an answer set. Verifying whether a claimed answer set is formally a correct answer set of the program can be decided in polynomial time for (normal) programs. However, it is far from immediate to verify whether a program that is claimed to be inconsistent, indeed does not admit any answer sets. In this paper, we address this problem and develop the new proof format ASP-DRUPE for propositional, disjunctive logic programs, including weight and choice rules. ASP-DRUPE is based on the Reverse Unit Propagation (RUP) format designed for Boolean satisfiability. We establish correctness of ASP-DRUPE and discuss how to integrate it into modern ASP solvers. Later, we provide an implementation of ASP-DRUPE into the wasp solver for normal logic programs. This work is under consideration for acceptance in TPLP.

READ FULL TEXT VIEW PDF

page 1

page 2

page 3

page 4

03/15/2012

Possibilistic Answer Set Programming Revisited

Possibilistic answer set programming (PASP) extends answer set programmi...
08/05/2016

The Power of Non-Ground Rules in Answer Set Programming

Answer set programming (ASP) is a well-established logic programming lan...
03/26/2015

ASPeRiX, a First Order Forward Chaining Approach for Answer Set Computing

The natural way to use Answer Set Programming (ASP) to represent knowled...
08/05/2020

Modular Answer Set Programming as a Formal Specification Language

In this paper, we study the problem of formal verification for Answer Se...
08/05/2016

Stable-Unstable Semantics: Beyond NP with Normal Logic Programs

Standard answer set programming (ASP) targets at solving search problems...
09/22/2020

LP2PB: Translating Answer Set Programs into Pseudo-Boolean Theories

Answer set programming (ASP) is a well-established knowledge representat...
09/17/2021

How to Split a Logic Program

Answer Set Programming (ASP) is a successful method for solving a range ...

1 Introduction

Answer Set Programming (ASP) [Brewka et al. (2011)] is a logic-based declarative modeling language and problem solving framework [Gebser et al. (2012)]

for hard computational problems and an active research area in artificial intelligence (AI) and knowledge representation and reasoning. It has been applied both in academia 

[Balduccini et al. (2006), Gebser et al. (2010), Gebser et al. (2011)] and industry [Gebser et al. (2011), Guziolowski et al. (2013), Ricca et al. (2012)]. In propositional ASP questions are encoded by atoms combined into rules and constraints which form a logic program. Solutions to the program consist of sets of atoms called answer sets; if no solutions exist then the program is inconsistent.

Knowledge representation languages like ASP are usually considered explainable AI, as they are based on deduction, which is an explainable procedure. For example, we can easily explain answer sets of a normal logic program in terms of program reducts and fixpoint operators [Liu et al. (2010)]. In this case, we may argue that answer sets are self-explanatory, and therefore ASP systems providing answer sets are explainable AI systems. However, modern ASP systems do not provide any explanation for inconsistent programs; there is no witness that can be checked or evidence of the correctness of the refutation of the input program. Hence, even if inconsistency of logic programs is anyhow explainable with mathematical rigour, ASP systems are essentially black-boxes in this case, and just report the absence of answer sets. We believe that adding inconsistency proofs in ASP systems is important to make them explainable for inconsistent programs, but also provides auditability for consistent programs. Thanks to a duality result in the literature [Pearce (1999)], such inconsistency proofs for ASP can be also used as a certificate for the validity of some formulas of intuitionistic logic and other intermediate logics. A further application of these inconsistency proofs is query answering in ASP, which is usually achieved by inconsistency checks. There, the goal is to provide a witness for cautiously true answers of a given query.

Modern ASP solvers have been highly influenced by SAT solvers, which solve the Boolean satisfiability problem and are often based on conflict-driven clause learning [Silva and Sakallah (1999)]. Typically, ASP solvers aim for computing an answer set of a given program, and therefore solve the consistency problem that asks whether a given program has an answer set. This problem is on the second level of the polynomial hierarchy when allowing arbitrary propositional disjunctive programs as input and on the first level when restricting to disjunction-free programs [Truszczyński (2011)]. As already stated, while consistency of a program can be easily verified given such a computed answer set, verifying whether a solver correctly outputted that a program is inconsistent, is not immediate. Given that ASP solvers are also used for critical applications [Gebser et al. (2018), Haubelt et al. (2018)], their correctness is of utter importance.

When looking at SAT solvers, various techniques have been developed to ensure correctness of unsatisfiability, such as clausal proof variants [Gelder (2008), Goldberg and Novikov (2003)] based on clauses that have RUP (reverse unit propagation) and RAT (Resolution Asymmetric Tautology) property. These proof formats share verifiability in polynomial time in the size of the proof and input formula and can be tightly coupled with modern solving techniques. A solver outputs such a proof during solving. Thereby, the correctness of solving can be verified by a relatively simple method for every input instance. While there are variants of these proofs for various problems, such as extensions to verify the validity of quantified Boolean formulas [Heule et al. (2013), Wetzler et al. (2014)] (QRAT [Heule et al. (2014)] and QRAT+ [Lonsing and Egly (2018)]), to our knowledge such a format is not yet available for verifying ASP solvers. One approach to certify inconsistency of a given normal program is to translate the program into a SAT formula in polynomial time [Lin and Zhao (2003), Janhunen (2006)] and obtain a proof from a SAT solver, e.g., via a RAT-based proof format, to verify that indeed the program is inconsistent. Unfortunately, this approach does not take the techniques into account that are employed by state-of-the-art ASP solvers and therefore seem to lack efficiency and scalability. Further, this is still not a suitable technique for disjunctive programs, nor to verify whether internally the ASP solver is able to correctly explain the obtained result.

We follow this line of research and establish the following novel results for ASP:

  1. We present the proof format ASP-DRUPE based on RUP for logic programs given in SModels [Syrjänen (2000)] or ASPIF [Gebser et al. (2016)] (restricted to ASP without theory reasoning) input format including disjunctive programs and show its correctness.

  2. We provide an algorithm for verifying that a given solving trace in the ASP-DRUPE format is indeed a valid proof for inconsistency of the input program. This algorithm works in polynomial time in the size of the given solving trace.

  3. We illustrate on an abstract ASP solving algorithm how one can integrate ASP-DRUPE into state-of-the-art ASP solvers like clasp [Gebser et al. (2012)] and wasp [Alviano et al. (2015)].

  4. We provide an implementation in a variant of wasp, where ASP-DRUPE is integrated for normal ASP. This variant of wasp is able to not only explain inconsistency for inconsistent logic programs, but also provides auditability in case of consistency for verifying whether the provided answer set was indeed correctly obtained.

Related Work.

Heule et al. [Heule et al. (2013)] presented a proof format based on the RAT property and subsequently a program to validate solving traces in this format [Wetzler et al. (2014)]. Extended resolution allows to polynomially simulate the DRAT format [Kiesl et al. (2018)] and vice-versa [Wetzler et al. (2014)]. Many advanced techniques, such as XOR reasoning [Philipp and Rebola-Pardo (2016)] as well as symmetry breaking [Heule et al. (2015)] can be expressed in DRAT and efficient, verified checkers based on RAT have been developed [Cruz-Filipe et al. (2017)]. Further, RAT is also available for QBF [Heule et al. (2014)] and has been extended to cover a more powerful redundancy property [Lonsing and Egly (2018)].

2 Preliminaries

2.1 Answer Set Programming (ASP)

We follow standard definitions of propositional ASP [Brewka et al. (2011)] and use rules defined by the SModels [Syrjänen (2000)] or ASPIF [Gebser et al. (2016)] (restricted to ASP without theory reasoning) input format, which is widely supported by modern ASP solvers. In particular, let , , be non-negative integers such that , , , propositional atoms, , , , non-negative integers. A choice rule is an expression of the form , a disjunctive rule is of the form , and a weight rule is of the form , where . A rule is either a disjunctive, a choice, or a weight rule. A (disjunctive logic) program  is a finite set of rules. For a rule , we let , , , and is a set of literals, i.e., an atom or the negation thereof. We denote the sets of atoms occurring in a rule  or in a program  by and . For a weight rule , let map literal  to its weight  in rule  if for , or if  for , and to otherwise. Further, let for a set  of literals and let be its bound. A normal (logic) program  is a disjunctive program  with for every . The positive dependency digraph  of is the digraph defined on the set  of atoms, where for every rule  two atoms and  are joined by an edge . We denote the set of all cycles (loops) in  by . A program  is called tight, if . While we allow programs with loops that might also involve atoms of weight rules, we consider weight rules only as a compact representation of a set of normal rules, similar to the definition of stable models in related work [Bomanson et al. (2016)]. In other words, we do not consider advanced semantics concerning recursive weight rules (recursive aggregates). In case of solvers with different semantics, one can restrict the input language to disregard recursive weight rules, which is also in accordance with the latest ASP-Core-2.03c standard [Calimeri et al. (2015)]. This restriction is motivated by a lack of consensus on the interpretation of recursive weight rules [Ferraris (2011), Faber et al. (2011), Gelfond and Zhang (2014), Pelov et al. (2007), Son and Pontelli (2007)].

2.2 Solving Logic Programs

Let  be a given program, be a rule, and . We define the set  of induced bodies with  in the head by the singleton if is a choice rule, by  if  is a disjunctive rule, and by the union over  for every (subset-minimal) set  of literals such that , if  is a weight rule. This allows us to define , and . A variable assignment is either  or , where variable  is either an atom, or an induced body, or a fresh atom that does not occur in . For a variable assignment , is the complementary variable assignment of , i.e., if and if . An assignment  is a set of variable assignments, where , , and such that . For a set  of literals, we define the induced assignment . A nogood  is an assignment, which is not allowed, where  refers to the empty nogood. Given a set  of nogoods. We define the least fixpoint  of unit propagated nogoods by the fixpoint computation , and  for . Nogood  is a consequence using unit propagation (UP) of set , denoted by , if . An assignment  satisfies a set  of nogoods (written ) if for every , we have . Set  of nogoods is a consequence of a set  of nogoods (denoted by ) if every assignment, which contains a variable assignment for all variables in , that satisfies  also satisfies . The set  of completion nogoods [Clark (1977), Gebser et al. (2012)] is defined by , where

Note that in practice, current ASP solvers do not fully compute . Instead, these solvers partially compute and add relevant nogoods lazily during solving [Alviano et al. (2018)].

Then, if is tight the set  is an answer-set if and only if there is a satisfying assignment  for  [Fages (1994), Gebser et al. (2012)]. The set  of external bodies of program  and set  of atoms are given by  [Gebser et al. (2012)]. We define the loop nogood  for an atom  on a loop  by . For a logic program , the set  is an answer set if and only if there is a satisfying assignment  for , where  [Lin and Zhao (2003), Faber (2005)].

3 ASP-DRUPE: RUP-like Format for Proof Logging

Inspired by RUP-style unsatisfiability proofs in the field of Boolean satisfiability solving [Goldberg and Novikov (2003)], we aim for a proof of inconsistency of a program. Since modern ASP solvers use Clark’s completion [Clark (1977)] to transform a program into a set of nogoods, we do so as well. Our aimed proof then has the following features:

  1. Existence of a simple verification algorithm. In order to increase confidence in the correctness of results, the algorithm that verifies the proof has to be fairly easy to understand and to implement.

  2. Low complexity. The proof is verifiable in polynomial time in its length and the size of the completion nogoods.

  3. Integrability into solvers that employ Conflict-Driven Nogood Learning (CDNL). The proof can stepwise be outputted during solving with minimal impact on the solving algorithm and hence the solver.

The method works as follows: We run the solver on the set  of completion nogoods for given input program . The solver outputs either an answer set or that has no answer set and a proof . We pass together with to the verifier in order to validate whether the solver’s assessment is in accordance with its outputted proof.

3.1 The Proof Format for Logic Programs

The basic idea of clausal proofs for SAT is the following: One starts with the input formula in CNF (given as a set of clauses). Every step of the proof denotes an addition or deletion of a clause to/from the set of clauses. For additions, the condition is to only add clauses that are a logical consequence of the current set of clauses and that it can be checked easily, e.g., use only unit propagation.

For our format ASP-DRUPE, we consider Clark’s completion as the initial set of nogoods corresponding to the input program . Besides addition and deletion of nogoods, we need proof steps that model how the solver excludes unfounded sets (loops).

Example 1

Consider program , which is inconsistent. contains only the positive loop , whose external support is given by the set  of rules, and thus . Set  induces two possible loop nogoods, and .

We describe the proof format ASP-DRUPE for logic programs and adapt the RUP property [Goldberg and Novikov (2003)] to nogoods as follows.

Definition 1 (nogood RUP)

Let be a set of nogoods. Then, a nogood  is RUP (reverse unit propagable) for  if , i.e., we can derive  using only unit propagation.

A proof step is a triple , where denotes the type of the step, is an assignment, and is an atom or . The type  indicates whether the step is an addition (), a completion rule addition (), a completion support addition (), an extension (), a deletion (), or a loop addition (). A proof sequence for a logic program is a finite sequence of proof steps. Initially, a proof sequence gets associated with a set  of nogoods. Note that although the set might be exponential in the size of the program , body definitions for body variables  that do not occur in the proof are never materialized. Then, each proof step  for  subsequently transforms  into the induced set  of nogoods, formally defined below. An ASP-DRUPE derivation is a proof sequence that allows for RUP-like rules for ASP and includes both deletion and extension. In an ASP-DRUPE derivation each step  has to satisfy a condition depending on its type as follows:

  1. An addition  inserts a nogood  that is RUP for .

  2. A completion rule addition  inserts a nogood .

  3. A completion support addition  inserts a nogood  if .

  4. An extension  introduces a definition that renders nogood  equivalent to a fresh atom , i.e., does not appear in . Formally, this rule represents the set of extension nogoods.

  5. A deletion  represents the deletion of from .

  6. A loop addition111There could be an exponential number of external bodies involving weight rules. However, both clasp and wasp treat weight rules differently [Alviano et al. (2015)]. Alternatively, one could easily modify the loop addition type to list also involved external bodies (as in the completion support addition type), which we did not for the sake of readability.  inserts a loop nogood for a loop .

Given an ASP-DRUPE derivation , we define the set  of nogoods induced by step  as the result of applying proof steps  to the initial completion body definitions  for . For our inductive definition in the following, we use multiset semantics for additions and deletions of nogoods, and write for the multiset sum.

Then, we say that an ASP-DRUPE derivation  is an ASP-DRUPE proof for the inconsistency of  if it actually derives inconsistency for , formally, . Note that might be exponential in the input program size, in the worst case. However, there is no need to materialize the set , as, intuitively, this set of body definitions only ensures that every induced body has a reserved auxiliary atom that can be used to “address” the body in a compact way. In an actual implementation of a solver that uses ASP-DRUPE, one needs to specify these used auxiliary atoms anyway, cf. Section 5, where implementational specifications of ASP-DRUPE are described.

Example 2

Consider program  from Example 1 and loop , which induces loop nogood . Then, the proof sequence  is an ASP-DRUPE proof for the inconsistency of with

We show that the proof step is correct, i.e., that is RUP for . To this end, we need to derive from by unit propagation. With the nogood , we derive the unit nogood . With we now get . With these unit nogoods, reduces to .

3.2 Correctness of ASP-DRUPE

Next, we establish soundness and completeness of the ASP-DRUPE format.

Lemma 1 (Invariants)

Let be a logic program and be a finite ASP-DRUPE derivation for program . Moreover, let be the accumulated set of nogoods introduced by the extension rules in for all . Then, the following holds: .

[Proof (Sketch).] We proceed by induction over the length  of the derivation. For the base case, we have . Hence, and the claim holds trivially. For the induction step, we assume that the statement holds for length  and consider step . It remains to do a case distinction for the type:

  1. Deletion with : Immediately, we have . Thus, transitivity of  and the induction hypothesis establishes this case.

  2. Addition with : Since  is RUP for , we know that is a logical consequence of . The remaining steps to draw the conclusion are similar to the deletion step case.

  3. Completion Rule Addition with : Since the resulting nogood  is contained in , the result follows.

  4. Completion Support Addition with : Since the resulting nogood  is contained in , the result follows.

  5. Extension with : According to the induction hypothesis we have . Then, , since is a fresh variable and . As is monotone, and , we know that . It then follows that .

  6. Loop addition with : By definition nogood  is already contained in , which immediately establishes this case.  

Theorem 1 (Soundness and Completeness)

Let be a logic program. Then, is inconsistent if and only if there is an ASP-DRUPE proof for .

[Proof (Sketch).] Let be a logic program. “”: Assume there is an ASP-DRUPE proof of . By definition, there is a finite sequence of proof steps such that and is inconsistent. From Lemma 1, we obtain that is inconsistent. As consists of extension nogoods with disjoint variables, we know that is inconsistent. We conclude from an earlier result [Gebser et al. (2012), Theorem 5.4] that  is inconsistent. “”: Suppose is inconsistent. According to earlier work [Gebser et al. (2012), Theorem 5.4], we know that is inconsistent. RUP is complete [Gelder (2008), Goldberg and Novikov (2003)], which means that for every propositional, unsatisfiable formula  there is a RUP proof for . Hence, we can construct an ASP-DRUPE proof for  as follows: (i) Output all completion rule additions for  and completion support additions for . (ii) Generate loop addition steps for all loops . (iii) Transform  into a propositional formula  by inverting all nogoods. (iv) Construct and use a RUP proof for . Then, output addition rules accordingly, where again all clauses need to be inverted to obtain addition proof steps using nogoods.

Note that in the only-if direction of the proof, one can also use RAT [Wetzler et al. (2014)] proofs without deletion information and afterwards translate RAT steps into extended resolution steps [Kiesl et al. (2018)].

Listing 1 presents the ASP-DRUPE checker, that decides whether a given ASP-DRUPE proof is correct. The input to the checker is both the original program and the proof . To check the proof, we encode into nogoods and then check each statement sequentially.

Input : A logic program and an ASP-DRUPE derivation .
Output : Success if proves that has no answer set, Error otherwise.
1 for  do
2      if  and  then
3            
4       else if  and  then
5            
6       else if  and  then
7            
8       else if  and is a fresh atom w.r.t.  then
9            
10       else if  then
11            
12       else if  then
13            if  then else return  
14       else
15            return
16      
if  then return else return  
Listing 1 ASP-DRUPE-Checker
Lemma 2

For a given logic program  and an ASP-DRUPE derivation , the ASP-DRUPE-Checker runs in time at most .

Corollary 1

Given a logic program  and an ASP-DRUPE derivation . Then, the ASP-DRUPE-Checker is correct, i.e., it outputs Success if and only if  is an ASP-DRUPE proof for the inconsistency of .

3.3 Extension to Optimization

Next, we briefly mention how to verify cost optimization. To this end, an optimization rule is an expression of the form , where  is a literal. Intuitively this indicates that if an assignment satisfies , then this results in costs . Overall, one aims to minimize the total costs, i.e., the goal is to deliver an answer set of minimal total costs. Therefore, if one wants to verify, whether a given answer set candidate is indeed an answer set of minimal costs, we foresee the following extension to ASP-DRUPE, where such a proof consists of the following two parts. (i) An answer set that shows a solution with costs  exists. (ii) An ASP-DRUPE proof that shows that the program restricted to costs  is inconsistent. Note that for disjunctive programs already the first part also needed to contain a second proof showing that indeed there cannot be an unfounded set for the provided answer set. Further, it is not immediate, how this extends to unsatisfiable cores. Hence, so far it only applies to progression based approaches.

4 Integrating ASP-DRUPE Proofs into a Solver

Input : A logic program .
Output : An answer set of or an ASP-DRUPE proof certifying that has no answer set.
1 , , , loop
         // deterministic cons.
2       if  for some  then // conflict
3            if  then
4                  return (INCONSISTENT, )
               // is RUP for
               // add conflict nogood
               // record nogood addition in proof
               // backjumping
5            
6       else if  then // answer set found
7            return (CONSISTENT, )
8       else
               // decision
9            
10      
11
Listing 2 CDNL-ASP-DRUPE: CDNL-ASP [Gebser et al. (2012), page 93] extended by proof logging
Input : A logic program , a set of nogoods, an assignment , and an ASP-DRUPE derivation .
Output : An extended assignment, a set of nogoods, and an ASP-DRUPE derivation (possibly empty).
1 loop
2       repeat
3            if  for some  then // conflict
4                   // record confl. completion nogood return
               // unit-resulting nogoods
5            for  do // record unit completion nogoods in proof
6                  
7            if  then
8                  let  for some  in
9                        
10                  
11            
12      until if  then return if  then  if  then return // no unfounded set let  in
               // add loop nogood
               // record loop in proof
13            
14      
15
Listing 3 NogoodPropagation [Gebser et al. (2012), page 101] adapted for proof logging

In the following, we describe the CDNL-ASP algorithm for logic programs  that we use as a basis for our theoretical model. Afterwards, we describe how proof logging can be integrated. In other words, during the run of an ASP solver, we immediately print the corresponding ASP-DRUPE rules that are needed later for verification in case the ASP solver concludes that the program is inconsistent. A typical CDNL-based ASP solver (cf., Listing 2) relies on unit propagation, since this is a rather simple and efficient way of concluding consequences. Thereby it keeps a set  of nogoods, a current assignment , and a decision level . In a loop it applies NogoodPropagation [Gebser et al. (2012), page 101] consisting of unit propagation and loop propagation (using UnfoundedSet [Gebser et al. (2012), page 104]) whenever suited. Then, if there is some nogood that is not satisfied, either the program is inconsistent (at decision level 0) or ConflictAnalysis [Gebser et al. (2012), page 108] triggers backtracking to an earlier decision level, followed by the learning of a conflict nogood . Otherwise, if all nogoods in  are satisfied and all the variables are assigned, an answer set is found, and otherwise some free variable is selected (Select).

Listings 2 and 3 contain a prototypical CDNL-based ASP solver that is extended by proof logging, where the changes for proof logging are highlighted in red. We use the element operator () to determine whether an element is in a sequence, and denote the concatenation of two proofs by the operator as follows: . The idea is to start with an empty ASP-DRUPE derivation. Whenever a new nogood, or loop nogood is learned and added to  accordingly, this results in an added addition or loop addition proof step, respectively. Note that in Listing 3 we add completion rule addition steps and completion support addition steps, whenever unit propagation (or conflicts) involve rules in  or , respectively. In particular, Lines 3 and 3 take care of adding involved parts of the completion to the proof (if needed) accordingly. At the end, when the ASP solver concludes inconsistency, the proof is returned including the empty nogood as last nogood. Note that advanced techniques (see, e.g., [Gebser et al. (2012)]) like forgetting of learned clauses and restarting of the ASP solver can also be implemented using deletion rules with ASP-DRUPE. As it turns out, preprocessing in ASP is less sophisticated as for SAT. In the literature, CDNL-based ASP solvers often use preprocessing techniques [Gebser et al. (2008)] similar to SAT solvers, i.e., SatElite-like [Eén and Biere (2005)] operations as variable and nogood elimination. For simple preprocessing operations restricted to variable and nogood elimination ASP-DRUPE suffices. Note that if Clark’s completion is exponential in the program size due to weight rules, also propagators [Alviano et al. (2018)] are supported. For details we refer to the implementation in Section 4.1.

Example 3 (Cdnl-Asp-Drupe)

We continue the previous Example 2 and indicate a possible CDNL-ASP-DRUPE run on that leads to the exemplary ASP-DRUPE proof given above. We use the notation ( to indicate that was assigned true (false) at decision level .

  1. Initially, nothing can be propagated.

  2. After the decision , unit propagation derives only