Conservative Maltsev Constraint Satisfaction Problems

One of the central open problems to classify the computational complexity of finite-domain constraint satisfaction problems within P is to prove better algorithmic results for CSPs with a Maltsev polymorphism; we do not even know whether these CSPs are in NC. Relatedly, the descriptive complexity of these problems is open as well. An important special case, previously studied by Carbonell from the perspective of uniform polynomial time-algorithms, are CSPs with a conservative Maltsev polymorphism. We show that for every finite structure B with a conservative Maltsev polymorphism, the CSP for B can be solved by a symmetric linear Z2-Datalog program, and in particular is in the complexity class parity-L. Previously, the best known algorithms just showed containment in P. In our proof we develop a structure theory for conservative Maltsev algebras which might be of independent interest.

💡 Research Summary

The paper addresses a long‑standing open problem in the fine‑grained complexity classification of finite‑domain constraint satisfaction problems (CSPs) that admit a Maltsev polymorphism. While such CSPs are known to be solvable in polynomial time (e.g., via the Bulatov‑Dalmau algorithm), it is unknown whether they belong to any smaller parallel complexity class such as NC, nor is their descriptive complexity fully understood. The authors focus on the subclass of conservative Maltese CSPs, where each variable may be restricted to an arbitrary subset of the domain (list‑homomorphism setting). Carbonell previously gave a uniform polynomial‑time algorithm for this subclass, but it relied on arc‑consistency, which is P‑complete, and therefore did not shed light on the true complexity class.

The main contribution is a complete classification of conservative Maltsev CSPs within the log‑space hierarchy: every such problem is either in deterministic log‑space (L) or is ⊕L‑complete (the class of problems solvable by a log‑space nondeterministic Turing machine that accepts on an even number of computation paths). Consequently, all conservative Maltsev CSPs lie in the class ⊕L, a strict subclass of P that is believed to be far from P‑complete.

To achieve this, the authors develop a multi‑layered theory:

1. Algebraic Reduction to Conservative Minority Operations.
   Using a lemma of Carbonell, they show that any clone containing a conservative Maltsev operation also contains a conservative minority operation. Hence, the study can be restricted to algebras equipped with a single conservative minority term (called conservative minority algebras).

2. Tree‑Based Structure Theory for Conservative Minority Algebras.
   They analyze congruences (homomorphic kernels) of such algebras and prove a strong “block decomposition” property: any congruence can be expressed as a collection of independent blocks, and there exists a unique maximal congruence. By iteratively factoring out maximal congruences, every finite conservative minority algebra can be represented as a rooted tree whose nodes are decorated with local simple algebras. This representation (the conservative minority tree) simultaneously encodes the algebraic operations and the hierarchy of its congruences (Theorem 24).

3. Primitive‑Positive (pp) Construction of General Templates.
   The authors show that any finite structure with a conservative Maltsev polymorphism can be pp‑defined from a small, well‑understood family of structures denoted (P_{n,k}). These basic structures have at most ternary relations and possess a very restricted tree representation where the underlying trees are essentially paths decorated by hereditarily simple algebras (Section 4.5). Section 5 proves that the whole class of conservative Maltsev templates can be built from the (P_{n,k}) structures using a sequence of mutations and iterative constructions.

4. Algorithmic Core – Solving (P_{n,k}) CSPs.
   For each basic template (P_{n,k}) the authors design a deterministic log‑space procedure called Solve. The procedure reduces an instance of (\operatorname{CSP}(P_{n,k+1})) to an instance of (\operatorname{CSP}(P_{n,k})) while preserving satisfiability, eventually reaching a trivial base case. The reduction crucially requires solving systems of linear equations over the two‑element field (\mathbb{Z}_2).

5. Cyclic‑Group Datalog and Symmetric Linear (\mathbb{Z}_2)‑Datalog.
   To capture the linear‑equation solving step within a logical framework, the authors extend Datalog with a cyclic‑group operator that can test feasibility of (\mathbb{Z}_2) linear systems. The resulting language, called Symmetric Linear (\mathbb{Z}_2)‑Datalog, is shown to be evaluable in ⊕L (Section 7.3). Moreover, primitive‑positive constructions preserve expressibility in this language, so the whole algorithm for any conservative Maltsev CSP can be expressed as a symmetric linear (\mathbb{Z}_2)‑Datalog program.

6. Complexity Classification.
   By combining the logical expressibility result with known characterisations of ⊕L (problems whose accepting computation paths have even parity), the authors prove that every conservative Maltsev CSP is either in L or ⊕L‑complete (Theorem 118). This yields a full classification up to log‑space reductions, improving the previous state of knowledge which only placed these problems in P.

The paper concludes by highlighting the broader significance of the tree‑based structure theory for conservative minority algebras. Although developed for the conservative Maltsev setting, the techniques may extend to general Maltsev CSPs, potentially leading to new insights about their placement within NC or other sub‑P classes. The work also bridges universal algebra, descriptive complexity, and algorithm design, offering a concrete example where algebraic decomposition directly informs the design of a low‑complexity logical program.