Exact Separation of Words via Trace Geometry

Abstract: A basic question in the theory of two-state measure-once quantum finite automata (MO-QFAs) is whether two distinct input words can be separated with certainty. In the setting considered here, this exact separation problem reduces to a trace-vanishing question in (SU(2)): given distinct positive words (u) and (v), find matrices (A,B\in SU(2)) such that the evaluated trace of (u^{-1}v) is zero. The central difficulty lies in the genuinely nonabelian regime where (u) and (v) have the same abelianization, so the obvious commutative information disappears and the fine structure of the word must be connected to the geometry of representations. This paper develops a slice-driven framework for that task and proves exact separation for every hard positive-word difference covered by four explicit certified conditions, thereby reducing the problem to a sharply delimited residual super-degenerate class. The method extracts algebraic data from the positive-word difference and uses them to select explicit low-dimensional families in (SU(2)^2) on which the trace becomes computable. On the algebraic side, the metabelian polynomial is decomposed into explicit interval blocks determined by prefix statistics, and a suitable slope specialization preserves nontrivial information. On the analytic side, the paper derives a computable quadratic trace identity on a visible one-parameter family and complements it with a Laurent-matrix sum-of-squares identity in a parallel algebraic model. These certified criteria are already strong in numerical experiments. This paper also shows that no method based only on finitely many finite-image tests can be universal.

• The paper presents a slice-driven framework for exact word separation in two-state MO-QFAs using trace geometry.
• It uses Fox calculus and combinatorial invariants to derive explicit slice families that force vanishing trace in SU(2).
• Numerical evaluations on 50,000 instances validate the method, though a super-degenerate residual class remains for future study.

Exact Separation of Words via Trace Geometry: Summary and Analysis

Problem Formulation and Context

The paper addresses the exact separation problem for measure-once quantum finite automata (MO-QFAs) with two states: given distinct input words $u$ and $v$, can one construct MO-QFA accepting $u$ and rejecting $v$ with certainty? Specifically, in the case where $u$ and $v$ have identical counts of letters (i.e., equal abelianization), the separation problem reduces to finding matrices $A,B \in SU(2)$ such that the trace of $u^{-1}v$ vanishes. This core problem exposes the nonabelian structure by focusing on positive-word differences in $F_2'$, the commutator subgroup of the free group $F_2$.

Prior work established universal separation in nondeterministic modes, while the zero-error question for two-state MO-QFAs remained open, particularly in the nonabelian “equal-abelianization” regime. The present study advances this frontier, relating automata-theoretic questions to concrete word map geometry in compact Lie groups.

Methodological Innovations

The main technical innovation is a slice-driven framework for the trace-vanishing problem. Rather than random enumeration over $v$0, the method leverages combinatorial invariants extracted from word structure to choose low-dimensional family (slice) of pairs $v$1 where the trace can be evaluated or forced to zero. On the algebraic side, the metabelian polynomial is computed using Fox calculus and decomposed into explicit interval blocks via prefix statistics, providing combinatorial coordinates for representation space. A slope specialization (substituting $v$2 for an integer $v$3) retains essential information and yields a one-variable polynomial.

Analytically, the paper places slope-specialized words onto an explicit one-parameter family in $v$4, exploiting a quadratic trace identity: the second-order trace deficit becomes the squared modulus of the specialized Fox polynomial. A complementary Laurent-matrix sum-of-squares identity further links the trace deficit to algebraic expressions, facilitating concrete witness constructions.

Explicit Witness Menu and Numerical Evaluation

Multiple certified slice families are constructed for witness generation:

• Dihedral slice: Alternating row-prefix invariant $v$5 allows explicit evaluation leading to trace-zero for nonzero $v$6.
• Quaternionic slice: Specialization at $v$7 yields a cosine trace formula; zero-trace is attained if $v$8 is nonzero for some slope $v$9.
• Mixed normalizer slice: Signed $u$0-count ($u$1) criteria yield trace-zero witnesses when $u$2.
• Local regime ($u$3): Classification places the difference word as conjugate to a commutator, permitting explicit witness constructions.
• Character coordinate tests: Evaluations of the trace polynomial at selected points in the $u$4 character variety region provide additional explicit criteria.

Numerical tests on $u$5 random hard pairs with $u$6 produced no misses for the combination of explicit slice families and interior-point criteria, indicating that these methods cover the overwhelming majority of hard instances.

Theoretical Implications and Limitations

The finite-menu approach demonstrates that combinatorial-algebraic slice families suffice for exact separation in nearly all cases, substantially reducing the residual class requiring further analysis. Importantly, the study proves that universal finite-image testing (i.e., using only a fixed finite list of finite subgroup evaluations) cannot distinguish all hard pairs; explicit counterexamples are provided, and surjectivity constraints are rigorously established.

Nevertheless, a “super-degenerate” residual class remains, defined by the simultaneous vanishing or inactivity of all current slice invariants and interior-point tests. The completion target articulated in Conjecture 5.6 is to develop finitely many explicit analytic slice families, each with compact, connected parameter spaces and a commuting basepoint, so that for any hard positive-word difference, at least one slice in the menu yields a trace-zero witness.

Future Directions

• Slice Family Expansion: The principal challenge is to construct analytic slice families capable of detecting the residual super-degenerate class. Harmonic-analytic or topological tools (e.g., Peter-Weyl theory, transfer matrices) may provide new forcing principles for trace-zero.
• Global Energy and Correlation Analysis: Recasting the trace-zero problem in the Fourier domain or via Toeplitz/Hardy-type operators could reveal global phase cancellation phenomena, complementing the algebraic slice approach.
• Representation Theory Integration: Leveraging character orthogonality and global matrix methods on $u$7 has potential to synthesize combinatorial and geometric insights for universal detection.

Conclusion

This paper establishes a robust algebraic-analytic framework for exact word separation in two-state MO-QFAs. By bridging combinatorial group theory, Fox calculus, and trace geometry on $u$8, it demonstrates that explicit slice-driven methods almost universally solve the separation problem for positive-word differences. The theoretical boundary is sharply defined: finite-image subgroup tests cannot provide a universal method, and the completion target is a finite menu of analytic slices guided by geometric or harmonic analysis. The practical impact is immediate—most hard pairs are exactly separated by current explicit tests—while the residual problem delineates precise future research avenues for quantum automata and nonabelian word map geometry.

Reference: "Exact Separation of Words via Trace Geometry" (2603.29411).