Testing Equivalence to the Hamiltonian Cycle Polynomial

Abstract: The Hamiltonian Cycle polynomial, denoted as $HC_n$, is defined to be the sum of the weighted Hamiltonian Cycles in an $n$-vertex complete digraph, with vertices labeled $1$ to $n$ and edges weighted by formal variables $x_{i,j}$. Valiant (STOC 1979) studied the Permanent and $HC$, defined as the family ${HC_n | \ n \geq 1}$, and showed both families are VNP-complete, the former over any field of characteristic other than $2$, and the latter over any field. Since its introduction, $HC$ has been studied from the perspective of lower bounds by Jerrum-Snir (JACM 1982), determinantal complexity by Huttenhain-Ikenmeyer (LAA 2016), and its relation to the Permanent by Goulden-Jackson (EJC 1981) and Grochow (ToC 2017). Its VNP-completeness over any field has been used in Malod (CCC 2007), Grochow-Mulmuley-Qiao (ICALP 2016) and Hrubes (ToCT, 2016). The Equivalence Testing problem for a polynomial $f(\mathbf{x})$ (ET for $f$) is as follows: Given $g(\mathbf{x}) \in \mathbb{F}[\mathbf{x}]$ as a black box, decide if there exists $A \in \mathrm{GL}_{|\mathbf{x}|}(\mathbb{F})$ such that $g = f(A\mathbf{x})$. Kayal (STOC 2012) gave a randomised polynomial time ET algorithm for the Permanent. In this work, we give a randomised polynomial time ET algorithm for $HC$ with mild constraints on the field. We show that, like the Permanent polynomial, the symmetries of $HC_n$ are generated by permutation and scaling matrices over large enough fields. We also show that $HC_n$ is not characterised by its symmetries, unlike the Permanent polynomial, Mulmuley-Sohoni (SIAM J. Computing, 2001). Nevertheless, like the Permanent polynomial, $HC_n$ is downward self-reducible, Zhang-Bai (TCS 2011), implying $HC_n$ is characterised by circuit identities and an efficient algorithm to test if a given circuit $\mathrm{C}$ computes $HC_n$. We also get a Flip theorem for $HC_n$ as a result of its circuit identities.

• The paper presents a randomized polynomial-time algorithm that tests equivalence to the Hamiltonian Cycle polynomial by leveraging invertible linear transformations.
• The methodology combines Lie algebraic analysis, symmetry structure characterization, and combinatorial techniques to overcome non-symmetry characterizations.
• The results imply efficient black-box circuit testing for HCₙ and highlight distinctions between Hamiltonian Cycle and Permanent complexity.

Detailed Summary of "Testing Equivalence to the Hamiltonian Cycle Polynomial" (2606.26653)

Introduction and Problem Statement

This work studies the equivalence testing (ET) problem for the Hamiltonian Cycle polynomial family ($HC_n$) over a field $F$. The $HC_n$ polynomial, defined on the adjacency matrix of an $n$-vertex complete digraph with formal variables $x_{i,j}$ as edge weights, encodes as its monomials the Hamiltonian cycles of the graph. It is VNP-complete over every field, in contrast to the Permanent ($\text{Perm}_n$), which is VNP-complete only for fields of characteristic $\neq 2$. ET asks: given black-box access to an $(n^2-n)$-variate homogeneous degree-$n$ polynomial $f$ over $F$0, does there exist an invertible linear transformation $F$1 (and optionally a shift vector) such that $F$2? This is strictly harder (in terms of generality) than simply testing if $F$3 computes $F$4 up to a permutation or scaling, since it considers arbitrary invertible linear changes of variables.

This problem is a natural analogue of previously studied ET problems for other polynomial families, especially the Permanent, for which efficient randomized ET algorithms are known via the use of symmetries and Lie algebraic techniques [Kayal12]. The $F$5 case is particularly interesting both for its universality (VNP-complete over all fields) and for deep differences: while zero-testing for the Permanent corresponds to perfect matching (in P), for $F$6 it is NP-complete (Hamiltonicity).

Main Results

The central contribution is a randomized polynomial-time algorithm for equivalence testing to $F$7 over large enough fields (roughly, size at least $F$8 and characteristic $F$9 or $HC_n$0), generalizing the comparison to the Permanent and extending the ET toolkit to a new, fundamental VNP-complete family. The approach is technical and builds out a full structural analysis of $HC_n$1's symmetries, Lie algebra, downward self-reducibility, and the relationships among these objects.

Highlights:

• Randomized Polynomial-Time ET Algorithm: There exists a randomized polynomial-time algorithm that, given black-box access to a candidate polynomial, either finds an invertible linear transformation to $HC_n$2 or determines that none exists. This covers both the $HC_n$3 and more general $HC_n$4 cases.
• Symmetry Structure: For large enough fields, all symmetries of $HC_n$5 are generated by permutation and scaling matrices, analogously to the Permanent.
• Lie Algebra Characterization: The full Lie algebra $HC_n$6 is characterized, and explicit bases are constructed for all $HC_n$7; the $HC_n$8 case is fully resolved by field characteristic.
• Non-Characterization by Symmetries: For all $HC_n$9 and large enough fields, $n$0 is not characterized by its symmetries (unlike the Permanent): there exist non-scalar multiples with the same symmetry group.
• Circuit Testing and Flip Theorem: Despite lack of symmetry characterization, $n$1 polynomials are characterized by circuit identities (via downward self-reducibility), enabling efficient black-box circuit testing for $n$2 and implying a Flip theorem analogous to prior work for the Permanent.

Algorithmic Approach and Technical Contributions

The algorithm closely follows the Lie algebraic approach successfully used for the Permanent [Kayal12], but requires several novel ingredients due to key structural differences:

1. Lie Algebra and Symmetry Group Analysis: A detailed investigation demonstrates that the Lie algebra consists solely of diagonal matrices satisfying specific combinatorial sum conditions. Over the base field, permutation and continuous (scaling) symmetries suffice to generate the full symmetry group.
2. Efficient Basis Construction: Explicit construction of a $n$3-dimensional basis for $n$4 is provided for all relevant $n$5, adapting ideas used for the Nisan-Wigderson polynomial in [GuptaS19], but generalizing beyond the field characteristic restrictions present in prior work.
3. Reduction to Permutation-Scaling Equivalence: The first major step is to efficiently reduce general ET to the problem of permutation and scaling equivalence using the Lie algebra structure and random diagonalization.
4. Combinatorial Recovery of Permutation and Scaling: The recovery of the correct permutation and scaling relies on a careful analysis of the vanishing pattern of second derivatives of $n$6, which differ materially from the Permanent (notably, monomials differ in $n$7 variables, complicating pairwise comparisons).
5. Verification via Downward Self-Reducibility: Since $n$8 is not characterized by symmetries, the final verification step relies on downward self-reducibility—a property that allows efficient checking of whether a candidate circuit actually computes $n$9.

Structural Differences vs the Permanent

A detailed table in the paper highlights several crucial contrasts:

• Symmetries: Permanent is characterized by its symmetries over almost all fields, $x_{i,j}$0 is not when $x_{i,j}$1.
• Zero Testing: Permanent testing is in P (perfect matching); $x_{i,j}$2 zero-testing is NP-complete (Hamiltonicity).
• Self-reducibility: Both polynomials possess downward self-reducibility, which is essential for circuit identity-based characterization.
• Circuit Testability & Flip Theorem: Both admit efficient black-box circuit testing and flip theorems.
• Lie Algebra Basis: Explicit, efficiently computable over all relevant fields for both.

Implications and Theoretical Impact

This work significantly extends the range of polynomial families for which efficient equivalence tests are possible, broadening the practical and theoretical reach of ET techniques and further exposing structure in VNP-complete polynomials. It shows that Lie-theoretic and symmetry-based tools, augmented with combinatorial and circuit-theoretic analysis, can resolve ET even in settings where monomial structure and symmetry characterization are much more subtle than in the Permanent case.

The failure of symmetry characterization for $x_{i,j}$3 is notable, as it is rare among prominent algebraic complexity polynomials and highlights the necessity of additional testing tools beyond symmetries.

Practically, the algorithm enables succinct canonical representations (modulo invertible transforms) and efficient checking of polynomial identity relative to $x_{i,j}$4, a tool of interest in computational invariant theory, symbolic computation, and complexity-theoretic reductions.

Theoretically, this places $x_{i,j}$5 on similar footing as the Permanent with respect to randomized ET, and prompts further questions:

• Can such ET algorithms be derandomized over natural fields?
• Are there natural VNP-complete polynomial families for which ET is provably hard?
• What other polynomial families lack symmetry characterization but admit circuit-identity characterizations suitable for ET?
• How do these results interact with recent progress on lower bounds, monotone projections, and arithmetic circuit complexity for $x_{i,j}$6?

Conclusion

This work establishes that ET for the Hamiltonian Cycle polynomial $x_{i,j}$7 is efficiently solvable over large enough fields, leveraging a comprehensive structural analysis that combines representation-theoretic, combinatorial, and circuit-theoretic insights. The results deepen the understanding of VNP-complete polynomials and the power and limitations of symmetry- and circuit-based classification tools in algebraic complexity. The approach and techniques open the door for both further efficient algorithms for equivalence and new complexity-theoretic separations among natural polynomial families.