A TQFT-based Platform for Efficient Computation of Knot Invariants

Technical Summary: A TQFT-based Platform for Efficient Computation of Knot Invariants

Problem Statement
The central challenge in knot theory is determining when two closed curves in three-dimensional space represent the same knot. While knot invariants provide a tractable approach to this classification problem, the most informative invariants—specifically colored HOMFLY–PT polynomials and higher-rank Chern–Simons invariants—are computationally expensive. Direct evaluation is hindered by the complexity of path integrals over gauge fields and the rapid growth of computational cost associated with higher representations and increasing crossing numbers. Existing methods often struggle to balance theoretical depth with practical accessibility, creating a gap between the strength of colored invariants and their utility for researchers.

Methodology
The authors introduce the TQFT Knot Explorer, an interactive web platform that unifies the diagrammatic construction of knots, tensor network evaluation, and invariant computation. The methodology focuses on arborescent (double-fat) knots, which can be described using Feynman Ribbon Diagrams (FRDs).

1. Diagrammatic Representation (FRDs):

   • Knots are modeled as tree-structured diagrams composed of vertices, fingers (legs), and propagators.
   • Fingers carry integer-valued braiding parameters (twists), while propagators connect vertices and record the twist between them.
   • The platform supports three modes of construction: Pretzel diagrams, two-vertex trees, and a free-form "Build your own" editor.
   • A key structural finding is the two-vertex FRD ($\Gamma(A_1; p_{12}; A_2)$), which consists of two vertices connected by a propagator, each with two fingers.

2. Topological Quantum Field Theory (TQFT) Mapping:

   • The FRD components are mapped to building blocks of the $SU(N)$ Wess–Zumino–Novikov–Witten (WZNW) conformal field theory.
   • Vertices correspond to amplitudes in the space of conformal blocks ($H^{\otimes n}$).
   • Fingers encode local braiding data via fusion and braiding operators.
   • Propagators transfer information between vertices.
   • The platform constructs a tensor network where the contraction of these tensors yields the knot invariant. The amplitude $A_R(\Gamma)$ is proportional to the expectation value of a Wilson loop $W_R[K]$ colored by representation $R$.

3. Algorithmic Implementation:

   • The system utilizes a hybrid C++/Python pipeline.
   • C++ (GiNaC): Handles symbolic evaluation of TQFT building blocks, including Racah matrices and braiding eigenvalues.
   • Python: Manages parameter enumeration, normalization, and database comparison.
   • Efficiency Optimization: The algorithm exploits the recursive structure of FRDs. For the two-vertex case, it employs an "independent-index generalization" where the bridge factor is replaced by a generalized propagator, allowing for efficient summation over internal states.
   • Search Procedure: The system enumerates parameter configurations, computes the resulting Laurent polynomials, and hashes them against curated databases (including colored Jones and Adjoint polynomials) to identify knots up to 13 crossings.

Key Contributions

• Unified Workflow: This is the first platform to combine diagrammatic construction, tensor-network evaluation, and knot identification in a single visual environment.
• The Two-Vertex FRD Completeness: The authors demonstrate that the evolution family of two-vertex FRDs is complete for all FRD-like knots with up to 10 crossings. Furthermore, this compact structure effectively identifies knots with higher crossing numbers (up to 13 and beyond), suggesting that minimal FRD architectures can encode rich topological information.
• Computational Efficiency: Benchmarks against the KnotTheory colored Jones algorithm and the braid-walk method show that the proposed TQFT-based approach is faster in most tested cases. The recursive nature of the algorithm reduces unnecessary overhead, making it scalable for complex knot families.
• Accessibility: The platform converts technically demanding calculations into intuitive, reproducible diagrammatics, lowering the barrier to studying higher-rank Chern–Simons invariants.

Results

• Performance: Running time analysis (Figures 5–8) confirms that the method maintains stable and efficient execution times as structural complexity increases, outperforming existing general-purpose algorithms in the tested range (up to 8 crossings, representations $r=1$ to $10$).
• Identification: The system successfully identifies knots by matching computed invariants against a database. It handles mirror images by comparing coefficient sequences and their reversals.
• Classification: The two-vertex FRD structure provides a systematic classification for FRD-like knots up to 10 crossings, a result not previously presented in this unified form.

Significance and Claims
The paper positions the TQFT Knot Explorer as a significant step toward a unified computational framework for quantum knot invariants. Its primary significance lies in:

1. Bridging Theory and Practice: It makes high-rank Chern–Simons invariants accessible by translating abstract algebraic constructions into visual, interactive workflows.
2. Scalability: It offers a practical framework for exploring colored Jones invariants beyond the range typically accessible via general algorithms.
3. Foundation for Future Research: The authors note that the framework serves as a foundation for emerging research directions, including the knot–quiver correspondence, machine learning methods for knot invariants, and quantum computing applications.

The authors remain modest regarding the current scope, noting that the 13-crossing limitation is due to the current database size rather than the computational engine's capabilities. Future updates are planned to support higher crossing numbers and $SU(N)$ gauge groups. The work explicitly acknowledges that while the invariants are strong, they are not complete (e.g., they may not distinguish all mutant knots), and the identification results are presented as candidate sets rather than absolute proofs of identity in ambiguous cases.