The proposal sends very heavy quarks into a strongly coupled three-dimensional gauge theory and treats them as probes. In two space dimensions, two heavy particles orbiting each other trace out braided worldlines in spacetime. If the theory's low-energy limit is topological, meaning only how the worldlines are knotted matters and not their exact shape, then the long-distance part of the heavy pair's physics is a knot invariant. Specifically, it is a colored invariant of the torus link the orbit traces. The standard way of splitting heavy-quark physics into a short-distance piece and a long-distance piece, NRQCD (non-relativistic QCD) factorization, then predicts that the long-distance piece is exactly computable. It also predicts that this piece is different in each proposed vacuum of the theory.
This matters for non-supersymmetric Seiberg duality. That is the conjecture that two different gauge theories, an "electric" one and a "magnetic" one, describe the same low-energy physics even without supersymmetry. In three dimensions it currently rests on one-loop effective potentials and string-theory intuition, and its authors state that conclusive evidence for the key ingredient, condensation of magnetic scalar quarks, is missing. Heavy-quark knot invariants give a sharp, falsifiable fingerprint of each conjectured phase. We would compute that fingerprint on both sides of the duality and check it against a weak-coupling calculation where the electric theory is under control.
Core Claim
Phases of QCD3 from non-SUSY Seiberg duality and brane dynamics (also; Phases of QCD 3 from non-SUSY Seiberg duality and brane dynamics; Phases of QCD 3 from non-SUSY Seiberg duality and brane dynamics) studies $USp(2N)$ Yang–Mills–Chern–Simons theory with $N_f$ fundamental flavors. A Chern–Simons term at level $k$ is a topological addition to the gauge action that gives the gluons a mass. Using a non-supersymmetric brane construction and a magnetic dual, the paper proposes four regions of parameter space:
- Region I: magnetic squarks condense, and the low-energy theory is a bosonic description.
- Regions II and II′: squark condensation Higgses the magnetic group and breaks flavor symmetry.
- Region III: the low-energy theory is an interacting conformal field theory.
Every gapped region ends at low energies in a definite Chern–Simons topological field theory (TQFT), whose gauge group and level are fixed by which magnetic squarks condensed. Region III ends in no TQFT at all.
Our claim is that a heavy quark of mass $m_Q$, far above every dynamical scale, acts as a Wilson line in the representation $R$. A Wilson line is the phase picked up by an infinitely heavy color charge along its path. A heavy quark–antiquark pair produced in color fusion channel $a$ (the 3D analogue of NRQCD's color-singlet and color-octet states) then obeys a factorized rate
$$\Gamma_a = C_a(m_Q,\alpha)\;\frac{d_a}{d_R\, d_{\bar R}}\;\big[1+O(m_Q v^2/M_{\rm gap})\big].$$
The symbols are as follows:
- $C_a$ is the perturbative short-distance coefficient. It is identical in electric and magnetic language because it is set at the scale $m_Q$.
- $d_a$ is the quantum dimension of channel $a$ in the low-energy TQFT. The ratio $d_a/(d_R d_{\bar R})$ plays the role of NRQCD's long-distance matrix element (LDME).
- $M_{\rm gap}$ is the mass gap of the light sector.
The correction factor follows from the usual NRQCD scale ordering $m_Q \gg m_Q v \gg m_Q v^2$. It requires the bound-state energy $m_Q v^2$ to lie below $M_{\rm gap}$ so that soft dynamics is purely topological.
A bound state that orbits $n$ half-turns traces the torus link $T(2,n)$. Its phase is the colored invariant of that link, so its angular momentum is shifted by the fractional amount $h_a - h_R - h_{\bar R}$, where $h$ denotes topological spin. In this toy setting the "LDMEs" are universal by construction and exactly calculable. They are also phase-dependent, so measuring them in one description and predicting them in the other is a test of the duality.
This is a different use of NRQCD from its home in four-dimensional collider physics. We do not claim to resolve the 4D LDME universality crisis:
- LDMEs fitted to Tevatron hadroproduction overpredict HERA photoproduction (Color-Octet Contributions to J/ψ Photoproduction).
- Next-to-leading-order color-singlet theory undershoots CMS and ATLAS $J/\psi$-pair data by factors of 10–40 while fitting LHCb at low transverse momentum (NLO corrections to double prompt J/ψ hadroproduction).
Four-dimensional QCD has no topological IR phase, so our matrix elements do not exist there. Those papers motivate the question of when long-distance matrix elements can be universal. Our 3D model answers it in one controlled corner, where universality holds exactly and fails only through the stated $m_Q v^2/M_{\rm gap}$ correction.
Relation to Prior Work
Three groups of papers play distinct roles.
The duality being tested. The Armoni–Niarchos phase analysis above supplies the conjecture we put under test. The 4D orientifold dualities of Nonsupersymmetric Seiberg duality, orientifold QCD, and noncritical strings supply the large-$N$ anchor. Those dualities are exact in the planar limit and conjectural at finite $N$. Our test is designed for finite $N$, where the open problem lies. Defects in Chern-Simons theory, gauged WZW models on the brane, and level-rank duality realizes level-rank duality with branes. We reuse level-rank duality to translate Wilson-line data between descriptions. Level-rank duality is the exact equivalence between Chern–Simons theories with group rank and level exchanged.
The knot invariants. Knots-quivers correspondence and its review Knot-quiver correspondence: a brief review give the colored invariants of torus knots and an integrality structure, Labastida–Mariño–Ooguri–Vafa (LMOV) integrality. That structure is proved only for symmetric representations of $U(N)$. We use it as an independent computational cross-check on a $U(N)$ analogue of our setup. The $USp$ case requires Kauffman-type invariants, which the quiver correspondence does not cover; that is a constraint we work around rather than ignore.
Topology meeting dualities. Multi-cover skeins, quivers, and 3d N = 2 dualities turns knot skein relations into supersymmetric 3D dualities. We do the reverse with less machinery: we use knot invariants to test a non-supersymmetric duality, and we need no supersymmetry because the topology lives in the low-energy TQFT rather than in protected BPS states. Brane dynamics and 3D Seiberg duality on the domain walls of 4D N = 1 SYM shows such 3D theories arise on domain walls of a supersymmetric 4D bulk. It does not probe them with heavy charges.
Research Program
Mapping each region to its low-energy TQFT. For each point on a grid in $(N, k, N_f)$ spanning Regions I, II, II′ and III, plus the boundary at $k=0$, we determine the IR Chern–Simons theory predicted by the magnetic description after squark condensation. We map the electric heavy-quark Wilson line to its magnetic image using level-rank duality, and confirm the map respects the one-form-symmetry and flavor 't Hooft anomalies already matched in the source paper.
Computing the fingerprint. From the modular data of each TQFT we compute the topological spins $h_a$, the quantum dimensions $d_a$, and the colored invariants of $T(2,n)$ for $n=3,5,7,9$. In knot-table notation these are the trefoil and the knots $5_1$, $7_1$, $9_1$, which are benchmark knots in the knot-quiver literature. The figure-eight knot serves as a parity control. It is amphichiral (identical to its mirror image), so its invariant must be unchanged under $k\to-k$, the time-reversal operation. That makes it a check on sign conventions that carries no phase information.
Computing the electric side independently. At large $|k|$ the electric theory is weakly coupled, with effective coupling of order $N/k$. In that regime the heavy-pair Wilson-line network can be computed perturbatively, including the one-loop shift of the level from integrating out massive matter. This calculation does not use the duality at all.
Characterizing the conformal region. In Region III the duality predicts no TQFT. The heavy-pair observables should then show conformal dependence on separation instead of pure topological phases. We will compute the leading falloff in the magnetic description at weak magnetic coupling.
Baselines
The naive reference point is the existing evidence base: 't Hooft anomaly matching plus the one-loop Coleman–Weinberg potential (the quantum correction to the scalar potential from loops of fields), as carried out in Phases of QCD3 from non-SUSY Seiberg duality and brane dynamics. We reproduce it on the same parameter grid. We beat it if heavy-probe invariants distinguish at least one pair of anomaly-consistent candidate phases that the potential analysis cannot separate. No reported quantitative result found on Lacuna for a phase-discrimination rate or any numerical confidence measure from that analysis, so the comparison is qualitative: we count discriminated cases.
The strongest existing method is direct level-rank matching of the IR Chern–Simons theories alone, using the brane picture of Defects in Chern-Simons theory, gauged WZW models on the brane, and level-rank duality. We run it over the same grid. Our proposal beats it only where heavy-probe data add a constraint: the Region III versus gapped boundary, the $k=0$ edge, and the large-$|k|$ overlap where the independent electric calculation applies.
Ablation of the key ingredient. We repeat the analysis using only the color-singlet channel $a=0$. In any TQFT that channel gives $d_0/(d_R d_{\bar R})$ with trivial spin shift, and it is blind to which phase is realized. If the singlet-only analysis discriminates phases as well as the full analysis, the non-singlet channels are not doing the work and the central claim fails.
Calibration of the knot computations. For the $U(N)$ analogue, colored invariants obtained from knot-quiver generating functions (Knots-quivers correspondence) must agree exactly with our modular-data computation on the same torus knots. The paper derives explicit formulas for the $(3,7)$ torus knot; no reported numerical benchmark beyond this found on Lacuna.
Why not just…?
Why not just do level-rank duality on the IR theories and skip the heavy quarks? This is the real threat. In Regions I, II and II′, once the IR TQFT is assumed, heavy-probe data are fixed by it, and part of our work reduces to bookkeeping. The added content lies elsewhere. Level-rank matching assumes which TQFT appears, while the heavy-pair rate and spins can be computed independently from the electric side at large $|k|$, so the TQFT assignment itself is tested rather than assumed. Heavy probes also separate gapped from conformal behavior, which no TQFT matching can do. Our first experiment quantifies the threat by counting the grid points where anomaly matching leaves more than one IR candidate with different torus-link invariants. If that count is zero, the topological part of the proposal adds little, and we will say so.
Why not simulate the theory on a lattice? A Chern–Simons term at nonzero $k$ makes the Euclidean action complex. That causes a severe sign problem: Monte Carlo weights are no longer positive, so standard sampling breaks down. At $k=0$, the boundary the source paper flags as anomalous, the sign problem is milder. A static-quark lattice study there would be valuable, but the file reports no lattice result for these theories. We treat it as a follow-up, not a competitor.
Why not test the factorization directly against quarkonium data? The CDF, HERA, LHCb, CMS and ATLAS data in the file probe 4D QCD, which has no topological IR phase, so our topological matrix elements have no counterpart there. Fitting them would be meaningless. The 3D model's only lesson for 4D is structural: it shows a controlled mechanism by which long-distance matrix elements are universal and states the parameter that breaks universality.
Why not use the knot-quiver correspondence as the main engine? It is proved for symmetric representations of $U(N)$, while the duality under test is for $USp(2N)$. We therefore use it only as a cross-check in the $U(N)$ analogue.
Proposed Experiments and Decisive Checks
The first experiment builds the phase map. It produces the IR TQFT, spins and quantum dimensions for every grid point, plus the count of anomaly-consistent ambiguities described above.
The second experiment is decisive. In the overlap where $|k|\gg N$ and the source paper places the theory in a gapped region, we compare the perturbative electric results for two quantities against the magnetic TQFT predictions. The first quantity is the spin shift $h_a-h_R-h_{\bar R}$. The second is the fusion weight, computed through second order in $N/k$. Agreement through that order supports the proposed squark-condensation phase. A mismatch in the leading spin shift, which is exactly rational and cannot be absorbed by higher-order corrections, falsifies that region's assignment.
The third experiment evaluates the torus links $T(2,3)$ through $T(2,9)$ and the figure-eight control on both sides. It checks orbit-number dependence, mirror symmetry under $k\to-k$, and agreement with knot-quiver results in the $U(N)$ analogue.
The fourth experiment tests Region III. Heavy-pair observables there must show dependence on separation. An exactly topological answer in Region III would contradict the claimed conformal phase.
The program succeeds if the electric weak-coupling calculation and the magnetic TQFT agree on spin shifts across the gapped regions, and the non-singlet ablation shows those channels carry the discriminating power. It fails informatively if any region's assignment is contradicted.
References
- Phases of <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:msub><mml:mrow><mml:mi>QCD</mml:mi></mml:mrow><mml:mrow><mml:mn>3</mml:mn></mml:mrow></mml:msub></mml:mrow></mml:math> from non-SUSY Seiberg duality and brane dynamics
- Color-Octet Contributions to<mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mi mathvariant="italic">J</mml:mi><mml:mi>/</mml:mi><mml:mi>ψ</mml:mi></mml:math>Photoproduction
- Next-to-leading-order QCD corrections to double prompt <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML" display="inline"><mml:mrow><mml:mrow><mml:mi>J</mml:mi></mml:mrow><mml:mo>/</mml:mo><mml:mi>ψ</mml:mi></mml:mrow></mml:math> hadroproduction
- Nonsupersymmetric Seiberg duality, orientifold QCD, and noncritical strings
- Defects in Chern-Simons theory, gauged WZW models on the brane, and level-rank duality
- Knots-quivers correspondence
- Knot-quiver correspondence: a brief review
- Multi-cover skeins, quivers, and 3d N = 2 dualities
- Brane dynamics and 3D Seiberg duality on the domain walls of 4D 𝒩 = 1 SYM