Monopole Deformations as a Finite-N Test of Non-Supersymmetric Seiberg Duality

Seiberg duality is the claim that two different-looking gauge theories, an "electric" one and a "magnetic" one, describe the same physics at long distances. When supersymmetry is present, strong mathematical control makes such claims checkable. Without supersymmetry, the proposed dualities in three and four dimensions rest mostly on two kinds of evidence. The first is a large-$N$ limit, where the non-supersymmetric theory mimics a supersymmetric one. The second is a small set of matching conditions on continuous symmetries. Neither kind of evidence says much about the finite-rank theories that matter. We propose a sharper and cheaper test. Deform both sides of a candidate duality by monopole operators, which are local operators that insert a quantized unit of magnetic flux. Then ask whether the protected quantities that survive the deformation still agree.

The protected quantities in question are 't Hooft anomalies. An 't Hooft anomaly is an obstruction to coupling a global symmetry to background fields. It is invariant under changes of energy scale, so any true duality must reproduce it exactly. A monopole deformation of charge $n$ breaks a magnetic $U(1)$ symmetry down to $\mathbb{Z}_n$. In the $\mathbb{CP}^{N-1}$ model this was shown to leave the mixed flavor–magnetic anomaly intact if and only if $\mathrm{GCD}(n,N)>1$ (Global anomaly matching in the higher-dimensional CP^{N-1} model). A duality predicts how monopole operators on one side map to operators on the other. It therefore predicts that the analogous arithmetic survival conditions coincide on both sides for every $n$ and every rank. Each deformation is a new, independent, finite-$N$ check that large-$N$ reasoning cannot supply. The checks are only legitimate once a gauge-anomaly question is settled: a bare monopole must be gauge invariant, and any discrete symmetry one gauges to repair a mismatch must be anomaly-free. That makes gauge anomaly cancellation the constraint that decides which repairs are allowed.

Core Claim

The claim is that non-supersymmetric Seiberg dualities can be tested at finite rank by requiring deformation covariance. Two families of monopole-deformed theories, labeled by monopole charge $n$ and gauge rank, must have identical surviving discrete 't Hooft anomalies, including mixed anomalies with charge conjugation, reflection, and one-form (center) symmetries. Where they fail to match, the duality must either be refined, typically by gauging a discrete symmetry, which gauge anomaly cancellation permits only if that symmetry is anomaly-free, or rejected at that rank.

What is new is using explicit symmetry-breaking deformations as a generator of constraints on dualities that lack supersymmetry. Existing checks only match anomalies of the undeformed theory.

What Prior Work Does and Does Not Do

The dualities we would test come from three papers that motivate the proposal by leaving finite-$N$ control open.

The four-dimensional orientifold QCD duality (Nonsupersymmetric Seiberg duality, orientifold QCD, and noncritical strings) proposes electric and magnetic $U(N_c)$ theories with two-index fermions. It supports them with planar equivalence to supersymmetric QCD, matching of the $SU(N_f)^3$ and $U(1)_R$ anomalies, and a non-critical string embedding. It predicts conformal windows such as $2N_c - 20/3 < N_f < 3N_c + 4/3$ for the antisymmetric case. It never considers magnetic symmetries or monopole deformations, even though the $U(1)$ factor of $U(N_c)$ carries a magnetic one-form symmetry in four dimensions.

The three-dimensional $USp(2N_c)$ Yang–Mills–Chern–Simons duality (Phases of QCD_3 from non-SUSY Seiberg duality and brane dynamics) maps out three regimes using one-loop Coleman–Weinberg potentials. In the bosonization regime, magnetic scalars condense and the low-energy theory becomes bosonic. In the flavor-breaking regime, scalar condensation breaks the flavor symmetry. In the conformal regime, the scalars stay massive and the endpoint is a conformal field theory. Its authors acknowledge that the one-loop analysis is not a non-perturbative proof.

The two-index $USp(2N)$ theory and its proposed mirror, an $SO(2N-1)$ theory with antisymmetric scalars (QCD_3 with two-index quarks, mirror symmetry, and fivebrane anti-BIons near orientifolds), come with supergravity evidence that $SO(4)$ flavor symmetry is unbroken. That evidence breaks down near the fivebranes.

The method is borrowed from two lines of work. The $\mathbb{CP}^{N-1}$ analysis supplies the template. It computes mixed anomalies involving a magnetic symmetry, including a reflection–magnetic anomaly detected on the non-orientable space $\mathbb{RP}^2$. It then derives the $\mathrm{GCD}(n,N)$ survival rule under monopole deformation and matches the result across the Néel, spin-liquid, and valence-bond-solid phases. That work is confined to abelian sigma models. Our contribution is to carry its logic into non-abelian gauge theories that are claimed to be dual.

The supersymmetric three-dimensional analysis (Generalized symmetries and anomalies of 3d N=4 SCFTs) shows that the center charges of monopole operators fix the global form of symmetries and their mixed anomalies with one-form symmetries, and it uses mirror symmetry as a cross-check. It relies on supersymmetric protection of monopole operators. We need only their gauge and center charges, which are topological data computable without supersymmetry.

The cobordism classification used for Standard Model proton stability (Proton stability: From the standard model to beyond grand unification) supplies the tool for discrete global anomalies that triangle diagrams miss. It also shows that such anomalies can differ between theories whose perturbative anomalies agree.

The gauge-anomaly side acts as a constraint rather than a mechanism. The work on symmetric mass generation (SMG), which gaps fermions through strong interactions without breaking any symmetry (Symmetric Mass Generation), establishes that a symmetric gapped phase exists exactly when all 't Hooft anomalies vanish. We use this only as the rule for reading results. When the surviving anomaly of a deformed theory is zero on both sides, a trivially gapped symmetric vacuum is allowed and the check is uninformative. When it is nonzero, both sides must realize it with the same kind of infrared phase.

Research Program

The first stage is calibration. We will build a small, reusable anomaly calculation for three-dimensional gauge theories with Chern–Simons terms. It records the gauge charge of bare monopole operators, which depends on the Chern–Simons level shifted by integrating out fermions. It also records their center and charge-conjugation quantum numbers and the resulting anomaly-inflow terms in four dimensions. The calculation must reproduce the $\mathrm{GCD}(n,N)>1$ rule and its matching across the three $\mathbb{CP}^{N-1}$ phases. It must also reproduce the monopole center charges of $T[SU(N)]$ reported in the supersymmetric three-dimensional work. If it fails either calibration it is not trusted further.

The second stage enumerates discrete symmetries in each non-supersymmetric pair. This is where the pairs become interesting. The electric $USp(2N)$ gauge group has no magnetic $U(1)$. With purely antisymmetric matter, which does not appear to carry center charge, it can instead have a $\mathbb{Z}_2$ one-form symmetry. The proposed $SO(2N-1)$ mirror has a trivial center but a $\mathbb{Z}_2$ magnetic zero-form symmetry, because its gauge group has a non-trivial fundamental group. In three dimensions, gauging a $\mathbb{Z}_2$ zero-form symmetry produces a dual $\mathbb{Z}_2$ one-form symmetry. The duality may therefore hold only between specific global forms, for example $SO$ versus $O$ or $Spin$. We will compute which global forms match, and we will check that the $\mathbb{Z}_2$ one would gauge is free of anomaly. That anomaly check is the gauge anomaly cancellation condition, and it decides whether the repair is legal.

For the $USp(2N_c)$ Chern–Simons theories with fundamentals, the relevant symmetries are charge conjugation, flavor, and, at special levels, time reversal. The gauge charge of monopoles is fixed by level quantization, which is the three-dimensional form of gauge anomaly cancellation.

The third stage is the deformation study proper. On the side where monopole operators are gauge-invariant local operators, we add charge-$n$ monopole deformations. We use the proposed duality map to find the corresponding operator on the other side, which is often a meson-like or scalar operator rather than a monopole. We then compute, as a function of $n$ and rank, which mixed anomalies survive on each side. The output is a pair of arithmetic survival rules, the analogues of $\mathrm{GCD}(n,N)$. The duality predicts that these rules are equal.

The fourth stage confronts the one-loop phase diagrams. In each of the three $USp(2N_c)$ regimes, the claimed infrared phase must carry any anomaly that survives the deformation. A trivially gapped endpoint in a region where the deformed anomaly is nonzero would falsify either the one-loop phase assignment or the duality. We will also carry out the analogous four-dimensional computation for the magnetic one-form symmetry of $U(N_c)$ orientifold QCD at finite temperature, following the $T_{\text{Néel}} \ge T_{\text{Mag}}$ style of argument. This stage is secondary.

Baselines

The naive reference point is the evidence the dualities already have: matching of continuous 't Hooft anomalies in the undeformed theories, as carried out for $SU(N_f)^3$ and $U(1)_R$ in the orientifold QCD paper and in the three-dimensional brane papers. We rerun exactly those checks with our calculation on the same list of pairs and ranks and count how many independent constraints they impose. The file reports no quantitative score for this baseline, since it is a pass/fail check that the authors report as passing. We found no reported result on Lacuna for any failure rate of continuous anomaly matching on these pairs. The quantitative predictions it supports are the conformal-window bounds quoted above.

The strongest existing competitor is the combination of planar equivalence with one-loop Coleman–Weinberg phase analysis (Phases of QCD_3 from non-SUSY Seiberg duality and brane dynamics). It is the only method in the file that makes phase predictions at moderate rank. We compare against it region by region, at the same Chern–Simons levels, flavor numbers, and ranks the paper uses. Beating it means one of two outcomes. We might identify a rank or level where its phase assignment is inconsistent with a surviving deformed anomaly. Alternatively, where it is consistent, we might derive the same phase boundary from anomaly arguments alone, independent of perturbation theory.

The key ablation isolates the deformation. We compute discrete and cobordism anomalies of the undeformed theories only, which is the program suggested by the finite-$N$ gap, and compare the number and strength of constraints against the deformed family. If the undeformed discrete checks already capture every mismatch the deformations find, the deformation ingredient adds nothing and we will say so.

Why not just…?

Why not just rely on planar equivalence at large $N_c$? Planar equivalence holds only as $N_c \to \infty$ with $N_f/N_c$ fixed, and it is blind to discrete anomalies and monopole dynamics. The two-index three-dimensional paper illustrates the limitation: large-$N$ reasoning suggested unbroken $SO(4)$, but confirming this at finite $N$ needed explicit supergravity solutions, and those broke down near the fivebranes. Our survival rules depend explicitly on rank through number-theoretic conditions. Those are precisely the effects planar equivalence cannot see.

Why not just check perturbative triangle anomalies? The reflection–magnetic anomaly in $\mathbb{CP}^{N-1}$ is invisible to triangle diagrams and requires non-orientable backgrounds. The cobordism analysis of the Standard Model shows that perturbatively identical theories can differ in global anomalies. The ablation above quantifies how many constraints the perturbative checks miss on these specific pairs.

Why not just compute discrete anomalies of the undeformed theories? This is the most serious alternative. It may catch the global-form mismatch in the $USp(2N)$/$SO(2N-1)$ pair without any deformation. Our answer is empirical. The deformation family turns one check into an infinite, rank- and charge-dependent sequence. It also probes the duality map on operators, not just the spectrum of symmetries. If the ablation shows no additional constraints, we will reduce the proposal to the undeformed computation.

Why not just train a classifier to recognize dual pairs? Classifiers for quiver Seiberg duality reach roughly 90–100% accuracy within known mutation classes but degrade on unseen classes (Quiver mutations, Seiberg duality, and machine learning). A classifier cannot certify an exact topological invariant. It is useless for deciding a new, non-supersymmetric pair at finite rank.

Decisive Checks

The first decisive check is calibration. The calculation must reproduce $\mathrm{GCD}(n,N)>1$ exactly for all tested $n$ and $N$, together with the $T[SU(N)]$ center charges. The second is the global form of the two-index mirror pair. We will determine whether the $\mathbb{Z}_2$ one-form symmetry on the $USp$ side matches the magnetic $\mathbb{Z}_2$ on the $SO$ side only after discrete gauging, and whether that gauging is anomaly-free. A negative answer would mean the mirror pair as stated fails at finite $N$. The third is the survival-rule comparison under monopole deformations for the Chern–Simons pairs. We report, for each rank and level, whether the electric and magnetic arithmetic conditions agree and whether the one-loop phase in each region can carry the surviving anomaly. A single robust mismatch is a publishable falsification. Agreement across a large grid of ranks, charges, and levels would be the strongest finite-$N$ evidence yet for these dualities.