[challenge]: Derive nonstandard BKT renormalization-group flows and discover Hamiltonians with α≠1/2

Released by

Youjin Deng, USTC

Contact email

yjdeng@ustc.edu.cn

Method

Analytical renormalization group / symbolic and ODE verification / Monte Carlo or tensor-network validation

Challenge issue

Generalized BKT transitions can have an essential correlation-length singularity with $\alpha\neq1/2$.

The convention used throughout this challenge is

$$
\xi(t)\sim \xi_0\exp\left[b|t|^{-\alpha}\right].
$$

This challenge has two explicit goals: (1) derive the corresponding RG flow for the long-range $q$-state Potts chain and connect its separatrix structure to $\alpha$; (2) identify and test other microscopic Hamiltonians whose defect content or RG normal form produces $\alpha\neq1/2$.

The research questions

For the conventional BKT transition, the massive-side correlation length diverges as

$$
\xi(t)\sim \xi_0\exp\left[b|t|^{-1/2}\right].
$$

The exponent $1/2$ is not universal across every transition with an essential singularity. A useful solvable starting point is the one-dimensional ferromagnetic $q$-state Potts chain with long-range $1/r^2$ interactions. Its dilute domain-wall description contains $q$-dependent kink-fusion processes and leads to

$$
\alpha(q)=\frac{2}{q+2}.
$$

The first task is to derive this result from the RG flow rather than insert it as a fit exponent. The second task is to determine what microscopic ingredients allow other Hamiltonians to realize a stable essential exponent different from $1/2$.

A successful result may be an independent derivation, a controlled numerical verification, a new Hamiltonian and RG mechanism, a counterexample to a proposed mechanism, or a justified conclusion that accessible scales cannot distinguish competing essential singularities.

Task 1 — derive the Potts kink RG flow

Use $L$ Potts spins $s_i\in\lbrace 1,\ldots,q\rbrace$ on a ring,

$$
\beta H_L
=-K\sum_{0\le i<j<L}
\delta_{s_i,s_j}\mathcal J_L(j-i),
$$

with periodic image-summed kernel

$$
\mathcal J_L(r)
=\sum_{n\in\mathbb Z}\frac{1}{(r+nL)^2}
=\left(\frac{\pi}{L}\right)^2
\csc^2\left(\frac{\pi r}{L}\right).
$$

Starting from a dilute gas of color-changing domain walls:

1. derive the logarithmic kink interaction and identify the running stiffness $\kappa$ and fugacity $y$;
2. enumerate the allowed two-kink fusion channels and their $q$-dependent multiplicities;
3. derive the lowest nontrivial RG equations, including the fusion term;
4. determine the critical separatrices near the marginal point;
5. integrate the flow to the scale $\ell_*$ where the dilute description breaks down;
6. show how $\xi\sim e^{\ell_*}$ produces $\alpha(q)=2/(q+2)$.

The expected quadratic flow, up to explicitly documented coupling conventions and analytic redefinitions, is

$$
\frac{d\kappa}{d\ell}=-2q\kappa y^2,
\qquad
\frac{dy}{d\ell}=(1-\kappa)y+(q-2)y^2.
$$

The derivation must explain the physical origin of every term. Algebraically reproducing the equations without identifying kink species and fusion rules is not sufficient.

RG validation

• recover $\alpha=1/2$ for $q=2$;
• recover $\alpha=2/5$ for $q=3$;
• recover $\alpha=1/3$ for $q=4$;
• integrate the RG equations numerically on both sides of the separatrix and verify the predicted escape-scale exponent;
• vary the operational definition of $\ell_*$ and show that it changes nonuniversal amplitudes but not $\alpha$;
• identify which higher-order terms can affect corrections to scaling without changing the leading exponent.

The result should clearly separate universal flow geometry from the nonuniversal matching between the lattice coupling $K$ and continuum RG coordinates.

Numerical test of the derived flow

Use a long-range Swendsen–Wang/Fortuin–Kasteleyn or equivalent cluster method for the pinned Hamiltonian. The minimum required models are:

• $q=2$, as the conventional $\alpha=1/2$ control;
• $q=3$, as the smallest nonstandard case;
• $q=4$, as a stretch test of the predicted $1/3$ exponent.

Measure the scaled order parameter, susceptibility, energy, cluster-size distribution, and full order-parameter histogram. Because a $1/r^2$ Hamiltonian can retain algebraic correlation tails, do not assume that a standard second-moment correlation length equals the RG crossover scale. Define the numerical proxy for $e^{\ell_*}$ before fitting and validate it on synthetic RG trajectories.

Use at least

$$
L=2^{10},2^{12},2^{14},2^{16},
$$

with larger sizes attempted after a wall-clock and autocorrelation audit. Compare:

$$
\log X=A+B|t|^{-\alpha}+C|t|^\alpha
$$

under the fixed RG prediction, a free-$\alpha$ essential form, the conventional $\alpha=1/2$ form, and an algebraic power law. Propagate uncertainty in $K_c$, use covariance-aware fits, and preregister the coupling window and excluded sizes.

The numerical goal is not merely to obtain a best-fit $\alpha$. It is to test whether the achieved dynamic range can distinguish the RG prediction from the alternatives.

Task 2 — explore Hamiltonians with $\alpha\neq1/2$

Search for microscopic models whose low-energy defects or marginal operators lead to a generalized BKT flow. Candidate mechanisms include:

• several defect species with nontrivial fusion rules;
• enlarged $SU(N)$, $\mathbb Z_N$, or coupled-order-parameter symmetry;
• multicritical points at which the leading BKT normal form is modified;
• marginal perturbations whose first nonvanishing beta-function term occurs at higher order;
• constrained or topological defects that forbid the conventional quadratic flow.

Two known starting points can be used as validation targets:

1. the bilinear–biquadratic spin-1 chain near the Uimin–Lai–Sutherland boundary, whose $SU(3)_1$ field theory predicts an essential gap scale with $\alpha=3/5$;
2. triangular-lattice three-spin interaction models with a reported generalized-BKT exponent $3/5$.

These examples do not complete the discovery task. A new candidate must provide:

1. an explicit microscopic Hamiltonian and normalization;
2. the relevant defect or operator content;
3. symmetry-allowed fusion and interaction terms;
4. the lowest nonvanishing RG equations;
5. the separatrix analysis and a derived $\alpha$;
6. a numerical observable that measures the RG escape scale;
7. a positive control and preregistered comparisons against $\alpha=1/2$ and a power law;
8. a prior-art check establishing what is genuinely new.

Changing the fitted exponent while retaining the ordinary BKT normal form is not evidence for a new universality class.

Validation

Before any new-exponent claim:

• reproduce the periodic kernel and small-$L$ partition function by exact enumeration;
• verify the $q=2$ Potts/Ising coupling conversion configuration by configuration;
• reproduce published $q=2$ and $q=3$ critical windows under the same convention;
• test the entire fitting pipeline on synthetic trajectories with known $\alpha$;
• demonstrate that the pipeline can distinguish at least two relevant competing exponents at the achieved noise and scale range;
• for a new Hamiltonian, reproduce one established limit or benchmark before approaching the proposed generalized-BKT point.

Failure of these controls is a validation failure, not evidence for an anomalous exponent.

Fairness and five-day contract

1. Day-0 scaffold: provide symbolic notebooks for the RG algebra, an ODE integrator, a tested long-range sampler, and exact small-size fixtures.
2. Locked conventions: freeze the Hamiltonian, periodic kernel, coupling normalization, reduced temperature, and definition of the crossover scale.
3. Raw-data release: retain per-size measurements, seeds, histograms, autocorrelation diagnostics, and code revision before exponent fitting.
4. Locked analysis: preregister critical windows, correction forms, size exclusions, covariance treatment, and model-selection score.
5. No one-fit verdict: a conclusion that changes after dropping the smallest size, removing a window endpoint, or changing the registered scale proxy is inconclusive.

If the numerical scaffold is unavailable, the five-day floor is the full analytical RG derivation, symbolic/ODE validation, and a technically justified survey of candidate Hamiltonians.

Adjudication

Allowed outcomes are:

1. RG derivation verified: the kink content, beta functions, separatrices, and $\alpha(q)$ follow consistently and pass symbolic/ODE checks.
2. Lattice support at accessible scales: the fixed RG exponent is stable and distinguishable from the registered alternatives.
3. New generalized-BKT candidate: an explicit Hamiltonian produces a different RG normal form and survives at least one controlled numerical test.
4. Tension with the proposed flow: analytical or numerical evidence identifies a reproducible failure after all controls pass.
5. Inconclusive: the scale range or corrections prevent discrimination.

An issue submission may complete Task 1 without discovering a new Hamiltonian, but a claim addressing Task 2 must include both a derived mechanism and an explicit model.

What is reproduction and what is new

• Re-deriving and numerically integrating the Potts kink flow is the core analytical reproduction.
• Reproducing $q=2,3$ lattice behavior validates the Monte Carlo and scaling pipeline.
• A controlled $q=4$ test can extend the published numerical evidence.
• A genuinely new result is a microscopic Hamiltonian with a derived $\alpha\neq1/2$, supported by symmetry analysis, RG flow, and an uncertainty-controlled numerical observable.
• Public derivations, code, raw data, and locked analysis records are required for citable positive, negative, or inconclusive results.

References

1. J. L. Cardy, “One-dimensional models with $1/r^2$ interactions,” J. Phys. A 14, 1407 (1981).
2. E. Luijten and H. Meßingfeld, “Criticality in one dimension with inverse square-law potentials,” arXiv:cond-mat/0104175.
3. J. M. Kosterlitz, “The critical properties of the two-dimensional XY model,” J. Phys. C 7, 1046 (1974).
4. C. Itoi and M.-H. Kato, “An extended massless phase and the Haldane phase in a spin-1 isotropic antiferromagnetic chain,” arXiv:cond-mat/9605105.
5. H. Otsuka, “Finite-size-scaling ansatz for the helicity modulus of the triangular-lattice three-spin interaction model,” arXiv:0807.3201.
6. H. Otsuka and K. Nomura, “Critical intermediate phase and phase transitions in a triangular-lattice three-spin interaction model,” arXiv:0803.3114.

The issue names no repository files, tests, or entry points. Start by identifying the requested symbolic notebooks, ODE integrator, long-range sampler, and exact small-size fixtures, then lock the Hamiltonian and analysis conventions. Done means the RG derivation and controls are reproducible, with numerical comparisons against the specified alternatives.