Improved XY Model: Advances and Extensions

• Improved XY model is a refined version of the classic planar spin system that suppresses leading scaling corrections through tuned Hamiltonians and enhanced symmetry features.
• It incorporates additional angular harmonics, lattice-pinning terms, and functional renormalization improvements to explore novel defects, phases, and critical behaviors.
• High-precision simulations and analytical treatments yield accurate universal amplitude ratios and critical exponents, validating its enhanced predictive power.

In lattice statistical mechanics, the standard XY model is the nearest-neighbor model of planar unit spins, typically written with $\mathbf S_i=(\cos\theta_i,\sin\theta_i)$ and Hamiltonian $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$ on a lattice (Lan et al., 2012). The expression “improved XY model” does not designate a single canonical modification. In the literature, it refers to several distinct constructions: tuned lattice realizations in which the leading correction to scaling is suppressed, generalized Hamiltonians with additional harmonics or lattice pinning, functional-renormalization treatments that retain the full field dependence of the effective action, nonequilibrium deformations driven by persistent noise or amplitude fluctuations, and quantum or topological generalizations that retain part of the XY structure while altering the phase structure or operator algebra (0810.2716).

1. Terminology and baseline usage

In the strict renormalization-group sense, an “improved” model is one for which the amplitude of the leading correction to scaling vanishes (Hasenbusch, 2019). In other settings, the same adjective denotes a richer microscopic Hamiltonian or a less truncated theoretical description. The 2D functional-RG treatment of the XY model, for example, uses “increasingly improved” to mean retention of the full field dependence of the effective potential $U_k(\rho)$ and of the gradient couplings $Z_k(\rho)$ and $Y_k(\rho)$, rather than a low-order $\phi^4$-type truncation (Jakubczyk et al., 2016). In the toric quantum construction, the “improvement” is of a different type again: gauge invariance is enforced energetically so that the non-topological local $U(1)$ degree of freedom is removed and only topological data remain in the ground space (Moses et al., 2023).

Usage of “improved”	Representative construction	Main effect
RG-improved lattice realization	Tuned $\phi^4$ or clock model (0810.2716, Hasenbusch, 2019)	Leading correction to scaling suppressed
Microscopic deformation	Added nematic, lattice-pinning, or higher-order terms (Basak et al., 2017, Samlodia et al., 2024)	New defects, phases, or universality classes
Improved field theory	Complete $\partial^2$ derivative expansion (Jakubczyk et al., 2016)	Accurate nonuniversal thermodynamics
Gauge-enforced quantum version	Toric $xy$ model (Moses et al., 2023)	Purely topological order

This plurality of meanings is central to the subject. A useful way to organize the topic is therefore by asking what is being improved: scaling behavior, microscopic symmetry structure, thermodynamic description, nonequilibrium realism, or quantum/topological content.

2. Improved realizations of the three-dimensional XY universality class

A paradigmatic use of “improved XY model” occurs in high-precision studies of the 3D XY universality class. In the two-component $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$0 model on the simple cubic lattice, there exists a value $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$1 such that the leading correction-to-scaling amplitude vanishes. The reported estimate is $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$2, and simulations at $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$3 and $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$4 were chosen because leading corrections there should be at least $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$5 times smaller than in the standard XY model (0810.2716). This tuning enabled unusually precise determinations of universal amplitude ratios, including $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$6, $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$7, $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$8, and $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$9, with the last quantity in excellent agreement with experiment on the $U_k(\rho)$0He $U_k(\rho)$1-transition (0810.2716).

A related construction replaces the microscopic $U_k(\rho)$2 variable by a generalized clock variable while maintaining XY criticality. The $U_k(\rho)$3-state clock model on the simple cubic lattice is tuned by a parameter $U_k(\rho)$4, and the improvement condition is $U_k(\rho)$5. For the $U_k(\rho)$6 model, the tuned value is $U_k(\rho)$7 (Hasenbusch, 2019). Although the microscopic symmetry is only $U_k(\rho)$8, the anisotropy is strongly irrelevant at the XY fixed point, with RG exponent $U_k(\rho)$9, so $Z_k(\rho)$0 symmetry emerges at large length scales at criticality (Hasenbusch, 2019). Finite-size scaling then yields high-precision exponent estimates $Z_k(\rho)$1, $Z_k(\rho)$2, and $Z_k(\rho)$3 on lattices up to $Z_k(\rho)$4 (Hasenbusch, 2019).

The importance of these improved realizations is sharpened by comparison with the plain 3D XY model. Large-scale GPU simulations of the classical XY model on a cubic lattice, without switching to an improved $Z_k(\rho)$5 action, nevertheless reached $Z_k(\rho)$6 and obtained $Z_k(\rho)$7 and $Z_k(\rho)$8, in agreement with the best microgravity $Z_k(\rho)$9He estimate $Y_k(\rho)$0 (Lan et al., 2012). This comparison shows that “improvement” is not necessary for precision, but it remains a standard route to cleaner finite-size scaling and tighter control of systematic errors.

3. Symmetry-deformed and generalized XY models

A second major meaning of improvement is the explicit deformation of the XY Hamiltonian by additional angular harmonics. The generalized $Y_k(\rho)$1 XY model supplements the usual nearest-neighbor term with a nematic second harmonic,

$Y_k(\rho)$2

with $Y_k(\rho)$3, so that $Y_k(\rho)$4 recovers the ordinary XY model and $Y_k(\rho)$5 gives a pure nematic model (Samlodia et al., 2024). The $Y_k(\rho)$6 sector introduces half-integer vortices in addition to the usual integer vortices. The resulting phase diagram contains ferromagnetic, nematic, and disordered phases, separated by BKT, half-BKT, and Ising transition lines; the lines meet in a small region around $Y_k(\rho)$7 and $Y_k(\rho)$8 (Samlodia et al., 2024). This model is a controlled deformation that interpolates between $Y_k(\rho)$9-dominated and $\phi^4$0-dominated regimes.

A closely related but physically distinct deformation arises in quantum Hall electron nematics. There the starting point is a 2D XY nematic with headless director field $\phi^4$1, order parameter $\phi^4$2, and Hamiltonian

$\phi^4$3

The improvement consists of adding the lattice-pinning term $\phi^4$4, which explicitly represents the fourfold symmetry of the GaAs/AlGaAs host crystal (Basak et al., 2017). With $\phi^4$5, the XY-plus-field description fits the transport anisotropy only down to about $\phi^4$6 mK; with nonzero $\phi^4$7, the entire temperature evolution can be matched, and the paper reports that $\phi^4$8, $\phi^4$9, and $U(1)$0 mK give a good fit over the full range (Basak et al., 2017). The substantial $U(1)$1 drives the system away from XY/BKT behavior and into the Ising universality class, leading to proposed diagnostics such as low-temperature hysteresis under a swept orienting field (Basak et al., 2017).

Other generalized XY models extend the same logic to still higher harmonics. In the 2D model with competing higher-order interactions of alternating sign and exponentially decreasing magnitude, the pair potential is

$U(1)$2

Its phase structure depends strongly on whether $U(1)$3 is odd or even: any odd $U(1)$4 gives two consecutive transitions to distinct ferromagnetic quasi-long-range-ordered phases, while even $U(1)$5 gives two transitions only when $U(1)$6, in which case a competition-induced canted ferromagnetic phase appears (Žukovič et al., 2021). In the limit $U(1)$7, only one finite-temperature transition remains, and for $U(1)$8 the transition temperature can be driven “practically to zero” (Žukovič et al., 2021).

In three dimensions, a different extension adds a biquadratic exchange term to the XY ferromagnet,

$U(1)$9

This model supports paramagnetic, quadrupole long-range order (QLRO), and dipole-quadrupole long-range order (DLRO) phases (Nagata et al., 2013). The single paramagnet-to-DLRO transition is second order at large $\phi^4$0, becomes first order for $\phi^4$1, and splits into separate paramagnet$\phi^4$2QLRO and QLRO$\phi^4$3DLRO transitions below $\phi^4$4 (Nagata et al., 2013). Here the additional interaction reorganizes the global phase diagram much more strongly than it changes the continuous-transition exponents.

4. Improved field-theoretic descriptions and thermodynamics

Improvement can also mean replacing a coarse continuum approximation by a more faithful field-theoretic flow. In the functional-renormalization treatment of the 2D XY model, the starting point is the actual lattice XY model mapped to a field theory whose bare local term is $\phi^4$5, so the initial potential contains all powers of the field (Jakubczyk et al., 2016). The central truncation keeps the full field dependence of the effective potential and of two independent gradient couplings,

$\phi^4$6

This “complete $\phi^4$7” derivative expansion is specifically designed to improve nonuniversal thermodynamics in the regime above $\phi^4$8 where a simple critical theory is insufficient (Jakubczyk et al., 2016).

The main payoff is quantitative control of entropy and specific heat. The free energy follows from the $\phi^4$9 effective action, and the specific heat peak is found to be pronounced and asymmetric, located at about $\partial^2$0 with height $\partial^2$1 (Jakubczyk et al., 2016). These compare well with Monte Carlo values $\partial^2$2 and $\partial^2$3 (Jakubczyk et al., 2016). By contrast, a simplified $\partial^2$4-type truncation can produce unphysical behavior, including negative specific heat in some temperature range, because it discards the infinitely many higher vertices encoded in the Bessel-function bare action (Jakubczyk et al., 2016). In this sense, the improvement is not a new universality class but a better representation of the same model’s nonuniversal thermodynamics.

A different analytical refinement concerns the low-temperature continuum limit. For the classical XY model in bounded domains with Dirichlet boundary conditions, if the inverse temperature satisfies $\partial^2$5, equivalently $\partial^2$6, then the properly rescaled gradient fluctuations converge to standard Gaussian white noise (Newman et al., 2016). The proof proceeds through a vortex-free coupling to a uniformly convex gradient model and a Helffer–Sjöstrand analysis. This result does not define a new Hamiltonian, but it sharpens the spin-wave approximation into a rigorous scaling-limit statement (Newman et al., 2016).

5. Nonequilibrium and application-driven deformations

Several improved XY models are motivated by nonequilibrium physics rather than equilibrium criticality. In the persistent-noise XY model, white thermal noise is replaced by time-correlated Ornstein–Uhlenbeck noise with persistence time $\partial^2$7,

$\partial^2$8

The stationary spin-wave theory acquires a persistence-induced $\partial^2$9 term, and the algebraic correlation exponent becomes $xy$0 (Shi et al., 25 Feb 2026). The model remains quasi-ordered even when $xy$1 exceeds the equilibrium bound $xy$2, yet the order-disorder transition remains BKT type, with the best consistency obtained for the standard BKT value $xy$3; the dynamic exponent stays near $xy$4 as $xy$5 varies (Shi et al., 25 Feb 2026). This is a nonequilibrium deformation that preserves the topological transition mechanism while changing quantitative scaling.

A very different application-driven modification is the modified planar rotator model for geostatistics,

$xy$6

with the specific choice $xy$7 (Žukovič et al., 2015). The aim is to replace the standard XY model’s quasi-long-range, power-law correlations by flexible short-range correlations suited to geophysical and environmental data. On an $xy$8 square lattice with vectorized checkerboard Metropolis updates and smooth-continuation boundary conditions, the resulting spin-angle fields are well fit by the Matérn covariance family, with temperature $xy$9 and simulation time $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$00 jointly controlling the correlation length $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$01 and smoothness $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$02 (Žukovič et al., 2015). Representative fits include $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$03 and $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$04 at $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$05, $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$06, and $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$07 at $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$08 on the same lattice (Žukovič et al., 2015). Here the improvement is judged by compatibility with short-range spatial statistics rather than by XY critical behavior.

The “simple XY model for cascade transfer” modifies the meaning of the spin itself. Each site carries a two-component real vector $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$09, interpreted as a local velocity-like degree of freedom, and the dynamics contains a projection operator

$$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$10

Amplitude fluctuations, forcing, and friction break detailed balance and allow steady flux across scales (Tanogami et al., 2021). The model exhibits a scale-independent inverse energy flux $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$11 in the inertial range and a one-dimensional energy spectrum $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$12, numerically confirmed in $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$13 and $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$14 (Tanogami et al., 2021). This places the model in a universality class different from standard fluid turbulence, while retaining a recognizable XY-like alignment structure (Tanogami et al., 2021).

6. Quantum, topological, and non-Hermitian generalizations

In quantum settings, improvement can mean enforcing constraints that remove the non-topological sector of the classical model. The toric $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$15 model assigns an $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$16 Hilbert space to each edge, with basis $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$17, and Hamiltonian

$$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$18

The ground states are exactly those satisfying gauge invariance at every vertex and the cocycle condition on every face, so the ground space is canonically associated to $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$19, or equivalently $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$20 (Moses et al., 2023). Excitations are stabilizer violations: face excitations are vortices labeled by integers $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$21, vertex defects are labeled by phases $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$22, and the braiding relation $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$23 implies anyonic mutual statistics (Moses et al., 2023). This construction recasts XY winding data as purely topological order.

A different quantum deformation is the non-Hermitian anisotropic XY chain,

$$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$24

which is invariant under intrinsic rotation-time reversal $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$25 rather than $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$26 (Zhang et al., 2012). After Jordan–Wigner, Fourier, and complex Bogoliubov transformations, the exact single-particle spectrum is

$$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$27

The spectrum is fully real in the $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$28-unbroken region and becomes complex beyond exceptional points; in the thermodynamic limit the phase boundary is the hyperbola $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$29 (Zhang et al., 2012). The Hermitian counterpart is approximately an isotropic nearest-neighbor XY chain away from criticality (Zhang et al., 2012).

Separate lines of work improve access to the quantum XY model without changing its basic Hamiltonian. In the spin-$$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$30 quantum XY model on the square lattice, an improved estimator inside the loop-cluster algorithm averages analytically over all $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$31 cluster flips and yields high-precision measurements of the magnetization distribution and constraint effective potential, in excellent agreement with low-energy effective field theory once $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$32, $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$33, and $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$34 are fixed with better than permille precision (Gerber et al., 2011). In one-dimensional spin chains, compressed quantum simulation maps the evolution of an $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$35-qubit XY model onto circuits on $$\mathcal H=-J\sum_{\langle i,j\rangle}\cos(\theta_i-\theta_j)$$36 qubits, allowing adiabatic evolution, quenching, and finite-time dynamics to be reproduced on an exponentially smaller register (Boyajian et al., 2013).

The surveyed literature therefore uses “improved XY model” in a family resemblance sense rather than as a unique model name. In the narrow RG sense, it denotes tuned representatives of the XY universality class with suppressed scaling corrections. In broader usage, it denotes deformations that add nematic harmonics, lattice pinning, higher-order competition, persistent noise, amplitude dynamics, gauge constraints, or non-Hermitian anisotropy. Across these constructions, the common theme is preservation of some recognizable XY sector together with a deliberate modification of the ultraviolet description, defect content, or symmetry structure to expose new universal behavior, improve numerical precision, or adapt the model to a different physical regime.