Monte Carlo study on Heisenberg model with local dipolar interaction Preprint: YITP-25-43, RIKEN-iTHEMS-Report-25

Etsuko Itou, Akira Matsumoto, Yu Nakayama, and Toshiki Onagi · 2025

Read Monte Carlo study on Heisenberg model with local dipolar interaction Preprint: YITP-25-43, RIKEN-iTHEMS-Report-25 by Etsuko Itou, Akira Matsumoto, Yu Nakayama, and Toshiki Onagi (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-15874.txt.

Introduction

Etsuko Itou, Akira Matsumoto, Yu Nakayama, and Toshiki Onagi revisit, in a 2025 paper, a long-standing question about how magnets change character at their tipping point. Theory from Aharony and Fisher says the pull between distant magnetic dipoles should push a ferromagnet to a new fixed point, distinct from the standard Heisenberg one. Whether that point is truly non-perturbative stayed open for decades.

The team simulated a local version of the model on a large lattice, tuning a coupling dial that stands in for the dipole strength. At a moderate setting, the key exponent already sat far from the Heisenberg value. The spin patterns also lost a symmetry that conformal theories require, matching the claim that the new point is scale-free but not conformal. The authors read this as firm numeric evidence for the dipolar picture.

Catalog record

CREATOREtsuko Itou, Akira Matsumoto, Yu Nakayama, and Toshiki Onagi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE80 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-15874

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Monte Carlo study on Heisenberg model with local dipolar interaction Preprint: YITP-25-43, RIKEN-iTHEMS-Report-25 Etsuko Itou, Akira Matsumoto, Yu Nakayama, and Toshiki Onagi Abstract Aharony and Fisher showed that non-local dipolar effects in magnetism destabilize the Heisenberg fixed point in real ferromagnets, leading to a new fixed point, called the dipolar fixed point. The non-perturbative nature of the new fixed point, however, has not been uncovered for many decades. Inspired by the recent understanding that the dipolar fixed point is scale-invariant but not conformal invariant, we perform the Monte Carlo simulation of the local Heisenberg-dipolar model on the lattice of 40 3 40^{3} by introducing the local cost function parameterized by a parameter λ \lambda , and study its critical exponents, which should become identical to the dipolar fixed point of Aharony and Fisher in the infinite coupling limit λ = ∞ \lambda=\infty . We find that the critical exponents become noticeably different from those of the Heisenberg fixed point for a finite coupling constant λ = 8 \lambda=8 (e.g. ν = 0.601 ( 2 ) ( − 2 + 0 ) \nu=0.601(2)(^{+0}_{-2}) in the local Heisenberg-dipolar model while ν = 0.712 ( 1 ) ( − 0 + 3 ) \nu=0.712(1)(^{+3}_{-0}) in the Heisenberg model), and the spin correlation function has a feature that it becomes divergence-free, implying the lack of conformal invariance. 1 Introduction The origin of magnetism has been one of the greatest mysteries in nature since the days of ancient Greece and China. Even after the development of Maxwell’s theory of electromagnetism and classical statistical mechanics, ferromagnetism remains a mystery. As Bohr argued in 1911 in his PhD thesis, since a static magnetic field does not work on charged particles there can be no net magnetism in statistical equilibrium. It was then Heisenberg who, in 1928, brought the brilliant idea that the origin of magnetism must be quantum mechanical Heisenberg:1928mqa : It is not the electromagnetic interaction but the quantum exchange effect that will cause the alignment of spins and explain ferromagnetism. His model, the Heisenberg model, has been the starting point for understanding ferromagnetism in nature from statistical mechanics. It provides the simplest statistical models to study the universal nature of Currie’s phase transition with the help of the renormalization group, Monte Carlo simulation, and conformal bootstrap. This is a familiar story. In 1973, Aharony and Fisher challenged the above conventional wisdom Aharony_1 ; Aharony_2 . They pointed out that the long-range electromagnetic dipolar exchange between magnetization vectors, rather than the short-range quantum mechanical exchange effect of Heisernerbg, should lead to a relevant effect in the renormalization group sense although it is typically too tiny to explain the ferromagnetism itself. Accordingly, they further argue that it can drastically change the nature of Currie’s phase transition because in the theory of renormalization group, it is not the smallness of the bare parameter but whether it is relevant or irrelevant will determine the fixed point and nature of the phase transition. The salient features of the new renormalization group fixed point caused by the long-range electromagnetic dipolar exchange effect, which we call dipolar fixed point, are (1) it demands the magnetization vector fields to be constrained to be transverse, (2) as a consequence it is scale-invariant but not conformal invariant, (3) the critical exponents must be different from those of the Heisenberg fixed point. There are some experimental verifications of (1) and (2). Certain ferromagnetic materials such as Europium compounds EuO and EuS experimentally show the spin correlation functions with suppressed transverse components J.Kotzler_1986 , which is not only a clear signal of (1) but also that there is no conformal invariance as argued in Ref. Gimenez-Grau:2023lpz . Regarding (3), the measured critical exponents of EuO and EuS are close to those of the Heisenberg model PhysRevB.14.4908 , and the difference, if any, is yet to be seen. We here point out that there have not been as many theoretical works to determine the critical exponents of the dipolar fixed points as those of the Heisenberg fixed points. In addition to the one or two-loop results in the original paper by Aharony and Fisher, we have only three-loop computations in Ref. 2022NuPhB.98515990K and the functional renormalization group analysis in Ref. Nakayama:2023wrx . The situation should be in stark contrast with the case of the Heisenberg model, where the competition among the renormalization group Guida:1998bx ; 2001JMP….42…52J , Monte Carlo simulations Hasenbusch:2011zwv ; Hasenbusch:2020pwj , and conformal bootstrap Kos:2015mba ; Kos:2016ysd ; Chester:2020iyt are very severe, now culminating up to more than six digits. We should emphasize here that the dipolar fixed point is not conformal invariant and the conformal bootstrap is not applicable. In this paper, we present the first Monte Carlo predictions of critical exponents of a local Heisenberg-dipolar model by performing the simulation on a cubic lattice of size up to 40^{3} . In our work, we impose the dipolar constraint not as a consequence of the long-range interaction, but as a consequence of the local energy cost. The strength of the local energy cost is measured by a parameter \lambda in our notation. The \lambda=0 corresponds to the Heisenberg fixed point and \lambda=\infty corresponds to the dipolar fixed point of Aharony and Fisher. The merit of the local Heisenberg-dipolar Hamiltonian is that we can evaluate the updating algorithm of the Monte Carlo simulation quickly. The demerit may be that taking larger \lambda demonstrates longer auto-correlation time and slower thermalization. Indeed, these will hinder us from taking the \lambda=\infty limit directly, but we study the theory up to \lambda=8 with numerical extrapolation to \lambda=\infty in mind. Our main finding is that the measured critical exponents of the dipolar fixed points can be very different from those of the Heisenberg fixed points: \nu=0.601(2)(^{+0}_{-2}) and \eta=0.132(8)(^{+0}_{-4}) at \lambda=8 , which should be compared with \nu=0.712(1)(^{+3}_{-0}) and \eta=0.0318(27)(^{+10}_{-7}) at \lambda=0 . In Figure 1 , we compare our results of \nu with the previous Monte Carlo studies Hasenbusch:2020pwj ; Hasenbusch:2011zwv ; Hasenbusch:2000ph ; Ballesteros:1996bd ; Chen:1993zz ; Holm:1993zz ; Peczak:1991zz ; Lau:1989zz as well as the conformal bootstrap Chester:2020iyt , the \varepsilon -expansion 2022NuPhB.98515990K , and the functional renormalization group Nakayama:2023wrx . As for \lambda=0 , our result is almost consistent with the other numerical works, while for \lambda=8 it shows a clear difference from the one at \lambda=0 . Thus, it strongly suggests that the dipolar fixed point realized at \lambda=\infty would be in a different universality class from the Heisenberg fixed point. Figure 1: Comparison plot on the critical exponent \nu (for \lambda=0 and 8 ) obtained in this paper and the previous studies by the Monte Carlo Hasenbusch:2020pwj ; Hasenbusch:2011zwv ; Hasenbusch:2000ph ; Ballesteros:1996bd ; Chen:1993zz ; Holm:1993zz ; Peczak:1991zz ; Lau:1989zz (colored open symbols), the conformal bootstrap Chester:2020iyt (star-yellow symbol), the \varepsilon -expansion (filled symbols), and the functional renormalization group Nakayama:2023wrx (cross-blue-gray symbols). The filled circle-black symbols depict the 1-loop results of the \varepsilon -expansion obtained for \lambda=0 and \infty in Ref. Aharony_1 and for \lambda=8 in this paper whereas the filled square-circle ones are the 3-loop results taken from Ref. 2022NuPhB.98515990K . The two results of the functional renormalization group are obtained with the LPA’ approximation, truncating the potential at the order n=6 or

Read the complete text