arXiv subjects

Kazuyuki Kanaya

Publications and source records attributed to Kazuyuki Kanaya.

At least 19 recordsLinked to original sources

arXiv2026

Mixed precision solvers with half-precision floating point numbers for Lattice QCD on A64FX processor

We investigate the use of half-precision floating-point numbers (FP16) in mixed-precision linear solvers for lattice QCD simulations. Since the emergence of GPUs for general-purpose, mixed-precision algorithms that combine single-precision (FP32) with double-precision (FP64) arithmetics have become widely used in this field and others. While FP32-based methods are now well established, we examine the practicality of using FP16. In this work, we introduce rescaling steps in both the outer iterative refinement step and the inner BiCGStab solver to avoid numerical instability. In our experiments with a simple Wilson kernel, the solver shows improved stability, and the additional iteration count compared to the FP64 version remains within 20\%, indicating that the FP16 version is practical for use. We believe that the proposed rescaling methods can also benefit other mixed precision preconditioners in avoiding underflows.

hep-lat↗

arXiv2026

Lee-Yang-zero ratio method in three-dimensional Ising model

By performing Monte Carlo simulations of the three-dimensional Ising model, we apply the recently proposed Lee-Yang-zero ratio (LYZR) method to determine the location of the critical point in this model. We demonstrate that the LYZR method is as powerful as the conventional Binder-cumulant method in studying the critical point, while the LYZR method has the advantage of suppressing the violation of the finite-size scaling and non-linearity near the critical point. We also achieve a precise determination of the values of the LYZRs at the critical point, which are universal numbers. In addition, we propose an alternative method that uses only a single Lee-Yang zero and show that it is also useful for the search for the critical point.

cond-mat.stat-mech↗

arXiv2025

Locating Critical Points Using Ratios of Lee-Yang Zeros

We propose a method to numerically determine the location of a critical point in general systems using the finite-size scaling of Lee-Yang zeros. This method makes use of the fact that the ratios of Lee-Yang zeros on various spatial volumes intersect at the critical point. While the method is similar to the Binder-cumulant analysis, it is advantageous in suppressing the finite-volume effects arising from the mixing of variables in general systems. We show that the method works successfully for numerically locating the CP in the three-dimensional three-state Potts model with a nonzero external field.

hep-lat↗

arXiv2025

Lee-Yang zeros in heavy-quark QCD

We explore the distribution of Lee-Yang zeros around the critical point that appears in the heavy-quark region of QCD at nonzero temperature in lattice numerical simulations. With the aid of the hopping-parameter expansion that is well justified around the critical point in our setting, our numerical analysis is capable of analyzing the partition function for complex parameters with high accuracy. This enables precise analyses of the Lee-Yang zeros around the critical point. We study their finite-size scaling around the critical point. We also propose new methods to utilize the scaling behavior of the Lee-Yang zeros for fixing the location of the critical point.

hep-ph↗

arXiv2025

Finite-size scaling of Lee-Yang zeros and its application to the 3-state Potts model and heavy-quark QCD

We propose a new general method to study critical points (CP) using the finite-size scaling of Lee-Yang zeros (LYZ). We first study the LYZ in the three-dimensional Ising model on finite lattices. We show that the ratios of multiple LYZ (Lee-Yang-zero ratios: LYZR) have useful scaling properties similar to the Binder cumulants, providing us with a novel method to study CP. In numerical simulations of the Ising model, we confirm that this method works well. We then apply the method to analyze the CP in the three-dimensional three-state Potts model and finite-temperature QCD in heavy-quark region, which are believed to belong to the same universality class as the Ising model. In these models, the partition function at complex parameters can be evaluated by the reweighting method, which allows us to determine the LYZ by varying coupling parameters continuously around the CP. We demonstrate that the LYZR method is powerful in determining the location of the CP in these models.

hep-lat↗

arXiv2024

Finite-temperature critical point of heavy-quark QCD on large lattices

We study the finite-temperature critical point of QCD in the heavy-quark region by a scaling study of the Binder cumulant on large lattices. Extending our previous study at $N_t=4$, we perform simulations on $N_t=6$ and 8 lattices with spatial volumes up to the aspect ratio $LT=N_s/N_t=18$ and 15 ($N_s=108$ and 120), respectively, to determine the critical point in the thermodynamic limit with a high precision. To enable simulations with large spatial volumes, we adopt the hopping parameter expansion combined with a method to effectively incorporate high order terms of the expansion. The reliability of the method is confirmed by examining the effect of high order terms. Using the results of the critical point at $N_t=4$, 6, and 8, we also attempt a preliminary continuum extrapolation of the critical point in physical units.

hep-lat↗

arXiv2024

High-precision analysis of the critical point in heavy-quark QCD at $N_t=6$

Binder-cumulant analysis of the critical point in the heavy-quark region of QCD is performed by Monte-Carlo simulations with the hopping-parameter expansion at $N_t=6$. We extend our previous analysis at $N_t=4$ to finer lattices and perform high-precision analyses on large spatial volumes up to the aspect ratio $LT=N_s/N_t=18$. Higher order terms in the hopping-parameter expansion are incorporated effectively up to 14th order. The numerical results show that the violation of the finite-size scaling becomes more prominent on the finer lattice at a given aspect ratio.

hep-lat↗

arXiv2023

Critical point in heavy-quark region of QCD on fine lattices

We perform a finite-size scaling analysis of the critical point in the heavy-quark region of QCD at nonzero temperature. Our previous analysis on the Binder cumulant at $N_t=4$ is extended to finer lattices with $N_t=6$ and $8$. The aspect ratio is also extended up to $15$ to suppress the non-singular contribution. High-precision analysis of the Binder cumulant is realized by an efficient Monte-Carlo simulation with the hopping-parameter expansion (HPE). Effects of higher-order terms in the HPE are incorporated by the reweighting method.

hep-lat↗

arXiv2023

Chemical potential dependence of the endpoint of first-order phase transition in heavy-quark region of finite-temperature lattice QCD

We determine the location of the critical point where the first-order deconfining transition in the heavy-quark region turns into a crossover in finite-temperature and density lattice QCD with 2+1 flavors of Wilson quarks. Combining a hopping parameter expansion (HPE) of the quark determinant with a reweighting method, we evaluate the chemical potential dependence of the critical point. By systematically calculating the coefficients of the hopping parameter expansion up to a high order of HPE at finite chemical potential, we find that the higher order terms are strongly correlated with the Polyakov loop, which is the leading-order term, on each configuration. Moreover, their complex phases themselves, which are important at finite density, are also found to be strongly correlated with the complex phase of the Polyakov loop. Using this property, we develop a method for estimating the critical point incorporating high-order terms from calculations with only low-order terms. We report that the first-order phase transition region in the heavy-quark region becomes narrower exponentially with increasing the chemical potential. Since the hopping parameter of the critical point decreases exponentially as the density increases, the sign problem does not become serious even when the density increases, and critical points can be evaluated up to high densities.

hep-lat↗

arXiv2023

Bridge++ 2.0: Benchmark results on supercomputer Fugaku

Bridge++ is a general-purpose code set for lattice QCD simulations aiming at a readable, extensible, and portable code while keeping practical high performance. The new version 2.0 employs machine-dependent optimization, enabling flexible data layout in float/double precision, while it was fixed layout and only with the double precision in previous versions. We report the performance on supercomputer Fugaku with Arm A64FX-SVE architecture by Fujitsu.

hep-lat↗

arXiv2022

Phase structure and critical point in heavy-quark QCD at finite temperature

We study phase structure and critical point of finite-temperature QCD in the heavy-quark region applying the hopping parameter expansion (HPE). We first study finite-size scaling on the critical point on $N_t=4$ lattices with large spatial volumes taking the leading order (LO) and the next-to-leading order (NLO) effects of the HPE, and find that the critical scaling of the Z(2) universality class expected around the critical point of two-flavor QCD is realized when the aspect ratio of the lattice is larger than about 9. This enables us to determine the critical point in the thermodynamic limit with high precisions. By a study of the convergence of the HPE, we confirm that the result of the critical point with the LO (NLO) approximation of the HPE is fairly accurate for $N_t=4$ (6), while we need to incorporate higher order effects for larger $N_t$. To extend the study to large $N_t$ lattices, we then develop a method to take the effects of higher-order terms of the HPE up to a sufficiently high order. We report on the status of our study on $N_t = 6$ lattice adopting the new method.

hep-lat↗

arXiv2022

Scope and convergence of the hopping parameter expansion in finite temperature QCD with heavy quarks around the critical point

Hopping parameter expansion is a useful tool to investigate heavy dynamical quarks in lattice QCD, while the range of its applicability has been sometimes questioned. We study the convergence and the valid range of the hopping parameter expansion in the determination of the critical point (critical quark mass) of QCD with heavy quarks at finite temperature and density. On lattices with sufficiently large spatial extent, the terms in the hopping parameter expansion are classified into Wilson loop terms and Polyakov-type loop terms. We first study the case of the worst convergence in which all the gauge link variables are unit matrices and thus the Wilson loops and the Polyakov-type loops get their maximum values. We perform explicit calculation up to more than 100th order of the hopping parameter expansion. We show that the hopping parameter expansion is convergent up to the chiral limit of free Wilson quarks. We then perform a Monte-Carlo simulation to measure correlation among Polyakov-type loop terms up to the 20th order of the hopping parameter expansion. In previous studies, strong correlation between the leading order Polyakov loop term and the next-to-leading order bent Polyakov loop terms was reported and used to construct an effective theory to incorporate the next-to-leading order effect by a shift of the leading order coupling parameter. We establish that the strong correlation among Polyakov-type loop terms holds also at higher orders of the hopping parameter expansion, and extend the effective theory to incorporate higher-order effects up to high orders. Using the effective theory, we study the truncation error of the hopping parameter expansion. We find that the previous next-to-leading order result of the critical point for $N_t=4$ are well reliable. For $N_t \ge 6$, we need to incorporate higher-order effects in the effective theory.

hep-lat↗

arXiv2021

Finite-size scaling around the critical point in the heavy quark region of QCD

Finite-size scaling is investigated in detail around the critical point in the heavy-quark region of nonzero temperature QCD. Numerical simulations are performed with large spatial volumes up to the aspect ratio $N_s/N_t=12$ at a fixed lattice spacing with $N_t=4$. We show that the Binder cumulant and the distribution function of the Polyakov loop follow the finite-size scaling in the $Z(2)$ universality class for large spatial volumes with $N_s/N_t \ge 9$, while, for $N_s/N_t \le 8$, the Binder cumulant becomes inconsistent with the $Z(2)$ scaling. To realize the large-volume simulations in the heavy-quark region, we adopt the hopping parameter expansion for the quark determinant: We generate gauge configurations using the leading order action including the Polyakov loop term for $N_t=4$, and incorporate the next-to-leading order effects in the measurements by the multipoint reweighting method. We find that the use of the leading-order configurations is crucially effective in suppressing the overlapping problem in the reweighting and thus reducing the statistical errors.

hep-lat↗

arXiv2021

General purpose lattice QCD code set Bridge++ 2.0 for high performance computing

Bridge++ is a general-purpose code set for a numerical simulation of lattice QCD aiming at a readable, extensible, and portable code while keeping practically high performance. The previous version of Bridge++ is implemented in double precision with a fixed data layout. To exploit the high arithmetic capability of new processor architecture, we extend the Bridge++ code so that optimized code is available as a new branch, i.e., an alternative to the original code. This paper explains our strategy of implementation and displays application examples to the following architectures and systems: Intel AVX-512 on Xeon Phi Knights Landing, Arm A64FX-SVE on Fujitsu A64FX (Fugaku), NEC SX-Aurora TSUBASA, and GPU cluster with NVIDIA V100.

hep-lat↗

arXiv2021

Latent heat and pressure gap at the first-order deconfining phase transition of SU(3) Yang-Mills theory using the small flow-time expansion method

We study the latent heat and the pressure gap between the hot and cold phases at the first-order transition temperature $T=T_c$ of SU(3) Yang-Mills theory, using the small flow-time expansion (SF$t$X) method based on the gradient flow. We first examine alternative procedures in the SFtX method -- the order of the continuum and vanishing flow-time extrapolations. We confirm that the final results adopting the two orders, as well as other alternatives in which the perturbative order of the matching coefficients and the renormalization scale of the flow scheme are varied, are all consistent with each other. We also confirm $Δp$ is consistent with zero, as expected from the dynamical balance of two phases at $T_c$. For the latent heat in the continuum limit, we find $Δε/T^4 = 1.117(40)$ for the spatial volume $L^3$ corresponding to the aspect ratio $N_s/N_t=T_cL=8$ and $1.349(38)$ for $N_s/N_t=6$. From hysteresis curves, we show that the entropy density in the hot phase is sensitive to the spatial volume, while that in the confined phase is insensitive.

hep-lat↗

arXiv2021

Latent heat and pressure gap at the first-order deconfining phase transition of SU(3) Yang-Mills theory using the small flow-time expansion method

We study latent heat and the pressure gap between the hot and cold phases at the first-order deconfining phase transition temperature of the SU(3) Yang-Mills theory. Performing simulations on lattices with various spatial volumes and lattice spacings, we calculate the gaps of the energy density and pressure using the small flow-time expansion (SFtX) method. We find that the latent heat $Δε$ in the continuum limit is $Δε/T^4 = 1.117 \pm 0.040$ for the aspect ratio $N_s/N_t=8$ and $1.349 \pm 0.038$ for $N_s/N_t=6$ at the transition temperature $T=T_c$. We also confirm that the pressure gap is consistent with zero, as expected from the dynamical balance of two phases at $T_c$. From hysteresis curves of the energy density near $T_c$, we show that the energy density in the (metastable) deconfined phase is sensitive to the spatial volume, while that in the confined phase is insensitive. Furthermore, we examine the effect of alternative procedures in the SFtX method - the order of the continuum and the vanishing flow-time extrapolations, and also the renormalization scale and higher-order corrections in the matching coefficients. We confirm that the final results are all very consistent with each other for these alternatives.

hep-lat↗

arXiv2020

Nf=2+1 QCD thermodynamics with gradient flow using two-loop matching coefficients

We study thermodynamic properties of Nf=2+1 QCD on the lattice adopting O(a)-improved Wilson quark action and Iwasaki gauge action. To cope with the problems due to explicit violation of the Poincare and chiral symmetries, we apply the Small Flow-time eXpansion (SFtX) method based on the gradient flow, which is a general method to correctly calculate any renormalized observables on the lattice. In this method, the matching coefficients in front of operators in the small flow-time expansion are calculated by perturbation theory. In a previous study using one-loop matching coefficients, we found that the SFtX method works well for the equation of state, chiral condensates and susceptibilities. In this paper, we study the effect of two-loop matching coefficients by Harlander et al. We also test the influence of the renormalization scale in the SFtX method. We find that, by adopting the mu_0 renormalization scale of Harlander et al. instead of the conventional mu_d=1/sqrt{8t} scale, the linear behavior at large t is improved so that we can perform the t -> 0 extrapolation of the SFtX method more confidently. In the calculation of the two-loop matching coefficients by Harlander et al., the equation of motion for quark fields was used. For the entropy density in which the equation of motion has no effects, we find that the results using the two-loop coefficients agree well with those using one-loop coefficients. On the other hand, for the trace anomaly which is affected by the equation of motion, we find discrepancies between the one- and two-loop results at high temperatures. By comparing the results of one-loop coefficients with and without using the equation of motion, the main origin of the discrepancies is suggested to be attributed to O((aT)^2)=O(1/N_t^2) discretization errors in the equation of motion at N_t =< 10.

hep-lat↗

arXiv2020

Four quark operators for kaon bag parameter with gradient flow

To study the CP-violation using the $K_0-\bar{K}_0$ oscillation, we need the kaon bag parameter which represents QCD corrections in the leading Feynman diagrams. The lattice QCD provides us with the only way to evaluate the kaon bag parameter directly from the first principles of QCD. However, a calculation of relevant four quark operators with theoretically sound Wilson-type lattice quarks had to carry a numerically big burden of extra renormalizations and resolution of extra mixings due to the explicit chiral violation. Recently, the Small Flow-time eXpansion (SFtX) method was proposed as a general method based on the gradient flow to correctly calculate any renormalized observables on the lattice, irrespective of the explicit violations of related symmetries on the lattice. To apply the SFtX method, we need matching coefficients, which relate finite operators at small flow-times in the gradient flow scheme to renormalized observables in conventional renormalization schemes. In this paper, we calculate the matching coefficients for four quark operators and quark bi-linear operators, relevant to the kaon bag parameter.

hep-lat↗