B_K with two flavors of dynamical overlap fermions
S. Aoki, H. Fukaya, S. Hashimoto, J. Noaki, T. Kaneko, H. Matsufuru, T. Onogi, N. Yamada
备注：
36 pages, 14 figures, comments and references added, analysis and systematic error revised, minor change in the final result. version to appear in PRD, reference corrected
We present a two-flavor QCD calculation of $B_K$ on a $16^3 \times 32$ lattice at $a\sim 0.12$ fm (or equivalently $a^{-1}\sim$1.67 GeV). Both valence and sea quarks are described by the overlap fermion formulation. The matching factor is calculated non-perturbatively with the so-called RI/MOM scheme. We find that the lattice data are well described by the next-to-leading order (NLO) partially quenched chiral perturbation theory (PQChPT) up to around a half of the strange quark mass ($m_s^{\rm phys}/2$). The data at quark masses heavier than $m_s^{\rm phys}/2$ are fitted including a part of next-to-next-to-leading order terms. We obtain $B_K^{\bar{\rm MS}}(2 {\rm GeV})= 0.537(4)(40)$, where the first error is statistical and the second is an estimate of systematic uncertainties from finite volume, fixing topology, the matching factor, and the scale setting.
We measure the low lying eigenmodes of an overlap Dirac operator on 2--flavor staggered configurations. By comparing the eigenmode distribution to the predictions of Random Matrix Theory we test to what accuracy staggered configurations describe continuum QCD. The agreement between the numerical data and RMT implies that at our quark mass values the lattice artifacts of the staggered configurations are comparable to overlap configurations. We identify the overlap valence mass that best matches the staggered sea quarks and predict the value of the infinite volume 2-flavor chiral condensate.
We present a set of related Hybrid Monte Carlo methods to simulate an arbitrary number of dynamical overlap fermions. Each fermion is represented by a chiral pseudo-fermion field. The new algorithm reduces critical slowing down in the chiral limit and for sectors of nontrivial topology.
We report on the status of the dynamical overlap QCD simulation project by the JLQCD collaboration. After completing two-flavor QCD simulation on a 16^3x32 lattice at lattice spacing a 0.12 fm, we started a series of runs with 2+1 flavors. In this report, we describe an outline of our algorithms, parameter choices, and some early physics results of this second phase of our project.
We report on our study of flavor-singlet mesons in three-flavor QCD with dynamical overlap quarks. Gauge ensembles are generated on a 16^3 x 48 lattice at a lattice spacing of 0.10 fm with the strange quark masses around its physical value m_s,phys and up and down quark masses down to m_s,phys/5. Connected and disconnected meson correlators are calculated using the all-to-all quark propagator. We present our preliminary results on the spectrum of flavor-singlet pseudoscalar and vector mesons.
In this talk I will emphasize the role of the Truncated Overlap Fermions in showing the equivalence between the Domain Wall and Overlap Fermions up to an irrelevant factor in the fermionic integration measure. I will also show how Domain Wall type fermions with a finite number of flavors can be used to accelerate propagator calculations of their light partner in the infinite flavor limit.
We study the sigma term and the strange quark content of nucleon in 2+1-flavor QCD with dynamical overlap fermions. We analyze the lattice data of nucleon mass taken at two different strange quark masses with five values of up and down quark masses each. Using the reweighting technique, we study the strange quark mass dependence of the nucleon and extract the strange quark content.
Staggered fermions with smeared links can have greatly improved chiral properties. In a recent paper we introduced a simple and effective method to simulate four flavors of staggered smeared link fermions. In this work we extend the four flavor method to two flavors. We define the two flavor action by the square root of the four flavor fermionic determinant and show that by using a polynomial approximation the two flavor action can be evaluated with the necessary accuracy. We test this method by studying the finite temperature phase structure with hypercubic smeared (HYP) link staggered action…
We study the finite temperature phase transition of four and two flavor staggered fermions with hypercubic smeared link actions. These preliminary studies suggest that the improved flavor symmetry of the fermionic action can have a significant effect on the finite temperature phase diagram.
We present a calculation of the nucleon sigma term on two-flavor QCD configurations with dynamical overlap fermions. We analyse the lattice data for the nucleon mass using the baryon chiral perturbation theory. Using partially quenched data sets, we extract the connected and disconnected contributions to the nucleon sigma term separately. Chiral symmetry on the lattice simplifies the determination of the disconnected contribution. We find that the strange quark content, which determines the neutralino dark matter reaction rate with nucleon through the Higgs boson exchange, is much smaller than…
I discuss minimally doubled fermions fermions as an ultra-local formulation on the lattice for sea quarks that realize a non-singlet chiral symmetry. I introduce a non-singlet mass term for Karsten-Wilczek fermions and identify the appropriate representation of the SU(2) flavor group at finite lattice spacing. I present an algebraic proof that the symmetry of the quark determinant under charge conjugation and reflections of the Euclidean axes is preserved for Karsten-Wilczek fermions as sea quarks. Finally, I discuss how the flavor components in meson correlation functions with Karsten-Wilczek…
We compare the performances of the exact one-flavor algorithm (EOFA) and the rational hybrid Monte Carlo algorithm (RHMC), for dynamical simulations of lattice QCD with domain-wall fermion.
At non-zero lattice spacing the flavor symmetry of staggered fermions is broken to a discrete subgroup. We show that in the chiral limit the flavor symmetry of the pion effective Lagrangian enlarges to an SO(4) subgroup of the continuum SU(4) symmetry. This provides an explanation for observed degeneracies in the pion spectrum.
We have studied the 3-flavor, finite temperature, QCD phase transition with staggred fermions on an $ N_t=4$ lattice. By studying a variety of quark masses we have located the critical point, $m_c$, where the first order 3-flavor transition ends as lying in the region $0.32 \le m_c \le 0.35$ in lattice units
We construct positive-definite pseudofermion actions for one fermion flavor in lattice field theory, for Wilson and domain-wall fermions respectively. The positive definiteness of these actions ensures that they can be simulated with the Hybrid Monte Carlo (HMC) method. For lattice QCD with optimal domain-wall quarks, we compare the efficiency of HMC simulations of 2-flavor and (1+1)-flavor, and find that the efficiency ratio is about 3:2.
We study improved Kogut-Susskind fermion actions focusing on flavor symmetry restoration. Several variants of fat actions suitable for dynamical simulations are considered, including an action with no tree level order a-squared errors. The spectrum of all the pions is computed and used as a measure of flavor symmetry violation. Finally, the Naik term is introduced to restore rotational symmetry.
Accurately calculating the mass of flavor-singlet meson states from numerical lattice simulations is an important milestone for lattice QCD. Careful measurement of the full pseudoscalar flavor-singlet propagator is also a crucial step in understanding the dynamics of the fermion sea on the lattice, in particular for potentially non-trivial formulations such as with 2+1-flavor staggered fermions. We briefly describe details of a dynamical QCD calculation using improved staggered fermions, with 30,000 trajectories, that was run for our studies of flavour singlet mesons.
We give an explicit proof for the locality of staggered overlap operators. The proof covers the original two flavor construction by Adams as well as a single flavor version. As in the case of Neuberger's operator, an admissibility condition for the gauge fields is required.
We discuss the impact of various improvements on simulations of dynamical overlap fermions using the Hybrid Monte Carlo algorithm. We focus on the usage of fat links and multiple pseudo-fermion fields.
We propose a new method for Hybrid Monte Carlo (HMC) simulations with odd numbers of dynamical fermions on the lattice. It employs a different approach from polynomial or rational HMC. In this method, gamma-five hermiticity of the lattice Dirac operators is crucial and it can be applied to Wilson, domain-wall, and overlap fermions. We compare HMC simulations with two degenerate flavors and (1 + 1) degenerate flavors using optimal domain-wall fermions. The ratio of the efficiency, (number of accepted trajectories) / (simulation time), is about 3:2. The relation between pseudofermion action of c…
We present further data in search of a parity-flavor-breaking phase in simulations of dynamical QCD with two flavors of light Wilson fermions in the strong coupling region. This is done on lattice sizes of 4^4 an up to 10^4 for a variety of values of beta and kappa as well as the coefficient, h, of an explicit breaking term included in the action. We confirm the existence of a region in the beta-kappa plane where such a phase exists at beta=6/g^2 smaller than 5.0 and above kappa_c.
We briefly review the overlap formalism for chiral gauge theories, the overlap Dirac operator for massless fermions and its connection to domain wall fermions. We describe properties of the overlap Dirac operator, and methods to implement it numerically. Finally, we give some examples of quenched calculations of chiral symmetry breaking and topology with overlap fermions.
An overview of the current status of algorithmic approaches to dynamical overlap fermions is given. In particular the issue of changing the topological sector is discussed.
In order to construct improved overlap fermions, we start from a short ranged approximate Ginsparg-Wilson fermion and insert it into the overlap formula. We show that its polynomial evaluation is accelerated considerably compared to the standard Neuberger fermion. In addition the degree of locality is strongly improved.
We present data testing the existence of a parity-flavor breaking phase in simulations of QCD with two flavors of light Wilson fermions. This is done by explicit simulations on lattice sizes of $6^4$, $8^4$ and $10^4$ for a variety of values of $β$ and $κ$ as well as the coefficient, $h$, of an explicit breaking term included in the action. We find that at $β=6/g^2 $ equal to or greater than 5.0 extrapolation in the parameter $h$ as well as in the lattice volume show no indication of a phase where parity and flavor are spontaneously broken in the limit of zero $h$.
We report our study on the critical endpoint of the finite temperature phase transition in 4-flavor QCD with Wilson-Clover fermions. Using the kurtosis intersection method, we determined the critical endpoint on lattices with $N_t$ = 4, 6 and 8. Our continuum extrapolated results show that the pseudo-scalar meson mass at the critical endpoint, $m_\mathrm{PS,E}$, for 4-flavor is clearly larger than that for 3-flavor. We also compared our results to those with staggered fermions and found that $m_\mathrm{PS,E}/T_E$ for 4-flavor with Wilson fermions might remain finite even in the continuum limit…
We discuss the kernel spectrum, locality properties and the axial-vector renormalization constant of UV-filtered overlap fermions. We find that UV-filtered overlap fermions have a better conditioned kernel, better locality and an axial-vector renormalization constant closer to 1 than their unfiltered counterparts, even if the shift parameter $ρ$ is simply set to 1.
We perform a systematic study of low-lying eigenmodes of $H_w$ with various gauge actions to find the optimal choice for dynamical overlap fermion simulations, with which one may achieve lower numerical cost for HMC and better locality property of the overlap kernel. For this purpose, our study is made with emphasis on the distribution of low-lying eigenvalues and the mobility edge with and without dynamical overlap fermions.
We report on our efforts to implement overlap fermions on NVIDIA GPUs using CUDA, commenting on the algorithms used, implemetation details, and the performance of our code.
In this contribution the costs of simulations employing domain wall and overlap fermions are estimated. In the discussion we will stay within the quenched approximation.
Recently a formulation of overlap fermions at finite density based on an analytic continuation of the sign function was proposed. We study this proposal by analyzing the energy and number densities for free fermions as a function of the chemical potential and the temperature. Our results show that overlap fermions with chemical potential give rise to the correct continuum behavior.
The RBC and UKQCD Collaborations continue to produce 2+1 flavor domain wall fermion ensembles, currently focusing on an ensemble with a $96^3 \times 192$ volume on SUMMIT at ORNL with $1/a \approx 2.8$ GeV, and smaller ensembles at stronger couplings. The $1/a \approx 2.8$ GeV ensemble uses the Exact One Flavor Algorithm for the strange quark, along with the Multisplitting Preconditioned Conjugate Gradient for solving the Dirac equation. We report on our progress and experience to date with the evolution of this ensemble.
We compute the difference in the renormalization of flavor singlet and nonsinglet fermion bilinear operators, to two loops in perturbation theory. Our results are applicable to a rather wide class of lattice actions with Symanzik improved gluons, stout links and clover fermions, including the Twisted Mass and SLiNC actions. A more detailed presentation of our results, along with relevant references, will appear in a forthcoming publication.
It is shown that free lattice Dirac fermions defined by overlap Dirac operator fulfill the Osterwalder-Schrader reflection positivity condition with respect to the link-reflection.
Studies of non-interacting lattice fermions give an estimate of the size of discretization errors and finite size effects for more interesting problems like finite temperature QCD. We present a calculation of the thermodynamic equation of state for free domain wall and overlap fermions.
I show how to avoid a two level nested conjugate gradient procedure in the context of Hybrid Monte Carlo with the overlap fermionic action. The resulting procedure is quite similar to Hybrid Monte Carlo with domain wall fermions, but is more flexible and therefore has some potential worth exploring.
We present results of a hybrid Monte-Carlo algorithm for dynamical, $n_f=2$, four-dimensional QCD with overlap fermions. The fermionic force requires careful treatment, when changing topological sectors. The pion mass dependence of the topological susceptibility is studied on $6^4$ and $12\cdot 6^3$ lattices. The results are transformed into physical units.
Using lattice overlap fermions, we have computed the 1-loop renormalization factors of several operators that measure DIS structure functions and weak amplitudes. Computer codes written in the algebraic manipulation language FORM have been used. The improvement of the operators is also discussed.
We study the lattice \njl~model with two flavors of Wilson fermions in the large $N$ limit, where $N$ is the number of `colors'. For large values of the four-fermion coupling we find a phase in which both, flavor symmetry and parity, are spontaneously broken. In accordance with general expectations there are three massless pions on the phase boundary, but only two of them remain massless inside the broken phase. This is analogous to earlier results obtained in lattice QCD, indicating that this behavior is a very general feature of the Wilson term.
We present studies of the SU(2) gauge theory with 4, 6 and 10 fermion flavors. These models are expected to lie on both sides of the edge of the conformal window, where the theory has an infrared fixed point. We observe that the coupling grows with the length scale at four flavors, implying QCD-like behavior. At ten flavors the results are compatible with a Bank-Zaks type fixed point. The results at six flavors remain inconclusive: the running is slow towards the infrared but the range and accuracy of the study are insufficient for determining the existence of a fixed point.
We study quenched QCD just above the phase transition temperature using overlap Fermions. Exact zero modes of the overlap operator are localized. Chiral symmetry is restored, as indicated by the behavior of the chiral condensate after subtracting the effects of zero modes. The vector and pseudo-scalar screening masses are close to ideal gas values.
We discuss the Schrödinger functional in lattice QCD with staggered fermions and relate it, in the classical continuum limit, to the Schrödinger functional regularized with Wilson fermions. We compute the strong coupling constant defined via the Schrödinger functional with staggered fermions at one loop and show that it agrees with the continuum running coupling constant in the Schrödinger functional formalism. We compute this running coupling in the ``weak coupling phase'' of many flavor QCD numerically at several values of the bare coupling and for several system sizes from $L/a=4$ to 12. Th…
I apply a recently developed algorithm for reweighting simulations of lattice QCD from one quark mass to another to simulations performed with overlap fermions in the epsilon regime. I test it by computing the condensate from distributions of the low lying eigenvalues of the Dirac operator. Results seem favorable.
We report on our ongoing study of nucleon properties, the nucleon mass and the strange quark content (<N|ss|N>) in particular. These calculations are done on (2+1)-flavor Domain Wall Fermion (DWF) ensembles with Iwasaki gauge action with 2 different lattice spacings (a ~ 0.08, 0.11fm) and DWF with Dislocation Suppressing Determinant Ratio(DSDR) (a ~ 0.14fm), generated by RBC/UKQCD collaborations.
This paper reviews the most popular methods which are used in lattice QCD to compute the determinant of the lattice Dirac operator: Gaussian integral representation and noisy methods. Both of them lead naturally to matrix function problems. We review the most recent development in Krylov subspace evaluation of matrix functions. The second part of the paper reviews the formal relationship and algebraic structure of domain wall and overlap fermions. We review the multigrid algorithm to invert the overlap operator. It is described here as a preconditioned Jacobi iteration where the preconditioner…
Employing the spin-flavor representation, we investigate the structures of the doubler-mixing symmetries and the mechanisms of their spontaneous breakdown in four types of lattice fermion formulation. We first revisit the $U(4)\timesU(4)A$ symmetries of the naive fermion with the vanishing bare mass m, and re-express them in terms of the spin-flavor representation. We apply the same method to the Wilson fermion, which possesses only the U(1) vector symmetry for general values of m. For a special value of m, however, there emerges an additional U(1) symmetry to be broken by pion condensation. W…
Comparison of the mixed-fermion-action Effects using different fermion and gauge actions with 2+1 and 2+1+1 flavors
Hangzhou Institute for Advanced Study, UCAS（杭州高等研究院，中国科学院大学）；Institute of Theoretical Physics, Chinese Academy of Sciences（中国科学院理论物理研究所）；University of Chinese Academy of Sciences（中国科学院大学）；The Cyprus Institute（塞浦路斯研究所）；Indiana University（印第安纳大学）
The action of the overlap-Dirac operator on a vector is typically implemented indirectly through a multi-shift conjugate gradient solver. The compute-time required depends upon the condition number, $κ$, of the matrix that is used as the overlap kernel. While the Wilson action is typically used as the overlap kernel, the FLIC (Fat Link Irrelevant Clover) action has an improved condition number and provides up to a factor of two speedup in evaluating the overlap action. We summarize recent progress on the use of FLIC overlap fermions.
We present two improvements to our previous dynamical overlap HMC algorithm. We introduce a new method of differentiating the eigenvectors of the Kernel operator, which removes an instability in the fermionic force. Secondly, by simulating part of the fermion determinant exactly, without pseudo-fermions, we are able to increase the rate of topological tunnelling by a factor of more than ten, reducing the auto-correlation.
Based on recent work by Adams, I construct a lattice fermion operator that fully lifts the staggered flavor degeneracy. The resulting operator is of Wilson type but smaller by a factor of 4, better conditioned and contains 3 instead of 15 doublers. It is further suggested that this operator may be used as a candidate kernel operator to an overlap construction. Prospects for practical applications and potential problems of the new discretizations are briefly discussed.
We examine the taste structure of eigenvectors of the staggered-fermion Dirac operator. We derive a set of conditions on the eigenvectors of modes with small eigenvalues (near-zero modes), such that staggered fermions reproduce the 't Hooft vertex in the continuum limit. We also show that, assuming these conditions, the correlators of flavor-singlet mesons are free of contributions singular in $1/m$, where $m$ is the quark mass. This conclusion holds also when a single flavor of sea quark is represented by the fourth root of the staggered-fermion determinant. We then test numerically, using th…
We describe an explicit construction of approximate Ginsparg-Wilson fermions for QCD. We use ingredients of perfect action origin, and further elements. The spectrum of the lattice Dirac operator reveals the quality of the approximation. We focus on beta =6 for optimisation. Such fermions are intended to be inserted into the overlap formula. Hence we also test the speed of convergence under polynomial evaluation of the overlap formula.
Numerical experiences are reported with overlap fermions which employ the Brillouin action as a kernel. After discussing the dispersion relations of both the kernel and the resulting chiral action, some of the physics features are addressed on quenched backgrounds. We find that the overlap with Brillouin kernel is much better localized than the overlap with Wilson kernel. Also a preliminary account is given of the cost of the formulation, in terms of CPU time and memory.
Simulations for the thermodynamics of the 2+1 flavor QCD are performed employing chiral fermions. The use of Möbius domain-wall fermions with stout-link smearing is more effective on the finer lattices where all the relevant chiral symmetries are realized more accurately. We report on the initial simulations near the (pseudo) critical point using the line of constant physics with an average $ud$ quark mass slightly heavier than physical at $a\lesssim 0.1$ fm.
Overlap formalism deals with the construction of chiral gauge theories on the lattice. These set of lectures provide a pedagogical introduction to the subject with emphasis on chiral anomalies and gauge field topology. Subtleties associated with the generating functional for gauge theories coupled to chiral fermions are discussed.
Staggered fermions with 4 tastes are expected to describe 4-flavor QCD in the continuum limit, therefore at finite lattice spacing the staggered determinant should be equivalent to an SU(4) flavor-symmetric system up to lattice artifacts. This equivalence is the starting point of the 4th root trick used to reduce the number of fermion flavors and provides the only consistent description of 2 or 1 flavor systems. In this paper we argue that the quark mass of the underlying flavor symmetric theory differs from the staggered mass by an additive term due to the taste breaking of the staggered acti…
We calculate the 1-loop renormalization of the fermion self-energy, all local fermion bilinears, as well as a set of extended bilinears which form a basis corresponding to moments of the parton distribution functions. We use the overlap action for fermions and Symanzik improved action for gluons. Our results are presented as a function of the overlap parameter rho and the parameters entering the Symanzik action.
