2+1 flavor full QCD configurations by the PACS Collaboration

The 2+1 flavor full QCD configurations listed below are generated by on supercomputer

This project was supported by the HPCI strategic program Filed 5 "The origin of matter and the universe".


Simulation parameters

These gauge configurations are generated with the Iwasaki gauge action at β = 6/g2=1.82 and a non-perturbatively O(a) improved clover quark action with the APE stout smeared links with the smearing parametor ρ=0.1 and the number of smearing steps Nsmear = 6. The ud and s quark masses are chosen to be near the physical point where Mπ = 146 MeV, MK = 525. The lattice spacing of a = 0.0846 fm is fixed by experimental values of Mπ, MK, and MΩ. We have one ensemble on 964 lattice.

The ud quarks are simulated with domain-decomposed hybrid Monte Carlo (DDHMC) algorithm geometrically separating the ud quark determinant into the UV and IR parts after the even-odd preconditioning. Furthermore, the twofold mass preconditioning on the IR part is employed by splitting the IR force into FIR, F'IR and F''IR. This decomposition is controlled by two additional hopping parameters K'ud= ρ1 Kud and K''ud= ρ1 ρ2 Kud. FIR is derived from the action preconditioned with K'ud. The ratio of two preconditioners with K'ud and K''ud gives F'IR. F''IR is from the heaviest preconditioners with K''ud. The multiple time scale integration scheme is adopted in the molecular dynamics steps. For the gauge force Fg, the UV force FUV and the three IR forces F''IR, F'IR and FIR, the number of steps for one trajectory (N0, N1, N2, N3, N4) is chosen for the associated step sizes: δτg=τ/N0N1N2N3N4, δτUV=τ/N1N2N3N4, δτ''IR=τ/N2N3N4, δτ'IR=τ/N3N4, δτIR=τ/N4, where τ is a trajectory length.

The s quark is simulated with UV-filtered polynomial hybrid Monte Carlo (UVPHMC) algorithm with a polynomial order Npoly where the action is UV-filtered after the even-odd preconditioning without domain decomposition. The step size is set as δτs=δτ''IR.

The table below shows the values of a, La and the simulation parameters.

ensemble # β cSW Kud
Ks 		(ρ, Nsmear) lattice
size 		a [fm] La [fm] τ (N0, N1, N2, N3, N4) (ρ1, ρ2) Npoly
1 1.82 1.11 0.126117
0.124790 		(0.1, 6) 964 0.0846(7) 8.12(6) 1.0 (8,2,2,2,15) (0.99975, 0.99400) 310

Configuration ensembles

The table below gives a list of ensembles of configurations. A consecutive set of DDHMC+UVPHMC trajectories for a given value of β, Kud, Ks, cSW on a particular lattice size is called an ensemble. A series name (h and i) is given to distinguish multiple DDHMC+UVPHMC runs.

Note that the ud and s quarks couple to the smeared link variables produced by the APE stout smearing method with ρ=0.1, Nsmear=6.

ensemble # β Kud Ks cSW lattice
size 		series traj. range missing
traj		 stored at
every
(traj)		 #config total
#config		 size/config
[GB]		 total
[TB]
1 1.82 1.11 0.126117 0.124790 964 h
i		 000960-002050
000610-001500		 --
--		 10
10		 110
90		 200 48.9229.784

Naming of Ensembles and Configurations (for ILDG users)

An ensemble is identified by markovChainURI named e.g. as A configuration is labeled by dataLFN named e.g. as Basename of the dataLFN is formmated as
RgClover spatial size x temporal size _ \beta\kappa_ud\kappa_s\c_sw - series - trajectory number

To get full list of markovChainURIs and LFNs, please use QCDml faceted navigation.

Location of Configurations (for JLDG users)

Configurations are placed in the gfarm directory
  /gfarm/public/ILDG/JLDG/PACS-Collab/RFCNF2p1/
and classified in terms of β and Kud, Ks. For example, the directory
  /gfarm/public/ILDG/JLDG/PACS-Collab/RFCNF2p1/RC96x96_B1820/Kud01261170Ks01247900/
contains configurations at β=1.820 and Kud=0.1261170, Ks=0.1247900. Filename is the same as basename of dataLFN. All ensemble and configuration XML files can be found in
  /gfarm/public/ILDG/JLDG/PACS-Collab/RFCNF2p1/QCDml/

Last update: 26 July 2022, H.Ohno