Updating a Gaugefield

To update a Gaugefield simply use the update! function that takes 2 positional arguments and however many keyword arguments specific to the update algorithm. The first positional is the actual update algorithm update_alg and the second is the Gaugefield U.

U = Gaugefield(...)
random_gauges!(U)

MAXIT = 100
numHB = 1
or_alg = MeatQCD.Updates.Subgroups
numOR = 4
update_alg = MeatQCD.Updates.Heatbath(U, MAXIT, numHB, or_alg, numOR)

update!(update_alg, U; ...)

Supported Update Algorithms

MetaQCD.Updates.HMC — Type

HMC(
    U,
    integrator,
    trajectory,
    steps,
    friction = 0,
    numsmear = 0,
    ρ_stout = 0;
    hmc_logging = true,
    fermion_action = QuenchedFermionAction,
    heavy_flavours = 0,
    num_cv = 0,
    logdir = "",
)

Create an HMC object, that can be used as an update algorithm.

Arguments

U: The gauge field on which the update is performed.
integrator: The integrator used to evolve the field.
trajectory: The length of the HMC trajectory.
steps: The number of integrator steps within the trajectory.
friction: Friction factor in the GHMC algorithm. Has to be in the range [0, 1].
numsmear: Number of Stout smearing steps applied to the gauge action.
ρ_stout: Step length of the Stout smearing applied to the gauge action.
hmc_logging: If true, creates a logfile in logdir containing information

on the trajectories, unless logdir = ""

fermion_action: An AbstratFermionAction to initialize the appropriate fermion fields
heavy_flavours: The number of non-degenerate heavy flavours, again to initialize the

right number of fermion fields

num_cv: If bigger than 0, additional fields are initialized that are needed for Stout

force recursion when using a bias.

Supported Integrators

Leapfrog
OMF2
OMF2Slow
OMF4
OMF4Slow

Supported Fermion Actions

WilsonFermionAction
WilsonEOPreFermionAction
StaggeredFermionAction
StaggeredEOPreFermionAction

source

MetaQCD.Updates.Metropolis — Type

Metropolis(U::Gaugefield{B,T,A,GA}, eo, ϵ, numhits, target_acc, or_alg, numorelax) where {B,T,A,GA}

Create a Metropolis object.

Arguments

U::Gaugefield{B,T,A,GA}: Gauge field object.
eo: Even-odd preconditioning.
ϵ: Step size for the update.
numhits: Number of Metropolis hits.
target_acc: Target acceptance rate.
or_alg: Overrelaxation algorithm.
numorelax: Number of overrelaxation sweeps.

Returns

A Metropolis object with the specified parameters. The gauge action GA of the field U determines the iterator used. For the plaquette or Wilson action it uses a Checkerboard iterator and for rectangular actions it partitions the lattice into four sublattices.

source

MetaQCD.Updates.Heatbath — Type

Heatbath(U::Gaugefield{B,T,A,GA}, MAXIT, numheatbath, or_alg, numorelax) where {B,T,A,GA}

Create a Heatbath` object.

Arguments

U: The gauge field on which the update is performed.
MAXIT: The maximum iteration count in the Heatbath update.
numheatbath: The number of Heatbath sweeps.
or_alg: The overrelaxation algorithm used.
numorelax: The number of overrelaxation sweeps.

Returns

A Heatbath object with the specified parameters. The gauge action GA of the field U determines the iterator used. For the plaquette or Wilson action it uses a Checkerboard iterator and for rectangular actions it partitions the lattice into four sublattices.

source

MetaQCD.Updates.Overrelaxation — Type

Overrelaxation(algorithm)

Create an Overrelaxation object, that can be used within a Metropolis or Heatbath update step.

Supported Algorithms

"subgroups": Cabibbo-Marinari SU(2) subgroup embedding scheme
"kenney_laub": Kenney-Laub projection onto SU(3)

source