Edit on GitHub

DQMC

The DQMC struct represents both the determinant quantum monte carlo algorithm and the simulation as a whole. Because of that it includes a bunch of options that aren't directly relevant to the algorithm. The minimal default is given by dqmc = DQMC(model, beta=beta) and the simulation can then be started with run!(dqmc). Additional keyword arguments include:

  • beta: The inverse temperature of the simulation.
  • delta_tau = 0.1: The imaginary time discretization.
  • slices = beta / delta_tau: The number of imaginary time slices.
  • safe_mult = 10: The number of save matrix multiplications.
  • thermalization = 100: The number of thermalization sweeps.
  • sweeps = 100: The number of measurement sweeps.
  • check_sign_problem = true: Enables or disables checks for sign problems. (negative or imaginary probabilities)
  • check_propagation_error = true: Enables or disables checks for time slices propagation errors. (Which may happen if safemult or deltatau is too large.)
  • silent = false: Enable or disable prints for the checks above.
  • measure_rate = 10: Sets the frequency of measurements. Every measure_rate sweeps a new measurement is taken.
  • print_rate = 10: Sets the frequency of general information prints (not checks).
  • seed = -1: Sets a random seed for the simulation. If set to -1 the seed will be chosen randomly.
  • last_sweep = 0: Sets the last finished sweep. Used internally for continued simulations.

Beyond this there are a couple of keyword arguments which are more involved and will be discussed in other chapters. These include

  • scheduler = SimpleScheduler(LocalSweep()): This sets up the sequence of updates performed by the simulation.
  • measurements = Dict{Symbol, AbstractMeasurement}(): A collection of named measurements that run during the simulation. These are usually added after creating the simulation.
  • thermalization_measurements = Dict{Symbol, AbstractMeasurement}(): Same as the above, but the measurements run during the thermalization stage. Might be useful to judge convergence or the number of necessary sweeps.
  • recorder = ConfigRecorder: A recorder for the configurations generated during the simulation.
  • recording_rate = measure_rate: The rate at which configurations are recorded.
  • field::Type = choose_field(model): The type of the field used by the simulation

Running a simulation also comes with a bunch of options via keyword arguments - most dealing with saving the simulation. The options for run(dqmc) include:

  • verbose = true: Enables or disables information printing during the runtime.
  • ignore = tuple(): Measurement keys to ignore during this run (none by default).
  • safe_before::TimeType = now() + Year(100): Sets a time stamp before which the simulation will cancel itself and save.
  • safe_every::TimePeriod = Hour(10000): Sets a time period interval for regular saves.
  • grace_period::TimePeriod = Minute(5): Sets a buffer time period for saving. By default saves at least 5 minutes before the requested time.
  • resumable_filename = "resumable_$datestring.jld2: Sets the name the savefile generated from safebefore and safeevery.
  • overwrite = true: Enables or disables overwriting of existing files.
  • min_update_rate = 0.001: Sets a minimum update acceptance rate under which a simulation is considered "failed". This will only trigger after 1 / min_update_rate sweeps and only if all updates fall below this rate.
  • fail_filename = "failed_$datestring.jld2": Sets the filename of a failed simulation.