The BdMove classes used by both the pscf_rpc and pscf_rpg programs implements a MC move that performs a short BD simulation, and always accepts the result. Unlike other MC moves, this move does not exactly satisfy detailed balance, and incurs the same sorts of systematic errors as those found in BD simulations.

The BdMove class is provided to allow users to mix BD and MC sampling strategies. For large systems, BD simulation is generally more efficient than MC simulation, at the cost of the existence of a systematic error that decreases with decreasing time step. By combining a BdMove with a one or more more specialized MC move, a user may combine the efficiency of BD simulation with specialized capabilities provided by another type of MC move.

Two closely analogous class templates named Pscf::Rp::BdMove and Pscf::Rpg::BdMove are used to perform this type of move in the pscf_rpc and pscf_rpg programs, respectively. Specializations of these templates are derived from specializations of the base class template Pscf::Rp::BdMove, which defines the implementation of the move in a generic form.

See also: Rp::BdMove (class API)

Step Algorithm

Each step of the BdMove algorithm runs a short BD simulation, using the Leimkuler-Matthews BD step algorithm that is also used by the LMBdStep BD step algorithm.

Parameter file

The only parameters needed by this algorithm are the "probability" parameter, which determines the probability that a move of this type will be attempted, a "mobility" parameter, which is the mobility parameter of the BD step algorithm, and parameter "nStep", which specifieds the number of BD steps that should be taken per MC move of this type. All three parameters are required. A typical example of an associated parameter file block is shown below.

BdMove{
  probability   0.5
  mobility      0.005
  nStep         100
}

The format for this block is thus:

BdMove{
  probability  float (in range [0,1])
  mobility     float
  nStep        int
}

The meanings of all parameters are also summarized briefly below:

Label	Description
probability	probability of attempting a step of this type
mobility	Mobility parameter of the underlying BD step algorith
nStep	number of BD steps per MC move of this type