Activating the plugin

To use the WESTPA string method plugin, first activate it in the west.cfg file, as in the following example:

---
west:
    ...
    plugins:
        - plugin: westext.stringmethod.StringDriver
          string_method: system.PeriodicLinkedStringMethod
          avgpos_method: system.average_position
          dfunc_method: system.dfunc
          do_update: True
          windowsize: 100
          update_interval: 10
          initial_update: 10
          priority: 1

The plugin block specific to the string method, must contain each of the following elements:

Setting name (default value)	What does it do?
plugin	Defines that the string method should be used. This should be set to westext.stringmethod.StringDriver.
string_method	Should be set to default or a custom subclass of the WESTStringMethod class.
avgpos_method	Method defining how the average position in a bin is to be calculated. This should be set to cartesian or a custom method.
dfunc_method	Custom method defining how the VoronoiBinMapper should calculate the distance between a test point in progress coordinates space and all of the bin generators.
do_update (True)	Update the string during an iteration corresponding to the update_interval.
update_interval (10)	The number of iterations of the WE method between string updates.
initial_update (20)	First iteration to attempt string update.
windowsize (10)	The number of iterations previous to the current iteration that should be considered when calculating the average position of replicas within each bin.
init_from_data (True)	Use the last stored string definition to initialize the string. Otherwise use the definition in the WESTSystem.initialize method.
priority (0)	If multiple plugins are specified for the same callback hook, execute them in ascending order based on the priority.

For the parameters of the plugin listed above, it is often convenient to define any custom methods within the system.py file as is shown in the example.

String Method parameters

While the string method plugin defines a generic interface for performing a string calculation in WESTPA, most users will likely use the DefaultStringMethod (specified by string_method: default, or a custom subclasss of DefaultStringMethod. Regardless of the implementation, the plugin driver expects a that a python dict can be found that is a instance variable of your system class, called system.sm_params, as in the following example:

class System(WESTSystem):

    def initialize(self):

        ...
        self.sm_params = {'slen': [slen, slen],
                          'kappa': 0.001,
                          'dtau': 0.15,
                          'fixed_ends': False,
                          'sciflag': True,
                          'mpairs': [[0, self.nbins - 1], [slen - 1, slen]],
                          'slabels': [2],
                          'fourierflag': True,
                          'fourier_P': 2}

The parameters available via the DefaultStringMethod are

Parameter name (default value)	What does it do?
slen	An iterable containing the number of centers in each string.
slabels (None)	A list containing the relative positions in each string of any state label progress coordinates if present. These progress coordinates will be ignored in the calculation. None if no labels.
mpairs (None)	A list of lists containing the indices of pairs of centers that should move together. None if strings move independently.
dtau (0.1)	Parameter controlling the rate at which centers move toward the average value in the bin.
kappa (0.1)	Parameter controlling the smoothing of the string.
fixed_ends (True)	Boolean flag specifying whether to fix the ends of the strings.
sciflag (None)	Boolean flag specifying whether to attempt to use SciPy methods which are generally more efficient. If None import SciPy modules if available.
fourierflag (False)	Boolean flag specifying whether to user fourier fitting method
fourier_P (2)	Integer value specifying how many fourier modes to use in fitting
fourier_maxiters (100)	Maximum number of iterations of fourier fitting procedure
fourier_tol (1.0E-6)	Tolerance for ending fourier fitting

If the dfunc method requires additional positional or keyword arguments, the plugin will take them from system.dfargs and system.dfkwargs respectively when updating the BinMapper. For example, if the dfunc method required a set of weights to calculate a weighted RMSD and a boolean flag parameter, with the call signature dfunc(p, centers, weights, flag=False), the system initialization might look like:

class System(WESTSystem):

    def initialize(self):

        ...
        weights = ... # Define weight array
        df_flag = True

        self.dfargs = (weights, )
        self.dfkwargs = {'flag': df_flag}
        self.bin_mapper = VoronoiBinMapper(dfunc, centers, dfargs=self.dfargs, dfkwargs=self.dfkwargs)