VMC is used to generate configurations for optimizations and following DMC sections.  sample parameter is used to set the total walker configurations stored during a VMC run.

VMC Optimization

One way to set the number of samples for an optimization is as follows,

• blocks = number of lines in the scalar.dat file.
• stepsbetweensamples = how often to dump configurations.
• walkersperthread = walkers per thread (defaults to 1).
• samples = number of samples to optimize on, this is across all nodes and threads.

VMC to DMC Run

One way to set the parameters for a VMC run leading up to a DMC run is

<qmc method="vmc">
  <parameter name="blocks">50</parameter>
  <parameter name="steps">15000</parameter>
  <parameter name="samples">10000</parameter>
  <-- other parameters -->
</qmc>

• steps = number of steps each block
• blocks = number of times to repeat steps (total steps for each walker is blocks x steps).
• walkers = number of walkers on each node. default is one walker per thread.
• samples = target number of walkers for the following DMC run. This is a global quantity across all nodes and threads.

Another way to set this is,

Other useful parameters

Some other things you might find useful to put in your VMC and DMC blocks.

VMC

• useDrift = turn drift on or off in your VMC run.

DMC

• MaxAge = the maximum number of times a walker can have a rejected move without reducing the probability to survive to the next step.
• nonlocalmoves = T-moves of Casula. default is no.
• fastgrad = default is yes. Might choose no for debugging?
• MaxCopy = maximum number of children a walker can spawn at each step.
• feedback = feedback parameter for adjusting trial energy. default = 1 (adjusted for tau).