Action: PATH_LOGMEANEXP

Module	pathbias
Description	Usage
Compute a stable log arithmetic mean across synchronized replicas.
output value	type
the dimensionless log arithmetic mean of replica ratios	scalar

Details and examples

Compute a stable log arithmetic mean across synchronized replicas.

ARG is one finite, nonperiodic dimensionless scalar ell_b on each replica. The result is log(sum_b exp(ell_b)/P). Its local derivative is exp(ell_b)/sum_c exp(ell_c), with no additional factor of 1/P. Only replica roots communicate across replicas; spatial MPI ranks are not extra beads. EXPECTED_REPLICAS is required to catch a missing or incorrect replica group. All replicas must execute the same graph at the same step and time. Numerical derivatives at the action level are unsupported: finite-difference tests must perturb one replica's input at a time.

For a common frozen energy field v(s), use ell=-v/kBT and apply -kBT times this result as a complete-path bias. This averages the effective target to unbiased probability ratio, not the density estimate or its logarithm.

Click on the labels of the actions for more information on what each action computes

#SETTINGS NREPLICAS=2
dThe DISTANCE action with label d calculates the following quantities:
 Quantity   
 Type   
 Description  
d
scalar
the DISTANCE between this pair of atoms: DISTANCECalculate the distance/s between pairs of atoms. More details ATOMSthe pair of atom that we are calculating the distance between=1,2 NOPBC ignore the periodic boundary conditions when calculating distances
vThe CUSTOM action with label v calculates the following quantities:
 Quantity   
 Type   
 Description  
v
scalar
an arbitrary function: CUSTOMCalculate a combination of variables using a custom expression. More details ARGthe values input to this function=d FUNCthe function you wish to evaluate=x*x PERIODICif the output of your function is periodic then you should specify the periodicity of the function=NO
ellThe CUSTOM action with label ell calculates the following quantities:
 Quantity   
 Type   
 Description  
ell
scalar
an arbitrary function: CUSTOMCalculate a combination of variables using a custom expression. More details ARGthe values input to this function=v FUNCthe function you wish to evaluate=-x/2.5 PERIODICif the output of your function is periodic then you should specify the periodicity of the function=NO
lmThe PATH_LOGMEANEXP action with label lm calculates the following quantities:
 Quantity   
 Type   
 Description  
lm
scalar
the dimensionless log arithmetic mean of replica ratios: PATH_LOGMEANEXPCompute a stable log arithmetic mean across synchronized replicas. More details ARGthe labels of the values from which the function is calculated=ell EXPECTED_REPLICASnumber of synchronized replicas in this path=2
energyThe CUSTOM action with label energy calculates the following quantities:
 Quantity   
 Type   
 Description  
energy
scalar
an arbitrary function: CUSTOMCalculate a combination of variables using a custom expression. More details ARGthe values input to this function=lm FUNCthe function you wish to evaluate=-2.5*x PERIODICif the output of your function is periodic then you should specify the periodicity of the function=NO
biasThe BIASVALUE action with label bias calculates the following quantities:
 Quantity   
 Type   
 Description  
bias.bias
scalar
the instantaneous value of the bias potential
bias.energy_bias
scalar
one or multiple instances of this quantity can be referenced elsewhere in the input file. these quantities will named with  the arguments of the bias followed by the character string _bias. These quantities tell the user how much the bias is due to each of the colvars. This particular component measures this quantity for the input CV named energy: BIASVALUETakes the value of one variable and use it as a bias More details ARGthe labels of the scalar/vector arguments whose values will be used as a bias on the system=energy

Verify identical field parameters/state hashes before a run or restart. This action has no bias history and does not make adaptive OPES learning compatible with this Hamiltonian. If v is an active frozen bias action, apply the correction V_path-v instead of adding V_path on top of it. The molecular dynamics adapter owns dynamical force scaling and single-path energy accounting. Uniform bead observables still use one path weight exp(V_path/kBT); the local derivative weights are not observable weights.

Input

The arguments that serve as the input for this action are specified using one or more of the keywords in the following table.

Keyword	Type	Description
ARG	scalar	the labels of the values from which the function is calculated

Full list of keywords

The following table describes the keywords and options that can be used with this action

Keyword	Type	Default	Description
ARG	input	none	the labels of the values from which the function is calculated
EXPECTED_REPLICAS	compulsory	none	number of synchronized replicas in this path
NUMERICAL_DERIVATIVESThis keyword do not have examples	optional	false	calculate the derivatives for these quantities numerically