STM:Server Controls

Navigation: Documentation / Methods / String Method / STM:Server / STM:Server Controls

---

Navigation: Documentation / Methods / String Method / STM:Server / STM:Path Controls

---

The STM path setup controls the global string path handled by the STM server. The setup defines input and output files, STM iteration periods, path update intervals, termination criteria, and optimisation-method-specific parameters.

[files]

Key	Default value	Type	Description
input	"{PATHS}"	string	Name of the input path file. The file defines the initial STM path.
output	"_stm.path"	string	Name of the output path file. The current or final STM path is written to this file.
trajectory	"_stm.traj"	string	Name of the path trajectory file. It stores the time evolution of the STM path if trajectory output is enabled.
summary	"_stm.results"	string	Name of the output summary file containing final STM results.
optlog	"_stm.log"	string	Name of the optimisation log file.

[stm]

Key	Default value	Type	Description
steps	100	integer	Maximum number of STM optimisation steps.
optmethod	adabelif	string	Optimisation method used to update the STM path. Allowed values are: gd, ngd, ngd-auto, adam, adabelif, amsgrad, amsgradbc. The value determines which optimiser-specific section is used.
shift2zero	true	logical	If enabled, the global minimum of the free energy along the path is shifted to zero.
init	10000	integer	Length of the initialisation period, expressed in MD steps.
soloinit	true	logical	If enabled, the initialisation period is performed as a separate stage before the regular STM cycle.
accu	5000	integer	Length of the accumulation period, expressed in MD steps.
equi	1000	integer	Length of the equilibration period, expressed in MD steps.
soloequi	true	logical	If enabled, the equilibration period is performed as a separate stage before accumulation.
prod	50000	integer	Length of the final production period, expressed in MD steps.
sfac	0.0	real	Path smoothing factor.
async	false	logical	Enables asynchronous STM updates. If disabled, the path is updated synchronously.

[termination]

Key	Default value	Type	Description
fplch	0.001	real	Termination threshold for the final relative change of the path length.
maxbmove	0.005	real	Termination threshold for the maximum bead displacement.
avebmove	0.005	real	Termination threshold for the average bead displacement.
maxppmf	10.00	real	Termination threshold for the maximum perpendicular mean force.
aveppmf	2.00	real	Termination threshold for the average perpendicular mean force.
tbuflen	3	integer	Length of the moving-average buffer used to evaluate termination criteria.

[intervals]

Key	Default value	Type	Description
reparam	1	integer	Interval, in STM steps, for path reparametrisation.
smooth	0	integer	Interval, in STM steps, for path smoothing. A value of zero disables periodic smoothing.
trajectory	0	integer	Interval, in STM steps, for writing the path trajectory. A value of zero disables trajectory output.
output	100	integer	Interval, in STM steps, for writing the updated path to the output file.

[gd]

Key	Default value	Type	Description
stepsize	0.003	real	Step size used by the gradient descent optimiser.

[ngd]

Key	Default value	Type	Description
stepsize	0.003	real	Step size used by the normalised gradient descent optimiser.
mingnormeps	1e-7	real	Minimum gradient-norm threshold used to regularise the normalised gradient.
maxgnormforgd	5.0	real	Maximum gradient norm for switching to ordinary gradient descent. This key is used only by the ngd-auto optimisation method.

[adam]

Key	Default value	Type	Description
stepsize	0.003	real	Step size used by the Adam optimiser.
beta1	0.7	real	Exponential decay factor for the first-moment estimate.
beta2	0.99	real	Exponential decay factor for the second-moment estimate.
mingnormeps	1e-7	real	Minimum gradient-norm threshold used to regularise the optimisation step.
maxresets	0	integer	Maximum number of optimiser resets. A value of zero disables this limit.
memsteps	0	integer	Number of steps after which the optimiser memory is reset. A value of zero disables periodic memory resets.

[adabelif]

Key	Default value	Type	Description
stepsize	0.003	real	Step size used by the AdaBelief optimiser.
beta1	0.7	real	Exponential decay factor for the first-moment estimate.
beta2	0.99	real	Exponential decay factor for the second-moment estimate.
mingnormeps	1e-7	real	Minimum gradient-norm threshold used to regularise the optimisation step.
maxresets	0	integer	Maximum number of optimiser resets. A value of zero disables this limit.
memsteps	0	integer	Number of steps after which the optimiser memory is reset. A value of zero disables periodic memory resets.

[amsgrad]

Key	Default value	Type	Description
stepsize	0.003	real	Step size used by the AMSGrad optimiser.
beta1	0.7	real	Exponential decay factor for the first-moment estimate.
beta2	0.99	real	Exponential decay factor for the second-moment estimate.
mingnormeps	1e-7	real	Minimum gradient-norm threshold used to regularise the optimisation step.
maxresets	0	integer	Maximum number of optimiser resets. A value of zero disables this limit.
memsteps	0	integer	Number of steps after which the optimiser memory is reset. A value of zero disables periodic memory resets.

[amsgradbc]

Key	Default value	Type	Description
stepsize	0.003	real	Step size used by the bias-corrected AMSGrad optimiser.
beta1	0.7	real	Exponential decay factor for the first-moment estimate.
beta2	0.99	real	Exponential decay factor for the second-moment estimate.
mingnormeps	1e-7	real	Minimum gradient-norm threshold used to regularise the optimisation step.
maxresets	0	integer	Maximum number of optimiser resets. A value of zero disables this limit.
memsteps	0	integer	Number of steps after which the optimiser memory is reset. A value of zero disables periodic memory resets.