- Use
exe_T.shto generate amonte.jsonand execute Monte Carlo - Run Monte Carlo:
casm monte -s monte.json - How to choose parametric chemical potential: do a quick scan of chemical potential, and make sure it covers the whole compositional space: e.g. -10 eV to 10 eV.
- Parametric chemical potential:
casm composition -d - How to understand chemical potential : energy decrease on the total number of specie N, i.e.,
- T down scan: check if the Monte Carlo can reproduce your convex hull using your cluster expansion model. You can start from the temperature close to the melting point down to a low temperature such as 10 K. Temperature step can be chosen as 5 or 10 K.
Output of Monte Carlo
-
What we want in output: Potential energy, order parameters such as and site fractions etc.
-
Ref: https://iopscience.iop.org/article/10.1088/0965-0393/10/5/304
-
ref:
casm format --monte
### monte ##################
LOCATION WHEN GENERATED:
User determined
DESCRIPTION:
The Monte Carlo input file does not need to be in any particular
location, as long as it is somewhere inside the CASM project
directory or subdirectories. The input file contains a JSON
object with "ensemble", "method", "model", "supercell",
"data", and "driver" attributes, as described below. An
optional attribute "debug" may also be included to print
information that may be useful for debugging an input file.
Input file parameters:
"ensemble" (string):
Possible options for "ensemble" are:
"GrandCanonical" or "grand_canonical": Semi-grand canonical
Monte Carlo calculation in which the total number of sites is
fixed, but the occupants on each site may vary. One occupant
change at a time is attempted.
"Canonical" or "canonical": Canonical Monte Carlo
calculation in which the total number of each type of occupant
is fixed. Each Monte Carlo step attempts to swap a pair of
occupants.
"method" (string):
Possible options for "method" are:
"Metropolis" or "metropolis": Run Monte Carlo calculations
using the Metropolis algorithm.
"LTE1" or "lte1": Single spin flip low temperature
expansion calculations.
"model": (JSON object)
/"formation_energy": (string, optional, default="formation_energy")
Specifies the cluster expansion to use to calculated formation
energy. Should be one of the ones listed by 'casm settings -l'.
"supercell": (3x3 JSON arrays of integers)
The supercell transformation matrix.
"data": (JSON object)
/"sample_by": (string)
Specify unit for the period between samples. May be either
"step" (sample after every "sample_period" proposed Monte
Carlo events), or "pass" (sample after the "sample_period"
number of passes), where 1 pass is a number of steps equal to
the number of sites in the supercell that have variable
occupation).
/"sample_period": (integer)
Specify how many steps or passes to wait between data samples.
/"measurements": (JSON array containing JSON objects)
Specifies which properties to sample. Each JSON object should
include "quantity" (string) specifying a property to be
sampled. Optionally, it may also include "precision" (number),
indicating the required (absolute) precision in the average of
the quantity for the calculation to be considered converged. If
a precision is given for any quantity, then the Monte Carlo
calculations run in automatic convergence mode and continue
until all quantities with a specified precision are converged
to level requested.
Possible options for "quantity" are:
"comp": composition, relative the composition axes
"comp_n": composition, number of atoms per unit cell
"site_frac": composition, normalized per basis site
"atom_frac": composition, normalized per total number of atoms
"formation_energy": formation energy (per unit cell)
"potential_energy": potential energy (per unit cell),
(= formation_energy - sum_i(mu_i*comp_i))
"non_zero_eci_correlations": correlations (per unit cell)
which have non-zero eci values.
"all_correlations": correlations (per unit cell)
"<anything else>": is interpreted as a 'casm query' query
/"confidence": (number, range (0.0, 1.0), default 0.95)
The confidence level used for calculating the precision in the
average value of sampled quantities.
/"min_pass", /"min_step", /"min_sample": (integer)
If in automatic convergence mode, prevents the calculation from
a minimum number of passes, steps, or samples have occurred.
/"max_pass", /"max_step", /"max_sample": (integer)
If in automatic convergence mode, stops the calculation if the
specified number of passes, steps, or samples have occurred.
/"N_pass", /"N_step", /"N_sample": (integer)
When not in automatic convergence mode (no precision has been
specified for any quantities being sampled), stops the
calculation when the specified number of passes, steps, or
samples have occurred.
/"equilibration_passes_first_run": (integer)
If included, the requested number of passes will be performed
at the initial conditions as a preliminary step before the
actual run begins. This may be useful when not running in
automatic convergence mode.
/"equilibration_passes_each_run": (integer)
If included, the requested number of passes will be performed
at each condition as a preliminary step before the actual run
begins. This may be useful when not running in automatic
convergence mode.
/"storage": (JSON object) Options for writing results.
/"output_format": (string or JSON array of string)
Specifies the type or types of output files. Current options
are "csv" or "json". Type names with either all lower
case or all upper case are accepted.
/"write_observations": (boolean, default false)
If true, all individual observations of the quantities
requested to be sampled will be written to compressed files:
"output_directory"/conditions.i/observations.ext.gz
where 'i' is the condition index and 'ext' is the output
format.
/"write_trajectory": (boolean, default false)
If true, the value of all degrees of freedom at the time of
each sample will be written to compressed files:
"output_directory"/conditions.i/trajectory.ext.gz
where 'i' is the condition index and 'ext' is the output
format.
/"enumeration": (JSON object, optional)
If included, save configurations encountered during Monte
Carlo calculations by keeping a 'hall of fame' of best scoring
configurations. After the calculation at a particular set of
thermodynamic conditions completes, the configurations in the
hall of fame are saved to the project configuration list.
/"check": (string, default="eq(1,1)")
A 'casm query'-like string that returns a boolean value
indicating if (true) a configuration should be considered for
for the enumeration hall of fame. The default always returns
true.
/"metric": (string, default="clex_hull_dist(ALL)")
A 'casm query'-like string that provides a metric for ranking
ranking configurations as they are encountered during a Monte
Carlo calculation. The resulting value is used to create a
hall of fame of 'best' configurations encountered during the
calculation. When the calculation is complete configurations
in the hall of fame are added to the [[CASM]] project config
list. The 'casm query'-like command should evaluate to a
number.
Besides the properties listed via 'casm query -h properties',
and 'casm query -h operators', both "check" and "metric"
can also use the property "potential_energy".
/"sample_mode": (string, optional, default="on_sample")
Indicate when to attempt to insert configurations into the
enumeration hall of fame. Options are:
"on_accept": after each accepted Monte Carlo event
"on_sample": after each data sample
/"check_existence": (bool, optional, default=true)
If true, only configurations that do not already exist in the
config list are inserted into the enumeration hall of fame.
/"insert_canonical": (bool, optional, default=true)
If true, configurations are inserted into the enumeration
hall of fame in their canonical form. If 'check_existence' is
true, this must be set to true.
/"N_halloffame": (integer, optional, default=100)
The number of configurations that are allowed in the
enumeration hall of fame.
/"tolerance": (number, optional, default=1e-8)
Tolerance used for floating point comparison of configuration
scores in the enumeration hall of fame.
"driver": (JSON object)
/"motif": (JSON object)
Specifies the initial occupation of the supercell.
For canonical ensemble Monte Carlo calculations an additional
step changes the occupants on random sites to make the actual
composition as close as possible to the requested composition.
/"configname": (string, optional)
A configuration name, "auto", "restricted_auto", or
"default".
Specifies the configuration that is tiled to fill the
supercell. If necessary, symmetry operations may be applied
An error is thrown if the specified configuration can not be
used to fill the "supercell".
Possible options for "configname" are:
A configuration name (ex. "SCEL3_3_1_1_0_2_2/0")
"auto": ("grand_canonical" ensemble only) Enumerated
configurations will be searched for the configuration with
the lowest potential energy to use as the motif.
"default": If the value "default" is used, the initial
motif occupation is determined from the occupation order in
the PRIM.
"restricted_auto": ("grand_canonical" ensemble only)
Same as "auto", but only configurations that can tile the
supercell are considered. As a last resort, "default" is
used.
/"configdof": (string, optional)
Specifies the path to a configdof JSON file, such as
"initial_state.json" or "final_state.json", containing
the degrees of freedom to initialize the supercell with
/"mode": (string)
Specify the drive mode.
Possible options for "mode" are:
"incremental": perform one or more calculations, starting
at the initial conditions and incrementing by the
incremental conditions up to (and including) the final
conditions.
"custom": perform one or more calculations, as specified by
the "custom_conditions".
/"dependent_runs": (boolean, default true)
If true, begin the next calculation with the final DoF from the
previous calculation. If false, begin each calculation with the
DoF specified for the "motif".
/"initial_conditions",
/"incremental_conditions",
/"final_conditions": (JSON object, optional)
Specifies the applied conditions for the calculation. For
"incremental_conditions", specifies the change in conditions
between individual calculations. Each JSON object should
include:
/"temperature": (number)
The temperature in K.
/"param_chem_pot" (JSON object, "grand_canonical" ensemble only)
The parametric chemical potential(s)
/"a", /"b", ...: (number)
The parametric chemical potentials conjugate to the parametric
compositions. The number of parametric chemical potentials
provided must match the number of independent compositions.
/"comp" (JSON object, "canonical" ensemble only, option 1)
The parametric composition(s)
/"a", /"b", ...: (number)
The parametric composition. The number of entries provided
must match the number of independent compositions.
/"comp" (JSON array, "canonical" ensemble only, option 2)
The parametric composition(s)
[number, ...]
An array containing the parametric composition. The number of
entries provided must match the number of independent
compositions.
/"comp_n" (JSON object, "canonical" ensemble only, option 3)
The mol composition per unitcell
/"A", /"B", ...: (number)
The mol composition per unitcell. The entries provided must
match occupant names in the 'prim.json' file. The values are
summed, normalized, and then converted to parametric composition.
/"comp_n" (JSON array, "canonical" ensemble only, option 4)
The mol composition per unitcell
[number, ...]
An array containing the mol composition per unitcell. The
number of entries provided must match the number components.
The values are summed, normalized, and converted to parametric
composition.
/"tolerance": (number)
Specifies a numerical tolerance for comparing conditions.
/"custom_conditions":
(JSON array of JSON objects) An array specifying a custom
path of conditions.
Restarts: Metropolis Monte Carlo calculations that are stopped
before the entire path has been calculated can be restarted as
long as the conditions of the existing calculations agree with
the conditions specified in the input settings. This means that
the "final_conditions" might be changed to increase the length
of a path, or additional "custom_conditions" might be added,
but the "incremental_conditions" may not be changed. Upon
restart, the results summary file is checked for the last
finished conditions. Then the path is resumed from the next set
of conditions. It is the responsibility of the user to ensure
that other important settings, such as the "model" are not
changed inappropriately.
"debug" (bool, default false):
If true, will print as much information as possible to assist in
debugging input file settings.
EXAMPLE: Settings for an incremental Metropolis calculation
with increasing temperature in automatic convergence mode.
-------
{
"comment" : "This is a sample input file. Unrecognized attributes (like the ones prepended with '_' are ignored.",
"debug" : false,
"ensemble" : "grand_canonical",
"method" : "metropolis",
"model" : {
"formation_energy" : "formation_energy"
},
"supercell" : [
[10, 0, 0],
[0, 10, 0],
[0, 0, 10]
],
"data" : {
"sample_by" : "pass",
"sample_period" : 1,
"_N_sample" : 1000,
"_N_pass" : 1000,
"_N_step" : 1000,
"_max_pass" : 10000,
"min_pass" : 1000,
"_max_step" : 10000,
"_max_sample" : 500,
"_min_sample" : 100,
"confidence" : 0.95,
"measurements" : [
{
"quantity" : "formation_energy"
},
{
"quantity" : "potential_energy"
},
{
"quantity" : "atom_frac"
},
{
"quantity" : "site_frac"
},
{
"quantity" : "comp",
"precision" : 1e-3
},
{
"quantity" : "comp_n"
}
],
"storage" : {
"write_observations" : false,
"write_trajectory" : false,
"output_format" : ["csv", "json"]
}
},
"driver" : {
"mode" : "incremental",
"motif" : {
"configname" : "auto",
"_configname" : "SCEL3_3_1_1_0_2_2/0",
"_configdof" : "path/to/final_state.json"
},
"initial_conditions" : {
"param_chem_pot" : {
"a" : -1.75
},
"temperature" : 100.0,
"tolerance" : 0.001
},
"final_conditions" : {
"param_chem_pot" : {
"a" : -1.75
},
"temperature" : 1000.0,
"tolerance" : 0.001
},
"incremental_conditions" : {
"param_chem_pot" : {
"a" : 0.0
},
"temperature" : 10.0,
"tolerance" : 0.001
}
}
}
-------
EXAMPLE: Settings for an custom drive mode LTE1 calculation with
increasing temperature.
-------
{
"comment" : "This is a sample input file. Unrecognized attributes (like the ones prepended with '_' are ignored.",
"debug" : false,
"ensemble" : "grand_canonical",
"method" : "lte1",
"model" : {
"formation_energy" : "formation_energy"
},
"supercell" : [
[9, 0, 0],
[0, 9, 0],
[0, 0, 9]
],
"data" : {
"storage" : {
"write_observations" : false,
"write_trajectory" : false,
"output_format" : ["csv", "json"]
}
},
"driver" : {
"mode" : "incremental",
"motif" : {
"configname" : "auto",
"_configname" : "SCEL3_3_1_1_0_2_2/0",
"_configdof" : "path/to/final_state.json"
},
"custom_conditions" : [
{
"param_chem_pot" : {
"a" : 0.0
},
"temperature" : 100.0,
"tolerance" : 0.001
},
{
"param_chem_pot" : {
"a" : 0.0
},
"temperature" : 200.0,
"tolerance" : 0.001
},
{
"param_chem_pot" : {
"a" : 0.0
},
"temperature" : 400.0,
"tolerance" : 0.001
},
{
"param_chem_pot" : {
"a" : 0.0
},
"temperature" : 800.0,
"tolerance" : 0.001
}
]
}
}
-------