Files

The initial conditions file, initial.dat

There are two ways to set the initial conditions values. The default option is to generated the initial conditions randomly at the beginning of the code. For that, prior values can be set in the chem_input.dat file.

Otherwise, the initial conditions can be set in an initial conditions file, where the file name must be set in the param.dat file.

The file must contain initial values for every used variable in the Equations.py file. The file can contain empty lines and comment lined, starting with a #

The units must agree with the implementation in the Equations.py file.

Example

The default initial conditions file looks like:

# phase components fractions

MgO_silicate = 2.12e-1
SiO2_silicate = 9.80e-2
MgSiO3_silicate = 5.37e-1
FeO_silicate = 3.42e-2
FeSiO3_silicate = 1.66e-2
Na2O_silicate = 1.66e-6
Na2SiO3_silicate = 6.27e-3
H2_silicate = 3.70e-2
H2O_silicate = 5.96e-2
CO_silicate = 5.92e-9
CO2_silicate = 2.20e-11
Fe_metal = 6.05e-1
Si_metal = 6.31e-2
O_metal = 1.99e-2
H_metal = 3.12e-1
H2_gas = 9.92e-1
CO_gas = 1.34e-6
CO2_gas = 1.57e-9
CH4_gas = 1.29e-3
O2_gas = 9.39e-15
H2O_gas = 6.55e-3
Fe_gas = 7.33e-7
Mg_gas = 2.41e-6
SiO_gas = 2.94e-6
Na_gas = 1.94e-5
SiH4_gas = 1.20e-4

# Other variables

Moles_atm = 2.32e3
Moles_silicate = 2.55e3
Moles_metal = 2.73e3

The chem_input.dat file

This file contains information for the used variables in the chemical network. The file can contain empty lines and comment lines, starting with a #

The units must agree with the implementation in the Equations.py file. The file contains several blocks of data:

Number of moles of elements

The next block contains the number of moles from all used elements set in section 2 of the Equations.py file.

Other parameters

This block contains all the other variable set in Section 2 of the Equations.py file

In the default version, these are:

Mplanet_Mearth: Mass of the planet in units of Earth masses.

T_AMOI = 4000: Surface temperature, in Kelvin.

T_SME = 4500: Temperature at the core-mantle-boundary, in Kelvin.

Pstd: value of Standard pressure in bar (default = 1 bar).

Boundaries

The next block contains the boundaries for all used species in the Equations.py file. The values must be positive.

Priors

The last block contains priors for all used species in the Equations.py file. The priors are only used it not initial conditions file is given, and when the initial conditions are generated randomly.

Example

The default chem_input.dat file looks like:

# Number of moles of elements

nSi = 1802.9040997359934
nMg = 1838.7303465715897
nO = 5854.040012483196
nFe = 1704.0538951977085
nH = 355.4836555326242
nNa = 1e-3
nC = 0.1164707804247013


# Other parameters

Mplanet_Mearth = 0.50499144
T_AMOI = 4000
T_SME = 4500
Pstd = 1.0


# Boundaries

bound_MgO_silicate = 1.0e-30, 0.9999999999
bound_SiO2_silicate = 1.0e-30, 0.9999999999
bound_MgSiO3_silicate = 1.0e-30, 0.9999999999
bound_FeO_silicate = 1.0e-30, 0.9999999999
bound_FeSiO3_silicate = 1.0e-30, 0.9999999999
bound_Na2O_silicate = 1.0e-30, 0.9999999999
bound_Na2SiO3_silicate = 1.0e-30, 0.9999999999
bound_H2_silicate = 1.0e-30, 0.9999999999
bound_H2O_silicate = 1.0e-30, 0.9999999999
bound_CO_silicate = 1.0e-30, 0.9999999999
bound_CO2_silicate = 1.0e-30, 0.9999999999
bound_Fe_metal = 1.0e-30, 0.9999999999
bound_Si_metal = 1.0e-30, 0.9999999999
bound_O_metal = 1.0e-30, 0.9999999999
bound_H_metal = 1.0e-30, 0.9999999999
bound_H2_gas = 1.0e-30, 0.9999999999
bound_CO_gas = 1.0e-30, 0.9999999999
bound_CO2_gas = 1.0e-30, 0.9999999999
bound_CH4_gas = 1.0e-30, 0.9999999999
bound_O2_gas = 1.0e-30, 0.9999999999
bound_H2O_gas = 1.0e-30, 0.9999999999
bound_Fe_gas = 1.0e-30, 0.9999999999
bound_Mg_gas = 1.0e-30, 0.9999999999
bound_SiO_gas = 1.0e-30, 0.9999999999
bound_Na_gas = 1.0e-30, 0.9999999999
bound_SiH4_gas = 1.0e-30, 0.9999999999
bound_Moles_atm = 1.0e-30, 100000.0
bound_Moles_silicate = 1.0e-30, 100000.0
bound_Moles_metal = 1.0e-30, 100000.0


# Priors

prior_MgO_silicate = 0.1, 0.3
prior_SiO2_silicate = 0.0, 1.0
prior_MgSiO3_silicate = 0.0, 1.0
prior_FeO_silicate = 0.0, 1.0
prior_FeSiO3_silicate = 0.0, 1.0
prior_Na2O_silicate = 1e-5, 1e-3
prior_Na2SiO3_silicate = 1e-5, 1e-3
prior_H2_silicate = 0.0, 1.0
prior_H2O_silicate = 0.0, 1.0
prior_CO_silicate = 0.0, 1.0
prior_CO2_silicate = 1e-5, 1e-3
prior_Fe_metal = 0.0, 1.0
prior_Si_metal = 0.0, 1.0
prior_O_metal = 0.0, 1.0
prior_H_metal = 0.0, 1.0
prior_H2_gas = 0.0, 1.0
prior_CO_gas = 1e-3, 1e-2
prior_CO2_gas = 1e-3, 1e-2
prior_CH4_gas = 1e-3, 1e-2
prior_O2_gas = 1e-3, 1e-2
prior_H2O_gas = 1e-3, 1e-2
prior_Fe_gas = 1e-3, 1e-2
prior_Mg_gas = 1e-3, 1e-2
prior_SiO_gas = 1e-3, 1e-2
prior_Na_gas = 1e-3, 1e-2
prior_SiH4_gas = 1e-3, 1e-2
prior_Moles_atm = 100.0, 10000.0
prior_Moles_silicate = 100.0, 10000.0
prior_Moles_metal = 100.0, 10000.0

The code parameters file param.dat

All code parameters can be set in the param.dat file. The parameters include:

Gibbs energy file:

<filename>.dat: when a file name is set (with ending .dat), then the Gibbs energies are used from this file. The filename can contain a path to a different directory. (See Gibbs energies)

<filename>.py: when a Python script is set (with ending .py), then no Gibbs energy file is used, and the Python code <filename>.py is used to calculate the Gibbs energies on the fly. The filename can contain a path to a different directory. (See Gibbs energies)

Initial conditions file:

when set to -, then no initial condition file is used, and the initial variables are generated randomly by using the priors set in the chem_input file.

when a file name is set, then the initial values are taken from this file (See The initial conditions file, initial.dat).

Output file: The name of the output file, (default = output.dat).

outputInterval: value at which steps an output is written (default = 10000).

Nwalker: Number of walkers (or chains). Must be larger than 1, default = 100 (See Using multiple walkers).

nSteps: Number of steps in the solver per solver iteration (default = 150000).

nIterations: Number of solver iterations, default = 2 (See Using multiple iterations).

nBestSolutions: Used to generate new initial values for multiple iteration runs, default = 1 (See Using multiple iterations).

method: used solver (default = 2):

1: AdamW (See Solvers).

2: AdamaxW (See Solvers).

AdamW and AdamaxW parameters:

eta: \(\eta\), learning rate in optimizer, default = 0.4, (See Solvers).

lambda_reg: \(\lambda_{reg}\), Weight decay in optimizer, default = 0.0001, (See Solvers).

Example

The default param.dat file looks like:

# General parameters
Gibbs energy file = -
Initial conditions file = -
Output file = output.dat
outputInterval = 10000
Nwalker = 100
nSteps = 150000
nIterations = 2
nBestSolutions = 1
method = 2

# Adam solver parameters
eta = 0.4
lambda_reg = 0.0001

The output file

The name of the output file can be set in the param.dat file. The default name is output.dat

The first line in the output file is a header line and contains a description of the used columns.

These are:

The iteration index

The chain index

The CTOT value (See The Cost Function)

The values of all used variables, with units corresponding to the implementation of the Equations.py file.

In the default implementation, these are \(x_{i,k}\) the mole fraction of component i in phase k,

Followed by \(N_k\), the number of moles of phase k.

When the code is run multiple times, then output is added to the file. The header is written only once at the very top.

The info.dat file

The info.dat file contains all relevant parameters and user settings of each run. The file is not overwritten, and new runs will append to it at the end.