Creating model grids

This module provides routines to generate simple (and small) ProDiMo model grids.

Please also look at the command line tool pgrid see Command-line utilities or just type pgrid -h in your command line.

You can make your own pythong script to generate the grid, however, usually it easier to use pgrid and a JSON file to define the parameters and ranges used for the grid. The example shows the different optioncs

Examples

JSON file

Here is an example for the content of JSON file that can be used with pgrid (or with the grid utilities in general).

[
  ["Lstar", 100, 300, 3, "lin"],
  ["env_Mdot", 1e-06, 0.0001, 3, "log"],
  ["Rout", 3000, 6000, 4, "lin"],
  ["new_escape"],
  ["ABS_CRI", [1, 3, 4]]
]

This JSON file is equivalent to the code in the Python script below. To generate a grid with this JSON file called gridparams.json type in the command line

pgrid make grid1 gridparams.json Init

For more details use pgrid -h (i.e. how to run the grid etc.).

Python script

import prodimopy.grid as pgrid

# the grid name
gridname="grid1"
# A director with the initial model (e.g. a normal ProDiMo model with the required input files.)
indir="Init"

# Create a Parameter list.
# The grid will be the same one, as for the example JSON file shown above.
params=list()
params.append(["Lstar",100,300,3,"lin"])
params.append(["env_Mdot",1.e-6,1.e-4,3,"log"])
params.append(["Rout",3000,6000,4,"lin"])
# a boolean value, just the name is required
params.append(["new_escape"])
# use values from a list, that also works with strings
params.append(["ABS_CRI",(1,3,4)])

# make the grid. One gets a list of modelnames (directories), and a tuple with the values for each parameter
modelnames,paramvalues=pgrid.make_grid(gridname, params, indir)

# print the Parameter values
for param,val in zip(params,paramvalues):
    print(f"{param[0]:20s} : {val.flatten()}")

# Run the grid, please not ParameterGrid.in as added automatically to the end of the command string
# $MODELNAME$ is replaced by the model directory name
pgrid.run_grid(gridname,"sbatch -J $MODELNAME$ -p short ../../prodimo.slurm")
prodimopy.grid.GRIDINPUTFILE = 'ParameterGrid.in'

The additional input file which includes all the parameters that are included in the grid.

prodimopy.grid.chgriddir(gridname)[source]

Changes the current working directory to the grid directory.

Parameters:

gridname (str) – The name of the grid (the directory with the models).

prodimopy.grid.genparamentry(name, value)[source]

Generates one Parameter entry in the ProDiMo Parameter.in format.

If name is fcarbon then it it is assumed that the DIANA standard opacities are used,and a proper ! NDUST entry is created (with a fixed vacuum fraction of 0.25).

Parameters:

  • name (str) – The name of the ProDiMo parameter.

  • value (Any) – The value of the Parameter, is converted to a string.

Returns:

The full str that can be written to a file in Parameter.in format.

Return type:

str

prodimopy.grid.genvalues(param)[source]

Generates the values for the given parameter, depending on type.

Parameters:

param (tuple) – A tuple with x entries depending on type. see make_grid().

Returns:

A numpy array with the generated values for that param. Can be of any type (str,float,int,boolean).

Return type:

NDArray[Any]

prodimopy.grid.get_modeldirs(modeldirs=None, gridname=None)[source]

Returns all grid models of the current grid (current directory) or for the grid given by gridname.

Parameters:

  • modeldirs (list | None) – The list of model directories. If None all directories within the current directory starting with model will be selected.

  • gridname (str | None) – The name of the grid (it’s directoryname to be precise). If gridname is given search is done with gridname+"/model*/"

Returns:

If modeldirs is not None modeldirs is returned (so nothing is done). Otherwise all directory names staring with [gridname/]model*/ are returned.

Return type:

list

prodimopy.grid.make_grid(gridname, params, indir=None)[source]

Generates a grid of prodimo models.

A directory with the name gridname is created. Within this directory for each parameter combination a directory is created. This directory is initially copied from the directory given by indir, where certain files (.out,.log,*.fits) are ignored. The generated parameters are added to a ParameterGrid.in file. If ParameterGrid.in does not exist it will be created.

Parameters:

  • gridname (str) – The name of the grid (the directory with the models).

  • params (list) –

    The list of parameters. Depending of the type different entries per Parameter are required.

    • numeric: Entry has to be like (paramname,startvalue,endvalue,numberofsteps,type). type can either be "lin" or "log" for linear or logarithmic spacing

    • boolean: Entry is simply (paramname)

    Special parameters:

    • fcarbon: Only works if DIANA opacities are used. Changes the carbon fraction and the silicate part accordingly.

  • indir (str | None) –

    an input or starting directory containing the usual ProDiMo input files. This directory is copied into a new model.

    If it is None only the ParameterGrid.in are created

Returns:

First entry is the list of modelnames (director values). Second one is a tuple with an entry for each parameter, giving the generated values for this parameter.

Return type:

tuple[list,tuple]

prodimopy.grid.run_grid(gridname, runProDiMo, modeldirs=None)[source]

Runs the grid.

Changes to the grid directory and runs each model by either calling a passed python function or a system command string (see parameters).

Parameters:

  • gridname (str) – The name of the grid (the directory with the models).

  • runProDiMo (str | Callable) –

    If it is a str the string is interpreted as a system command. Any occurrence of $MODELNAME$ in the given string is replaced by the actual model name (model directory).

    If the parameter is a python function (a Callable). This function is called with the modeldir as a parameter.

  • modeldirs (list | None) – A list of all models in the grid (the directory name of each model). If None the list is created automatically using prodimop.grid.get_modeldirs(). Only useful to run a selection of models from the grid (i.e. the failed ones).

prodimopy.grid.check_grid(gridname, modeldirs=None)[source]

Checks if all models look okay.

This routine checks if finished.out exists for all models of the grid.

Parameters:

  • gridname (str) – The name of the grid (the directory with the models).

  • modeldirs (list) –

    a list of all models in the grid (directory name of each model).

    If modeldirs is None all directories with names starting with model are considered as potential grid models.

prodimopy.grid.sel_lowest_chisquare(gridname, modeldirs=None, tolerance=None)[source]

Selects the model(s) with the lowest chi squared. If a tolerance is given all models within this tolerance, measured relative to the minimum chi square are selected.

Parameters:

  • gridname (str) – The name of the grid (the directory with the models).

  • modeldirs (list | None) –

    a list of all models in the grid (directory name of each model).

    If modeldirs is None all directories with names starting with model are considered as potential grid models.

  • tolerance (float | None)

Return type:

tuple[list, list]

prodimopy.grid.sel_param_val(gridname, param, value, modeldirs=None)[source]

Very primitive and inefficient method to select models having a certain value of a parameter. It only searches in grid.GRIDINPUTFILE.

Parameters:

  • gridname (str) – The (directory) name of the grid.

  • param (str) – The parameter name to search for.

  • value (float | int | str) – The value to search for. In case of numbers the read value from the grid Parameter file is converted to the type of value.

  • modeldirs (list[str] | None) – List of model directories of the grid. If None get_modeldirs() is used.

Return type:

list[str]