Setting custom parameters

Introduction

The possibilities offered by study-DA to scan parameters might sometimes be limited. For instance, you might want to input a parameter as a complex function of another parameter. Or you might want to load parametric values from a file. In such cases, you can set custom parameters instead of using a keyword to scan them in the scan configuration file.

Scan configuration

For this example, we decide to use a scan configuration that is very simple, based on the template configuration provided for HL-LHC v1.6:

config_scan.yaml

# ==================================================================================================
# --- Structure of the study ---
# ==================================================================================================
name: example_custom_parameters

# List all useful files that will be used by executable in generations below
# These files are placed at the root of the study
dependencies:
  main_configuration: config_hllhc16.yaml

structure:
  # First generation is always at the root of the study
  # such that config_hllhc16.yaml is accessible as ../config_hllhc16.yaml
  generation_1:
    executable: generation_1.py
    common_parameters:
      # Needs to be redeclared as it's used for parallelization
      # And re-used ine the second generation
      n_split: 5

  # Second generation depends on the config from the first generation
  generation_2:
    executable: generation_2_level_by_nb.py

Generating script

We're now going to introduce how to input custom parameters in the generating script. In this example, we want to scan the tunes qx and qy for the LHCb1 and LHCb2 beams. In addition, we have to pay attention not to neglect the n_split parameter used for parallelization.

custom_parameters.py

# ==================================================================================================
# --- Imports
# ==================================================================================================

from study_da import create

# ==================================================================================================
# --- Script to generate a study
# ==================================================================================================
# Generate the arrays of parameters to scan
l_qx = []
l_qy = []
l_qx_for_naming = []
l_qy_for_naming = []
l_n_split = []
l_n_split_for_naming = []

for qx in [62.310, 62.311]:
    for qy in [60.320, 60.321]:
        for n_split in range(5):
            l_qx.append({key: qx for key in ["lhcb1", "lhcb2"]})
            l_qy.append({key: qy for key in ["lhcb1", "lhcb2"]})
            l_qx_for_naming.append(qx)
            l_qy_for_naming.append(qy)
            l_n_split.append(f"{n_split:02d}.parquet")
            l_n_split_for_naming.append(n_split)

# Store the parameters in a dictionary
dic_parameter_all_gen = {
    "generation_2": {
        "qx": l_qx,
        "qy": l_qy,
        "n_split": l_n_split,
    }
}

dic_parameter_all_gen_naming = {
    "generation_2": {
        "qx": l_qx_for_naming,
        "qy": l_qy_for_naming,
        "n_split": l_n_split_for_naming,
    }
}

# Generate the study in the local directory
path_tree, name_main_config = create(
    path_config_scan="config_scan.yaml",
    dic_parameter_all_gen=dic_parameter_all_gen,
    dic_parameter_all_gen_naming=dic_parameter_all_gen_naming,
)

As you can see, the first part of the script is used to define the parameter space. We basically do the usual cartesian product of the tunes and n_split parameters manually. Because we don't want the lhcb1 or lhcb2 strings to appear in the filenames when generating the study, we also define a dictionnary of parameters for the naming of the files.

In practice, you can customize the parameters as you wish, as long as you respect the structure of the dictionary dic_parameter_all_gen and dic_parameter_all_gen_naming (although the latter is optional, but can lead to very long or inconstistent filenames if not used).

You can then generate your study as usual.