Core Parameters

These parameters control IGM's workflow, hardware configuration, and logging. They are set under the core key and apply globally to every run.

Config Structure

core:
  # Input / Output
  url_data: ""
  folder_data: data
  logging_file: ""

  # Logging
  logging: False
  tf_logging_level: WARNING
  igm_logging_level: WARNING

  # Hardware
  hardware:
    gpu_info: False
    visible_gpus: [0]
    strategy: none
    profile: False

  # Misc
  print_params: False
  print_comp: False
  print_info: True
  print_imported_modules: False
  check_compat_params: True
  check_module_needs: False

  # State variable aliases
  aliases:
    builtin: []
    extra_files: []

  # Custom Modules
  structure:
    root_foldername: user
    conf_foldername: conf
    code_foldername: code
    input_modules_foldername: inputs
    process_modules_foldername: processes
    assimilation_modules_foldername: assimilations
    output_modules_foldername: outputs

Parameters

Description of the parameters:

Name	Description	Default value	Units
url_data	URL from which to download .nc data files automatically.		—
folder_data	Local folder that holds input data downloaded or used by IGM.	data	—
logging_file	Path to a file for logging output. Leave empty to log to stdout only.		—
logging	Enable file-based logging.	False	—
tf_logging_level	TensorFlow logging verbosity level (DEBUG, INFO, WARNING, ERROR).	WARNING	—
igm_logging_level	IGM logging verbosity level (DEBUG, INFO, WARNING, ERROR).	WARNING	—
print_params	Print the full resolved parameter configuration at startup.	False	—
print_comp	Print computational performance information (timing, throughput).	False	—
print_info	Print live simulation progress (time, time step, ice volume).	True	—
print_imported_modules	Print the list of imported modules at startup.	False	—
check_compat_params	Check that user-defined parameter names match known defaults, catching typos early.	True	—
check_module_needs	Check, right after init, that every module's declared input variables exist on state. Set False to skip (e.g. when a needed variable is produced during update rather than at init).	False	—

hardware

Name	Description	Default value	Units
hardware.gpu_info	Print GPU information at startup.	False	—
hardware.visible_gpus	List of GPU device indices made visible to IGM. Only one GPU is used at a time.	[0]	—
hardware.strategy	TensorFlow distribution strategy (none, mirrored, etc.).	none	—
hardware.profile	Enable TensorFlow profiling for performance analysis.	False	—

aliases

Name	Description	Default value	Units
aliases.builtin	Built-in alias sets to load ('descriptive', 'pism'). Empty by default — no aliases are active unless explicitly listed.	[]	—
aliases.extra_files	Paths to custom alias YAML files (alias_name: canonical_name pairs). Resolved relative to the working directory.	[]	—

structure

Name	Description	Default value	Units
structure.root_foldername	Root folder for custom modules, relative to the working directory.	user	—
structure.conf_foldername	Sub-folder inside the root folder containing Hydra configuration stubs.	conf	—
structure.code_foldername	Sub-folder inside the root folder containing Python module source files.	code	—
structure.input_modules_foldername	Sub-folder name for custom input modules.	inputs	—
structure.process_modules_foldername	Sub-folder name for custom process modules.	processes	—
structure.assimilation_modules_foldername	Sub-folder name for custom assimilation modules.	assimilations	—
structure.output_modules_foldername	Sub-folder name for custom output modules.	outputs	—