You can interact with this notebook online: Launch notebook

How to use Adaptive Damping for Convergence in TARDIS¶

This notebook demonstrates how to run the TARDIS example configuration with the existing damped convergence strategy and the new adaptive_damped strategy, and illustrates their convergence behavior across iterations.

Damped Convergence Strategy¶

This example uses the standard TARDIS damped convergence strategy specified in the configuration file tardis_example.yml with the damping constant set to 0.5 (or another user-specified value between 0.1 and 1.0).

[1]:

!cat tardis_example.yml

# Example YAML configuration for TARDIS
tardis_config_version: v1.0

supernova:
  luminosity_requested: 9.44 log_lsun
  time_explosion: 13 day

atom_data: kurucz_cd23_chianti_H_He_latest.h5

model:
  structure:
    type: specific
    velocity:
      start: 1.1e4 km/s
      stop: 20000 km/s
      num: 20
    density:
      type: branch85_w7

  abundances:
    type: uniform
    O: 0.19
    Mg: 0.03
    Si: 0.52
    S: 0.19
    Ar: 0.04
    Ca: 0.03

plasma:
  disable_electron_scattering: no
  ionization: lte
  excitation: lte
  radiative_rates_type: dilute-blackbody
  line_interaction_type: macroatom

montecarlo:
  seed: 23111963
  no_of_packets: 4.0e+4
  iterations: 20
  nthreads: 1

  last_no_of_packets: 1.e+5
  no_of_virtual_packets: 10

  convergence_strategy:
    type: damped
    damping_constant: 1.0
    threshold: 0.05
    fraction: 0.8
    hold_iterations: 3
    t_inner:
      damping_constant: 0.5

spectrum:
  start: 500 angstrom
  stop: 20000 angstrom
  num: 10000

Running the Simulation¶

To run the simulation, import the run_tardis function and create the sim_damped object.

[2]:

from tardis import run_tardis

[3]:

sim_damped = run_tardis("tardis_example.yml",
                 virtual_packet_logging=True,
                 show_convergence_plots=True,
                 export_convergence_plots=True,
                 log_level="INFO")

Auto-detected Sphinx build environment

Auto-detected Sphinx build environment

Initializing tabulator and plotly panel extensions for widgets to work

Embedding the final state for Jupyter environments

Adaptive Damped Convergence Strategy¶

This example uses the adaptive_damped convergence strategy specified in the configuration file tardis_adaptive_damped_example.yml, where the damping factor is adjusted dynamically during the simulation, removing the need for manual tuning.

[4]:

!cat tardis_adaptive_damped_example.yml

# Example YAML configuration for the adaptive damped convergence strategy in TARDIS
tardis_config_version: v1.0

supernova:
  luminosity_requested: 9.44 log_lsun
  time_explosion: 13 day

atom_data: kurucz_cd23_chianti_H_He_latest.h5

model:
  structure:
    type: specific
    velocity:
      start: 1.1e4 km/s
      stop: 20000 km/s
      num: 20
    density:
      type: branch85_w7

  abundances:
    type: uniform
    O: 0.19
    Mg: 0.03
    Si: 0.52
    S: 0.19
    Ar: 0.04
    Ca: 0.03

plasma:
  disable_electron_scattering: no
  ionization: lte
  excitation: lte
  radiative_rates_type: dilute-blackbody
  line_interaction_type: macroatom

montecarlo:
  seed: 23111963
  no_of_packets: 4.0e+4
  iterations: 20
  nthreads: 1
  last_no_of_packets: 1.e+5
  no_of_virtual_packets: 10

  convergence_strategy:
    type: adaptive_damped
    threshold: 0.05
    fraction: 0.8
    hold_iterations: 3
    stop_if_converged: no

spectrum:
  start: 500 angstrom
  stop: 20000 angstrom
  num: 10000

Running the Simulation¶

To run the simulation, import the run_tardis function and create the sim_adaptive object.

[5]:

sim_adaptive = run_tardis("tardis_adaptive_damped_example.yml",
                 virtual_packet_logging=True,
                 show_convergence_plots=True,
                 export_convergence_plots=True,
                 log_level="INFO")

Embedding the final state for Jupyter environments

[ ]: