Plot parameters

Plot parameters#

Starting with version 11.2.3, it is possible to provide additional plotting parameters to speed up the plot model calculation when running pyorbit_results.

The keywords can be listed in the parameters section or in a new plot_parameters section:

parameters:
  Tref: 2456000.00
  cpu_threads: 128
  safe_reload: False
plot_parameters:
  low_ram_plot: True
  plot_split_threshold: 1000
  ...

The default keyword (i.e., the value used by PyORBIT if not provided in the yaml configuration file) is highlighted in boldface.

save_pdf

  • accepted values: True | False

  • If True, files are saved to PDF instead of PNG format.

plot_config_parameters

  • accepted values: any integer | 9

  • Default font size for the majority of the plots Plots have been generated with matplotlib using figure.figsize=(9,9) and the Arial font family. Some codes may override these settings.

rv_lnlike_samplings

  • accepted values: any integer | n_sample

  • aliases: rv_loglike_samplings', rv_lnlike_samplings, rv_like_samplings`

  • Number of samplings to be used to compute the RV log-likelihood. The computation is not parallelised, so computing the RV log-likelihood for all the samples could take too much time. If a number is provided, the RV log-likelihood will be computed on a random selection of rv_lnlike_samplings samples.

correlation_plot

  • accepted values: pygtc | corner | getdist

  • Default code to compute the corner plot, among pygtc, corner, getdist.

force_full_correlation_plot

  • accepted values: True | False

  • When the model has more than 30 parameters, the full correlation plot is not generated anymore as the crowding would render it useless. Set this flag to True to disable this behaviour.

Plot step size#

It is common practice to generate a model with much denser sampling than the original data, so that smooth variations in the model can be appreciated.

PyORBIT will automatically decide the step size according to this scheme:

  • The shortest orbital period among all the planets in the sample is taken as reference as minimum_planet_period, even if that planet is not used in the modelling of a specific dataset.

  • The minimum non-zero step size of the sorted data is computed, and divided by two to compute the minimum_step_size

  • For photometry data, the step size is automatically increased to 5 minutes if the range of the dataset is larger than three times ‘minimum_planet_period’, otherwise minimum_step_size is used

  • For all the other kinds of datasets, the code will use the minimum value among the minimum_planet_period divided by 20, the minimum_step_size, and the range of the dataset divided by ten times the number of points

use_shared_axis_for_rv

  • accepted values: True | False

  • All radial velocity datasets will share the same axis

use_shared_axis_for_activity

  • accepted values: True | False

  • All activity index datasets will share the same axis. Photometry is never considered an activity index even if it is modelled with an activity model.

GP regression#

Computing the model with denser sampling than the original data can be computationally expensive for datasets with a large temporal baseline, as the computational time can increase with the cube of the covariance matrix size. These options should reduce runtime and RAM usage.

Tip

If you get a segmentation fault error or the code stops abruptly while generating plots, you most likely exceeded the maximum RAM usage.

plot_split_threshold

  • accepted values: any integer | 10000

  • The calculation of models involving the covariance matrix (e.g., GP regression) will be split into smaller chunks of \(N=\)plot_split_threshold data points. Helpful to speed up the calculation of GP regression for large datasets or multivariate GP analysis.

low_ram_plot

  • accepted values: True | False

  • If True, the plot_split_threshold is further enforced for larger datasets.

compute_gp_variance

  • accepted values: True | False

  • If True, the variance of the GP regression is computed together with the model.

progress_bar

  • accepted values: True | False

  • It will show a progress bar when computing the model using a split threshold. Note_ the progress bar is not entirely accurate, so it may not reach 100% even for a successful job.

              #if dataset.kind in activity_datatype or dataset.kind == 'radial_velocity':
          #    input_step_size = plot_config_parameters['model_step_size'].get('activity', input_step_size)                
          #if dataset.kind == 'radial_velocity':
          #    input_step_size = plot_config_parameters['model_step_size'].get('radial_velocity', input_step_size)                
          #if dataset.kind == 'photometry':
          #    input_step_size = plot_config_parameters['model_step_size'].get('photometry', input_step_size)     
          #input_step_size = plot_config_parameters['model_step_size'].get(dataset_name, input_step_size)