Inspecting inference results

This example notebook will show how to load the results after an inference to inspect the properties. For this, we can use the functionalities in the InferenceResult class.

NOTE: This notebook assumes you have ran a jester inference and can point to a results.h5 file by adjusting this notebook. If you have not done this, you can run the chiEFT likelihood with the SMC sampler easily on a laptop. We will use that inference result for the demonstration here.

from jesterTOV.inference.result import InferenceResult

results_path = "../../../examples/inference/smc_random_walk/chiEFT/outdir/results.h5"
result = InferenceResult.load(results_path)

print("Available keys:")
for key, value in vars(result).items():
    print(f"{key}")

[INFO] jester: Loading inference results from ../../../examples/inference/smc_random_walk/chiEFT/outdir/results.h5
[INFO] jester: Successfully loaded blackjax_smc_rw results
Available keys:
sampler_type
posterior
metadata
histories
fixed_params

sampler_type = result.sampler_type
print(f"sampler_type: {sampler_type}")

posterior = result.posterior
print("posterior keys")
print(list(posterior.keys()))

metadata = result.metadata
print("metadata keys")
print(list(metadata.keys()))

# These are the histories of the metrics that are tracked during the inference
histories = result.histories
print(f"\nhistories: {histories}")

fixed_params = result.fixed_params
print(f"\nfixed_params: {fixed_params}")

sampler_type: blackjax_smc_rw
posterior keys
['E_sat', 'E_sym', 'K_sat', 'K_sym', 'L_sym', 'Q_sat', 'Q_sym', 'Z_sat', 'Z_sym', 'cs2_CSE_0', 'cs2_CSE_1', 'cs2_CSE_2', 'cs2_CSE_3', 'cs2_CSE_4', 'cs2_CSE_5', 'cs2_CSE_6', 'cs2_CSE_7', 'cs2_CSE_8', 'n_CSE_0_u', 'n_CSE_1_u', 'n_CSE_2_u', 'n_CSE_3_u', 'n_CSE_4_u', 'n_CSE_5_u', 'n_CSE_6_u', 'n_CSE_7_u', 'nbreak', 'Lambdas_EOS', 'cs2', 'e', 'masses_EOS', 'n', 'n_TOV', 'p', 'radii_EOS', '_sampler_specific', 'log_prob']
metadata keys
['config_json', 'annealing_steps', 'creation_timestamp', 'final_ess', 'final_ess_percent', 'fixed_params', 'kernel_type', 'logZ', 'logZ_err', 'mean_acceptance', 'mean_ess', 'min_ess', 'n_mcmc_steps', 'n_particles', 'n_samples', 'parameter_names', 'sampler', 'sampling_time', 'seed', 'target_ess']

histories: {'acceptance_history': array([0.5309    , 0.53073545, 0.53777032, 0.54399167, 0.53440824,
       0.54470018, 0.53606017, 0.53249006, 0.52549488, 0.52222282,
       0.52849188, 0.52766487, 0.53030358, 0.52324858]), 'ess_history': array([0.98863491, 0.99563471, 0.99846088, 0.99962016, 1.        ,
       0.91933081, 0.93660557, 0.95136335, 0.96040997, 0.96982907,
       0.97403149, 0.97722058, 0.98010417, 0.99572858]), 'log_evidence_history': array([-0.01667338, -0.02296372, -0.02492735, -0.02536324, -0.02536324,
       -0.37465495, -0.63846425, -0.83897929, -1.00144874, -1.13179534,
       -1.24482355, -1.34410429, -1.43345886, -1.4719266 ]), 'tempering_param_history': array([5.83924020e-11, 1.18049570e-10, 1.94830818e-10, 2.97461611e-10,
       4.36159553e-10, 8.77990727e-02, 1.75294419e-01, 2.64565375e-01,
       3.60310130e-01, 4.65103075e-01, 5.88935108e-01, 7.34955474e-01,
       9.12061643e-01, 1.00000000e+00])}

fixed_params: {}