get_runtime_config#

jeanspy.model_jax.get_runtime_config

jeanspy.model_jax.get_runtime_config()[source]#

Return the current model_jax runtime configuration.

Notes

  • Device selection is controlled by JAX itself and must be requested before importing JAX. This helper reports both the requested and active backend.

  • Precision can be toggled at runtime through configure_runtime for future traces/compilations.

  • Numerical integration choices are per-call options on sigmalos2 and ConstantAnisotropyModel.kernel rather than process-wide settings.

Inputs and units. configure_runtime accepts jax_enable_x64; get_runtime_config takes no arguments. Device selection should be set through environment variables before JAX initialization.

Returns and shape. Dictionary of effective runtime settings and numerical defaults.

Validity. Changing precision after tracing can create a different numerical analysis and restart identity.

Errors. Device initialization errors from JAX propagate.

Backend. Host configuration of JAX.

Differentiation. Configuration choices are static, not differentiable.

Examples. examples/docs_jax_spherical.py

Return type:

Dict[str, Any]