bilby-laplace¶
A Bilby sampler plugin that estimates posteriors via the Laplace approximation — a Gaussian fitted at the maximum a posteriori (MAP) point — followed by optional resampling to correct for non-Gaussianity.
The method is fast, scales well to moderate dimensions, and produces asymptotically exact posterior samples when the true posterior is close to Gaussian. It is useful as a cheap cross-check against nested sampling results.
Originally derived from bilby PR #933.
Where to go next¶
- Installation — install the package and let Bilby discover the sampler.
- Quick start — run your first Laplace analysis.
- How it works — the MAP → covariance → proposal → resampling pipeline.
- Choosing a resampling method —
rejection,importance,inprior,smc, or none. - Covariance estimation — the Hessian vs. waveform-Fisher routes and the options that shape the proposal.
- Background — what is actually computed, and why it is not the Fisher information matrix.
- Configuration & API — the full, auto-generated reference.
At a glance¶
import bilby
result = bilby.run_sampler(
likelihood=likelihood,
priors=priors,
sampler="laplace",
outdir="outdir",
label="my_run",
)
The sampler registers through Bilby's plugin entry-point system, so no extra configuration is needed once it is installed.