AeX

以下功能目前可用

import aesara.tensor as at
import aex

srng = at.random.RandomStream(0)

sigma_rv = srng.normal(1.)
mu_rv = srng.normal(0, 1)
Y_rv = srng.normal(mu_rv, sigma_rv)

sampler = aex.prior_sampler(Y_rv, mu_rv)
sampler(rng_key, 1_000_000)

即将推出

使用Blackjax的NUTS采样器从后验分布中进行抽样

sampler = aex.mcmc({Y_rv: 1.}, aex.NUTS())
samples, info = sampler(rng_key, 1000, 1000)

通过任意组合Blackjax步骤函数从后验分布中进行抽样

sampler = aex.mcmc({Y_rv: 1.}, {[mu_rv, sigma_rv]: aex.NUTS(), Y_rv: aex.RMH()})
samples, info = sampler(rng_key, 1000)

从后验预测分布中进行抽样

sampler = aex.posterior_predictive(trace, Y_rv)
sampler(rng_key, 1000)