pymc-devs/pymc

Add rewrite for Mixture when `comp_dists` can be "fused"

Open

#6,803 创建于 2023年6月29日

在 GitHub 查看
 (0 评论) (1 反应) (0 负责人)Python (7,926 star) (1,902 fork)batch import
feature requesthelp wantedlogprob

描述

Description

The following distributions are equivalent:

import pymc as pm

pm.Mixture.dist(w=[0.5, 0.5], comp_dists=[pm.Normal.dist(-1), pm.Normal.dist(1)])
pm.Mixture.dist(w=[0.5, 0.5], comp_dists=pm.Normal.dist([-1, 1]))

But the second one is more efficient, because the logp is vectorized among a single batched Normal.

We could add a rewrite in the logprob_rewrites to convert the former to the second, so that users are not penalized from using the first form (which may be more intuitive for some).

Actually that sort of rewrite stack([rv1, rv2]) -> rv3 could be useful in many places in the logprob submodule

贡献者指南

技术栈
python
领域
backend
议题类型
feature
难度面向新贡献者的预计实现难度，1 表示很小改动，5 表示专家级工作。
3
预计时间有经验贡献者完成调查、实现、测试并准备 pull request 的粗略时间范围。
1-3 hours
活动状态议题当前的可参与程度：新鲜、活跃、陈旧、阻塞或等待维护者输入。
fresh
清晰度议题是否清楚说明期望改动、验收标准和下一步。
clear
前置要求
Familiarity with PyMC's Mixture distributionUnderstanding of logp rewrites in PyMC
新手友好度1-100 的估计分数，表示该议题对首次贡献者的友好程度。
30
研究方向
Investigate the logprob rewrites module in the PyMC codebase (likely in pymc/logprob/rewrites.py or similar). The goal is to add a rewrite that detects when a Mixture's comp dists are all instances of the same distribution (e.g., Normal) with different parameters, and replaces them with a single batched distribution. Look at existing rewrites for patterns. Implement the transformation and add corresponding tests to ensure correctness. Check if any linked PRs or comments provide guidance; currently none.