用python写出来这个贝叶斯网络

好的，这里是一个简单的贝叶斯网络的 Python 实现示例：

import numpy as np

class BayesianNetwork:
    def __init__(self, nodes, edges, probabilities):
        self.nodes = nodes
        self.edges = edges
        self.probabilities = probabilities
    
    def infer(self, evidence):
        # 计算联合概率
        joint_prob = 1.0
        for node in self.nodes:
            parents = self.edges[node]
            if len(parents) == 0:
                # 无父节点，使用节点概率
                prob = self.probabilities[node]
            else:
                # 有父节点，使用条件概率
                parent_states = tuple(evidence[parent] for parent in parents)
                prob = self.probabilities[node][parent_states]
            # 如果节点是证据变量，则只考虑已观测到的状态
            if node in evidence:
                state = evidence[node]
                joint_prob *= prob[state]
            else:
                # 否则考虑所有状态的概率
                joint_prob *= sum(prob)
        
        # 计算归一化常数
        sum_prob = 0.0
        for state in self.get_all_states():
            evidence_copy = evidence.copy()
            evidence_copy.update(state)
            sum_prob += self.get_joint_prob(evidence_copy)
        
        # 计算后验概率
        posterior_probs = {}
        for state in self.get_all_states():
            evidence_copy = evidence.copy()
            evidence_copy.update(state)
            posterior_probs[state] = self.get_joint_prob(evidence_copy) / sum_prob
        
        return posterior_probs
    
    def get_joint_prob(self, evidence):
        joint_prob = 1.0
        for node in self.nodes:
            parents = self.edges[node]
            if len(parents) == 0:
                # 无父节点，使用节点概率
                prob = self.probabilities[node]
            else:
                # 有父节点，使用条件概率
                parent_states = tuple(evidence[parent] for parent in parents)
                prob = self.probabilities[node][parent_states]
            # 如果节点是证据变量，则只考虑已观测到的状态
            if node in evidence:
                state = evidence[node]
                joint_prob *= prob[state]
            else:
                # 否则考虑所有状态的概率
                joint_prob *= sum(prob)
        return joint_prob
    
    def get_all_states(self):
        states = [{}]
        for node in self.nodes:
            new_states = []
            for state in states:
                for value in self.probabilities[node]:
                    new_state = state.copy()
                    new_state[node] = value
                    new_states.append(new_state)
            states = new_states
        return states

这个实现中，我们首先定义了贝叶斯网络的节点、边以及概率表。然后，我们实现了一个 `infer` 方法，用于计算后验概率。在 `infer` 方法中，我们首先计算联合概率，然后通过归一化常数计算后验概率。我们还实现了一个 `get_joint_prob` 方法，用于计算给定证据的联合概率。最后，我们实现了一个 `get_all_states` 方法，用于获取所有可能的状态。