Open cthoyt opened 3 years ago
simplify
Simplification of an atomic expression $A=<T,S>$ given graph G and topological ordering $\pi$join
: Construction of the joint distribution of the set $J$ and a variable $V$ given their conditional sets $D$ and $C$ using d-separation criteria in $G$. $S$ is the current summation variable, $M$ is the set of variables not contained in the expression and $\pi$ is a topological ordering. insert
: Insertion of variable $M'$ into the joint term $P(J |D)$ using d-separation criteria in $G$ . $S$ is the current summation variable and $\pi$ is a topological ordering. deconstruct
Recursive wrapper for the simplification of an expression $B=< B , A , S >$ given graph $G$ and topological ordering $\pi$ . extract
Extraction of terms independent of the summation indices from a expression B = 〈B , A, S 〉 given graph G and topological ordering π . q-simplify
Simplification of a quotient $P_{B1}/P{B_2}$ given by the values of two expressions $B_1 = 〈 B_1 , A_1 , S_1 〉$ and $B_2 = 〈 \mathbf{B}_2 , \mathbf{A}_2 , \mathbf{S}_2 〉 given graph G and topological ordering π . It would be really great to have a y0 DSL to causaleffect converter, then we could create objects to pass directly to the causaleffect simplify implementation as an oracle for the tests
The simplify algorithm takes a graph, a probabilistic expression, and a topological sort order and produces a simplified probabilistic expression
Originally published with https://github.com/cran/causaleffect