r-jy / tensor-DFAs

Functions to investigate learning deterministic finite automata (DFA). Summer 2019 Harvey Mudd College AMISTAD lab.

tensor-DFAs

Functions to investigate learning deterministic finite automata (DFA). Summer 2019 Harvey Mudd College AMISTAD lab.

Represent DFA as a tensor

• DfaTensor: Contains class to represent DFA as a q*q*a tensor, where q is the number of states and a is the size of the alphabet of input characters. The [i][j][k] entry of this tensor is 1 if there is a transition from the i^th to the j^th state on the k^th input character in the alphabet, 0 otherwise.
• TensorGenerator: Expands on DfaTensor class to contain functions that read an input string and output whether or not it is accepted.

Find accepting strings

• AcceptingStringsGenerator: Generates all accepting strings of a given length. Uses a dynamic programming approach.
• STPTools: Generates all accepting strings of a given length. Code is based on algorithm 1, pg. 952 of Yan, Chen, and Liu’s “Semi-tensor product of matrices approach to reachability of finite automata with application to language recognition.” See also STP, SwapMatrix.
• StringGenerator: Generates the subset of all accepting strings for which no state in the DFA is visited twice while reading the string. This is implemented by a path-finding algorithm for graphs. See also Graph, a modification of the adjacency list graph class from geeks for geeks.