Skip to content

Latest commit

 

History

History
86 lines (68 loc) · 4.64 KB

CHANGELOG.md

File metadata and controls

86 lines (68 loc) · 4.64 KB
Raw

CHANGELOG

0.1.2

  • Added new docker latest, devel images. Updated installation instructions in documentation.
  • Added is_isomorphic_to function to Graph class. Updated docs and example.

0.1.3

  • Added logic.pl module with simplify, evaluate, to_plformula and all_sat methods for PL formulas.
  • Moved sigma method to Automaton class with default implementation.
  • Fixed the representation of Automaton acceptance conditions.
  • Added new module: automata, with DFA, Monitor, DBA, DCBA, DPA classes. Docs and example added.
  • Automaton can be constructed by passing components (Q, AP, Trans, q0, F) to the constructor.

0.1.4

  • Bugfix in `SpotAutomaton.is_semi_deterministic() function.
  • SpotAutomaton.acc_cond adheres to ggsolver.models.Automaton convention.
  • Made init arguments to classes in ggsolver.automata optional.
  • Added Automaton.from_automaton() function to construct DFA, DBA ... from SpotAutomaton.
  • Added example for translating LTL to DFA.

0.1.5

  • Pointed graphify implemented for GraphicalModel.
  • SubGraph class added.
  • Solvers now operate on SubGraph on given graph to construct node_winner, edge_winner properties.
  • GraphicalModel.states() is now REQUIRED to return a list of hashable objects.
  • dtptb package added for deterministic two-player turn-based games.
  • SWin, ASWin for algorithms for reachability and safety added.
  • mdp package added for qualitative Markov decision processes. ASWin, PWin algorithms for reachability added.
  • Added progress bars to graphify and solvers in dtptb package.
  • [Bugfix] input_domain stores the name of function (= graph property) that stores the input domain. Thus, the reconstructed graphical model has the same input domain functions as the original model.

0.1.6

  • (logic) [Added] atoms parameter to SpotAutomaton constructor.
  • (logic) [Added] Formula base class to logic package.
  • (logic) [Added] PL (propositional logic) module to logic package.
  • (logic) [Added] simplify, sat2formula, allsat, evaluate, substitute functions to PL class.
  • (logic) [Added] LTL logic to logic package.
  • (logic) [Added] ScLTL logic to logic package.
  • (logic) [Added] PL, LTL, ScLTL classes are hashable.
  • (logic) [Added] translate method to LTL, ScLTL classes. ScLTL translates to DFA object, LTL translates to SpotAutomaton object.
  • (logic) [Enhance] Automaton edges are labeled with simplified PL formulas.
  • (logic) [Enhance] Specialized Automaton's graphify method using PL formulas as labels.
  • (logic) [Bugfix] Fixed the circular imports: models.Automaton depends on pl.PL. Class pl.PL depends on i_spot.SpotAutomaton. And class i_spot.SpotAutomaton depends on models.Automaton.
  • (ggsolver core) [Bugfix] The base_only flag for GraphicalModel.graphify results in only underlying graph being constructed.
  • (ggsolver arch) [Enhance] Changed the import statements throughout ggsolver: from ggsolver.**.** import ** -> import ggsolver.**

[Unreleased]

  • Documentation for changes in v0.1.6.
  • Reorganized the examples folder v0.1.6.
  • (logic) [Added] Parser for PrefLTL logic.
  • (logic) [??] PrefLTL and PrefScLTL are hashable.
  • (logic) [Added] PrefModel construction given PrefScLTL formula using Formula2Model transformer.
  • (logic) [Added] null_assumption option to PrefLTL class to enforce the condition that "satisfying some formula is better than satisfying none."
  • (logic) [Added] PrefScLTL to DFPA translation.
  • (example) [Added] an example to demo PrefScLTL to DFPA translation.
  • (logic) [Algorithm] Ranking scheme on DFPA's preference graph.
  • (logic) [Refactored] inc_pbp package containing safe and almost-sure/positive improving strategies.
  • (logic) [Added] Product of DFPA and MDP.
  • (gridworld) [Designed] New architecture for pygame based gridworld simulator.
  • (gridworld) [Added] color_util module with color names.
  • (gridworld) [Added] StateMachine class to track progress of a game state. StateMachine can be stepped forward as well as backward. The action, state history is stored (with ability to limit length of history).
  • (gridworld) [Added] Window class that creates a new pygame window. Window handles event system, rendering and state-machine updates.
  • (gridworld) [Added] GWSim class that maintains the state-machine corresponding to game and updates the window.
  • (gridworld) [Added] Control class that represents any renderable element on Window. Each control handles a set of events.
  • (gridworld) [Added] Grid control with ability to create and manipulate gridworlds.
  • (gridworld) [Added] Cell control representing a cell in a Grid control.