Skip to content

Latest commit

 

History

History
242 lines (133 loc) · 5.53 KB

ChangeLog.md

File metadata and controls

242 lines (133 loc) · 5.53 KB

Change Log

[TOC]

Release 1.2 - 2023-09-18

Documentation

  • Various minor improvements of documentation

  • Update documentation layout

  • Clean up documentation of Python classes and functions in infrared namespace

  • Improve application example notebook

  • Add applications examples: graph coloring, automata and network parsimony

  • Elaborate documentation of def_constraint_class, def_function_class

Python library

  • New defaults and additional parameters for targeted sampling (multi-dimensional Boltzmann sampling)

  • Add error checking in def_constraint_class, def_function_class

C++ library

  • Rename class ired:ConstraintNetwork to ired:FeatureNetwork

Release 1.1 - 2022-11-15

This release improves support for stochastic optimization based on resampling (among some minor improvements and adapations)

C++ library

  • Introduce support for resampling by restricted traceback. The functionality is directly supported by the Infrared engine without need for recomputation of the partition function.

Python interface

  • Resampling functionality is provided by methods resample() of ir.Sampler and ir.PFClusterTree

  • Add ir.mc_optimize for optimization

Documentation

  • Update the bookchapter tutorial - use new resampling mechanism

Release 1.0 - 2022-11-15

This release facilitates the modeling of sampling and optimization problems from a much broader range of applications by

  • introducing a new modeling syntax which facilitates the definition of constraint models and supporting the definition of new constraints and functions with simple Syntax in Python

  • enabling modeling in a truly declarative and compositional style

  • supporting domain bounds, which can be performance critical for many applications

  • optimizing space requirements

  • enabling optimization in addition to sampling, where both can be based on the same constraint model

  • improving the API documentation

  • adding tutorials and examples in Jupyter notebooks that demonstrate the modeling of design problems in the framework

Python interface

  • Introduce new modeling syntax to define constraint network models based on a new class Model

  • Old class ConstraintNetwork is merged with new class Model

  • Support targeting of only selected features

  • Revise syntax for the definition of function and constraints

  • Add support for optimization (in addition to sampling)

    • allow to select evaluation algebra
    • introduce weighted functions

  • Support IUPAC sequence constraints with a specific form of constraint simplification

  • Rename functions and constraints in module rna

  • Support named variables in constraint models

C++ library

  • Support upper and lower bound of variable domains; add class FiniteDomain

  • Remove dependency on boost::graph

  • Implement traceback for constraint optimization

  • Optimize space consumption of materialized functions and constraints by using a specialized replacement SimpleMap of standard sparse containers like std::hash_map

  • Add evaluation in arctic semiring for constraint optimization

Documentation

  • Revisit API documentation

  • Adapt tutorial notebook to new Python interface

  • Add example for multi target design from Python code (Jupyter notebook)

  • Add Jupyter notebook as supplement of a book chapter on Infrared with new example code

  • Add tags to sync with book chapter

Build system

  • Update setup.py

Bug fixes

  • Fix bug of repeated dependencies in the specification of models

  • Fix undefined behavior in the materialization of functions/constraints

Release 0.5.1 - 2021-07-10

Python Interface

  • Export algebra for constraint optimization

  • Redefine constraints and functions in Python

C++ library

Materialization of functions /and/ constraints

  • supports the definition of functions and constraints in Python without performance penalties

  • the existing mechanism for functions was extended to constraints

Bug Fix

  • Update behavior of BoltzmannSampler.plot_td()

Documentation

  • Re-enable API documentation via doxygen

  • Update tutorial jupyter notebook

Release 0.5 - 2021-05-04

C++ library

Use pybind11

Installation and dependencies

  • Change build mechanism to setup.py and cmake

  • Enable installation via pip and conda

Release 0.4 - 2020-11-08

Python Interface

  • Refactor and streamline the Python interface

  • Support tree decomposition by networkx, improve the tree decomposition interface

  • Add access to partition function and provide consistency check in

Documentation

  • Add tutorial of the Python interface (Jupyter notebook)

C++ library

  • Add tests, fix handling of inconsistent networks

  • Revert to using boost::python (to resolve a bug in the use of pybind11)

Installation and dependencies

  • Improve installation via autotools

  • Replace dependency on libhtd by networkx

Release 0.3 - 2018-08-17

  • Generalize Python classes for Boltzmann sampler construction and provide them in Python module infrared for simpler use of the lib

  • Implement multi-dimensional Boltzmann sampling and provide this functionality via module infrared

Release 0.2.1 - 2018-08-08

  • Add/cleanup Doxygen documentation of Python modules

Release 0.2 - 2018-08-07

  • Improve program comments, support doxygen (make doxygen-doc)

  • Revise code, fix make check

Release 0.1 - 2018-08-07

  • Optimize infrared evaluation/sampling engine

  • Optimize tree decomposition using libhtd

  • Cleanup treedecomp and redprint Python code and add features

  • Bugfixes

  • Implement some tests

Release 0.0.1 - 2018-08-02

Initial release