Changelog¶

Release 2.2.0¶

New features

Introduce Pathfinder, a graph-based approach to analyze how compounds are connected via reactions while considering kinetic and stoichiometric constraints.

New features

Introduce Flasks to the reaction networks (aggregates of stable non-bonded complexes)
Elementary-step gear that uses the current minimum-energy conformer for reaction trial generation.
Added a gear that sets up kinetic modeling jobs.
Allow the refinement of a subset of elementary steps per reaction. The subset is given through an energy cut-off above the lowest lying transition state.
Introduce possibility to efficiently explore barrierless dissociations.

Python rewrite, and open source release with the following initial features:

Scriptable framework including a base set of features for the automated exploration of chemical reaction networks
Initial chemical reaction networks consisting of
- Structures aggregated into Compounds
- Elementary Steps aggregated into Reactions
- Properties tagged to Structures
- Calculations that generated the network
Definitions of Engines with perpetually running Gears to continuously perform tasks with chemical reaction networks (see list below)
Storage and expansion of chemical reaction networks in a SCINE Database
Automated job set up and execution via SCINE Puffin
Definitions of basic filters to reduce number of Elementary Step trials (see list below)

Initial Engines/Gears:

Basic bookkeeping jobs:
- Sorting Structures into Compounds (BasicCompoundHousekeeping)
- Sorting Elementary Steps into Reactions (BasicReactionHousekeeping)
- Basic Scheduling and prioritization of Calculations (Scheduler)
Data completion jobs:
- Conformer generation per compound (BruteForceConformers)
- Hessian generation per transition state and minimum energy Structure (BasicThermoDataCompletion)
Elementary Step exploration based on existing Compounds:
- For one Structure per Compound (MinimalElementarySteps):
  
  Based on atoms/fragments (AFIR, NT1)
  
  Based on bonds (NT2)
- For all combinations of Structures per Compounds (BruteForceElementarySteps):
  
  Based on atoms/fragments (AFIR, NT1)
  
  Based on bonds (NT2)
Steering of network growth via simple kinetic analyses:
- Based on connectivity to user input (MinimalConnectivityKinetics)
- Based on barrier heights of Elementary Steps (BasicBarrierHeightKinetics)

Initial set of filters:

Compound filtering possible:
- Base class, allows all compounds (CompoundFilter)
- By element counts (ElementCountFilter, ElementSumCountFilter)
- By atom counts or molecular weights (MolecularWeightFilter, AtomNumberFilter)
- By database IDs (IDFilter, OneCompoundIDFilter, SelectedCompoundIDFilter)
- By context (SelfReactionFilter)
- By Hessian evaluation (TrueMinimumFilter)
- By composition (CatalystFilter)
Reactive site filtering possible:
- Base class, allows all reactive sites (ReactiveSiteFilter)
- By fixed, simple rankings (SimpleRankingFilter, MasmChemicalRankingFilter)
- By custom user rules (AtomRuleBasedFilter, FunctionalGroupRule)
- By atom types (ElementWiseReactionCoordinateFilter)
All filters of the same type can be chained with logical operations to tailor the behaviour

Closed source C++ prototype implementation.