Simultaneously improving reaction coverage and computational cost in automated reaction prediction tasks.

Area of Science:

• Computational Chemistry
• Chemical Reaction Prediction
• Materials Science

Background:

• Automated reaction prediction is crucial for understanding complex reaction networks in fields like combustion and materials degradation.
• High computational cost and limited reaction coverage hinder the exploration of extensive reaction networks.

Purpose of the Study:

• To reduce computational cost and increase reaction coverage in automated reaction prediction.
• To develop and benchmark a novel reaction prediction package, YARP (Yet Another Reaction Program).

Main Methods:

• Modifying reaction enumeration, geometry initialization, and transition state convergence algorithms.
• Implementing these modifications in the YARP package.
• Benchmarking YARP on organic single-step reactions, thermal degradation, and competing ring-closures.

Main Results:

• YARP significantly reduces reaction characterization costs by nearly 100-fold.
• The package increases the convergence of transition states.
• YARP successfully rediscovers known and identifies novel reaction pathways and products.

Conclusions:

• Optimized algorithms in YARP enable simultaneous reduction in computational cost and increase in reaction coverage.
• This advancement facilitates the exploration of larger systems and complex reaction networks.
• The findings are particularly impactful for applications like chemical degradation where computational cost is a major limitation.