ParAMS ReaxFF parametrization challenge
What: Fit a ReaxFF force field using the new ParAMS module of AMS2022.
1st prize: 300 Euro discount on the next AMS license fee + 250 Euro Amazon voucher or equivalent
2nd prize: 200 Euro discount on the next AMS license fee + 100 Euro Amazon voucher or equivalent
3rd prize: 100 Euro discount on the next AMS license fee + 50 Euro Amazon voucher or equivalent
Eligibility: All AMS2021 customers and free trial users (request a 30-day trial) are allowed to participate. You need a valid ParAMS and ReaxFF license. Existing customers can request the needed licenses for the duration of the challenge by e-mailing email@example.com (including SCM username).
Expected time investment: 2-10 hours of human time depending on experience and preference. 4+ hours of computer time. You can spend as much or as little time as you want.
Submission deadline: 1 July 2022 (extended!)
Submission deliverable: Send the optimized force field (ffield.ff, but rename it as firstname_lastname.ff) to firstname.lastname@example.org, together with your name, affiliation, and SCM username.
Questions: Send any questions about the challenge to email@example.com
Tip 1: Before attempting the challenge, check out the Getting Started, ReaxFF: Gaseous H2O, and ReaxFF: Adsorption on ZnS tutorials.
Tip 2: Try to guard against overfitting. Do not optimize too many parameters.
Tip 3: Some bugs have been fixed in the latest bug-fixed binaries.
SCM provides a training set, that you will use during the parametrization. However, the goal of the challenge is to get the smallest error on the test set, that will only be made public after the challenge ends.
Download and unzip params_challenge.zip. It includes a training set, job collection, engine collection, and a ReaxFF parameter interface based on CHON2017_weak_bb.ff.
- modify the job collection in any way you like
- modify the training set in any way you like
- split the training set into a training and validation set
- change the initial force field, parameters to optimize, parameter ranges, …
- use any optimizer and other parametrization settings
- use the GUI or the Python library
The test set will be evaluated with the ReaxFF TaperBO (tapered bond orders) option enabled.
Only one submission is permitted per person.
If you win the challenge, SCM would like to post an interview with you on its website.
SCM may choose to extend the duration of the challenge, and may disqualify an entry for any reason at its sole discretion.
The training set contains data for describing an epoxide-amine polymerization reaction.
- Bond lengths, angles, and charges from geometry optimization jobs
- Conformer energies generated with AMSConformers
- Energies and forces from a bond boost trajectory (reactive events identified by ChemTraYzer2)
- Bond scans
- Reaction enthalpies (isomerization and polymerization reactions)
The test set contains similar types of data, for similar types of molecules.
All training and test data was calculated using the PBE functional with D3(BJ) dispersion corrections and a TZP basis set. The reference atomic charges are Mulliken charges.Key conceptsParametrization ReaxFF