Update README.md

submissions/2024-09-18-OpenFF-NAGL2-ESP-Timing-Benchmark-v1.1/README.md

@@ -1,7 +1,7 @@
 # OpenFF NAGL2 ESP Timing Benchmark v1.1
 
 ## Description
-A dataset of 1005 molecules, sub-sampled from the [OpenFF multi-Br ESP Fragment Conformers v1.1](https://github.com/openforcefield/qca-dataset-submission/tree/master/submissions/2023-11-30-OpenFF-multi-Br-ESP-Fragment-Conformers-v1.1-single-point), the [OpenFF Iodine Fragment Opt v1.0](https://github.com/openforcefield/qca-dataset-submission/tree/master/submissions/2024-09-10-OpenFF-Iodine-Fragment-Opt-v1.0), and the [OpenFF ESP Fragment Conformers v1.0](https://github.com/openforcefield/qca-dataset-submission/tree/master/submissions/2022-01-16-OpenFF-ESP-Fragment-Conformers-v1.0) datasets. Calculated at the PBE0/def2-TZVPPD level of theory, in vacuum and PCM water using the DDX package in Psi4.
+A dataset of 1005 molecules, sub-sampled from the [OpenFF multi-Br ESP Fragment Conformers v1.1](https://github.com/openforcefield/qca-dataset-submission/tree/master/submissions/2023-11-30-OpenFF-multi-Br-ESP-Fragment-Conformers-v1.1-single-point), the [OpenFF Iodine Fragment Opt v1.0](https://github.com/openforcefield/qca-dataset-submission/tree/master/submissions/2024-09-10-OpenFF-Iodine-Fragment-Opt-v1.0), and the [OpenFF ESP Fragment Conformers v1.0](https://github.com/openforcefield/qca-dataset-submission/tree/master/submissions/2022-01-16-OpenFF-ESP-Fragment-Conformers-v1.0) datasets. Calculated at the PBE0/def2-TZVPPD level of theory, in vacuum and PCM water using the DDX package in Psi4. Wavefunctions were saved for the vacuum dataset, but not the PCM dataset. This is because a) we just want to benchmark how much storage space we'd need, and the two sets of wavefunctions should require the same amount of space and b) we expect there may be some errors due to possible numerical issues with I + diffuse functions + PCM that we may need to debug, and don't want to store potentially problematic wavefunctions.
 
 Sub-sampling was done in `subsample_esp_ds.ipynb` by selecting 945 conformers at random from the ESP Fragment dataset, 53 from the I fragment dataset, and 11 from the multi-Br dataset.
 
@@ -36,6 +36,8 @@ The purpose of this dataset is to estimate timing and storage requirements for t
 * `br_subsample_filtered.json`: subset of multi-Br dataset selected for this benchmark
 * `i_subsample_filtered.json`: subset of I fragment dataset selected for this benchmark
 * `esp_subsample_filtered.json`: subset of ESP50k dataset selected for this benchmark
+* `generate-compute.ipynb`: Notebook used to generate the compute expansion `compute.json`
+* `compute.json`: compute expansion specs
 
 ## Metadata
 * elements: {'H', 'P', 'C', 'N', 'O', 'S', 'Cl', 'I', 'Br', 'F'}