trivial updates

Author: jla-gardner

paper/paper.bib

@@ -190,17 +190,16 @@ @article{Larsen-17-06
   urldate = {2022-11-08}
 }
 
-@misc{Kovacs-25-01,
-  title         = {{{MACE-OFF}}: {{Transferable Short Range Machine Learning Force Fields}} for {{Organic Molecules}}},
-  author        = {Kov{\'a}cs, D{\'a}vid P{\'e}ter and Moore, J. Harry and Browning, Nicholas J. and Batatia, Ilyes and Horton, Joshua T. and Pu, Yixuan and Kapil, Venkat and Witt, William C. and Magd{\u a}u, Ioan-Bogdan and Cole, Daniel J. and Cs{\'a}nyi, G{\'a}bor},
-  year          = {2025},
-  month         = jan,
-  number        = {arXiv:2312.15211},
-  eprint        = {2312.15211},
-  primaryclass  = {physics},
-  doi           = {10.48550/arXiv.2312.15211},
-  urldate       = {2025-01-30},
-  archiveprefix = {arXiv}
+@article{Kovacs-25-05,
+  title     = {{{MACE-OFF}}: {{Short-Range Transferable Machine Learning Force Fields}} for {{Organic Molecules}}},
+  author    = {Kov{\'a}cs, D{\'a}vid P{\'e}ter and Moore, J. Harry and Browning, Nicholas J. and Batatia, Ilyes and Horton, Joshua T. and Pu, Yixuan and Kapil, Venkat and Witt, William C. and Magd{\u a}u, Ioan-Bogdan and Cole, Daniel J. and Cs{\'a}nyi, G{\'a}bor},
+  year      = 2025,
+  journal   = {Journal of the American Chemical Society},
+  volume    = {147},
+  number    = {21},
+  pages     = {17598--17611},
+  doi       = {10.1021/jacs.4c07099},
+  copyright = {https://creativecommons.org/licenses/by/4.0/}
 }
 
 @article{Batatia-25-11,
@@ -279,19 +278,18 @@ @misc{Rhodes-25-04
   archiveprefix = {arXiv}
 }
 
-@misc{Mahmoud-25-02,
-  title         = {Assessing Zero-Shot Generalisation Behaviour in Graph-Neural-Network Interatomic Potentials},
-  author        = {Ben Mahmoud, Chiheb and {El-Machachi}, Zakariya and Gierczak, Krystian A. and Gardner, John L. A. and Deringer, Volker L.},
-  year          = {2025},
-  month         = feb,
-  number        = {arXiv:2502.21317},
-  eprint        = {2502.21317},
-  primaryclass  = {physics},
-  doi           = {10.48550/arXiv.2502.21317},
-  urldate       = {2025-03-04},
-  archiveprefix = {arXiv}
+@article{Mahmoud-25-11,
+  title   = {Assessing Zero-Shot Generalisation Behaviour in Graph-Neural-Network Interatomic Potentials},
+  author  = {Mahmoud, Chiheb Ben and {El-Machachi}, Zakariya and Gierczak, Krystian A. and Gardner, John L. A. and Deringer, Volker L.},
+  year    = 2025,
+  journal = {Digital Discovery},
+  volume  = {4},
+  number  = {11},
+  pages   = {3389--3399},
+  doi     = {10.1039/D5DD00103J}
 }
 
+
 @article{Pickard-22-07,
   title   = {Ephemeral Data Derived Potentials for Random Structure Search},
   author  = {Pickard, Chris J.},

paper/paper.md

@@ -47,7 +47,7 @@ The `graph-pes` package provides a **unified interface and framework** for defin
 
 # Related work
 
-`graph-pes` is beginning to drive projects within our research group, and we hope that it will be useful to many others. In recent work, we have described the use of `graph-pes` for fitting NequIP models to datasets created using `autoplex` [@Liu-25-08], for assessing zero-shot performance of different graph-network MLIPs [@Mahmoud-25-02], and for distilling atomistic foundation models [@Gardner-25-06].
+`graph-pes` is beginning to drive projects within our research group, and we hope that it will be useful to many others. In recent work, we have described the use of `graph-pes` for fitting NequIP models to datasets created using `autoplex` [@Liu-25-08], for assessing zero-shot performance of different graph-network MLIPs [@Mahmoud-25-11], and for distilling atomistic foundation models [@Gardner-25-06].
 
 A number of existing packages offer training and validation pipelines for particular ML-PES architectures, including `schnetpack` [@schutt2019schnetpack; @schutt2023schnetpack], `deepmd-kit` [@Wang-18-07; @Zeng-23-08], `nequip` [@Batzner-22-05], `mace-torch` [@Batatia-22-10], `torchmd-net` [@TorchMDNet], and `fairchem` [@fairchem]. 
 These frameworks focus on their associated model families and do not share a common interface for training.
@@ -70,7 +70,7 @@ Full API details are available in the project documentation.
 All MLIP models in `graph-pes` are implemented as subclasses of the `GraphPESModel` base class. 
 Implementations need only define a forward pass that returns a local energy for each atom or a total energy for the structure; 
 the framework handles the calculation of forces and stress tensors in a conservative manner via automatic differentiation. 
-We also support models that return direct force and stress tensor predictions (e.g., `TensorNet` or `orb-v3-*` with their optional direct force readout heads).
+We also support models that return direct force and stress tensor predictions (_e.g._, `TensorNet` or `orb-v3-*` with their optional direct force readout heads).
 
 Building on the `GraphPESModel` class, we provide independent re-implementations of popular MLIP architectures, including `PaiNN` [@Schutt-21-06], `EDDP` [@Pickard-22-07], `NequIP` [@Batzner-22-05], `MACE` [@Batatia-22-10], and `TensorNet` [@Simeon-23-06]. 
 We use building blocks provided by `e3nn` [@Geiger-22-07] to implement models that act on spherical tensor decompositions.
@@ -90,7 +90,7 @@ Because all models conform to the same interface, all training features can be u
 
 ## Easy access to foundation models
 
-A recent area of research is the development of "foundational" MLIPs that can describe the potential-energy surface of a wide range of systems. `graph-pes` integrates directly with the `mace-torch`, `mattersim`, and `orb-models` packages to provide access to, among others, the `MACE-MP` [@Batatia-25-11], `MatterSim` [@Yang-24-05], `orb-v2` [@Neumann-24-10], `MACE-OFF` [@Kovacs-25-01],  `Egret-v1` [@Mann-25-05], and `orb-v3` [@Rhodes-25-04] families of models. Each of these integrations generates `GraphPESModels` that are directly compatible with all relevant `graph-pes` features.
+A recent area of research is the development of "foundational" MLIPs that can describe the potential-energy surface of a wide range of systems. `graph-pes` integrates directly with the `mace-torch`, `mattersim`, and `orb-models` packages to provide access to, among others, the `MACE-MP` [@Batatia-25-11], `MatterSim` [@Yang-24-05], `orb-v2` [@Neumann-24-10], `MACE-OFF` [@Kovacs-25-05],  `Egret-v1` [@Mann-25-05], and `orb-v3` [@Rhodes-25-04] families of models. Each of these integrations generates `GraphPESModels` that are directly compatible with all relevant `graph-pes` features.
 
 # Acknowledgements