DOC: JOSS final editorial changes

paper/paper.bib

@@ -1167,17 +1167,15 @@ @article{preCICEv2
 }
 
 @InProceedings{Dehning2014,
-author="Dehning, Carsten
-and Bierwisch, Claas
-and Kraft, Torsten",
-editor="Griebel, Michael
-and Schweitzer, Marc Alexander",
-title="Co-simulations of Discrete and Finite Element Codes",
-booktitle="Meshfree Methods for Partial Differential Equations VII",
-year="2015",
-publisher="Springer International Publishing",
-address="Cham",
-pages="61--79",
-abstract="This paper describes methods and algorithms implemented for the co-simulation between a mesh-free discrete element method (DEM) code and a finite element analysis (FEA) code under control of a co-simulation middleware software environment.",
-isbn="978-3-319-06898-5"
+  author    = "Dehning, Carsten and Bierwisch, Claas and Kraft, Torsten",
+  editor    = "Griebel, Michael and Schweitzer, Marc Alexander",
+  title     = "Co-simulations of Discrete and Finite Element Codes",
+  booktitle = "Meshfree Methods for Partial Differential Equations VII",
+  year      = "2015",
+  publisher = "Springer",
+  address   = "Cham",
+  pages     = "61--79",
+  abstract  = "This paper describes methods and algorithms implemented for the co-simulation between a mesh-free discrete element method (DEM) code and a finite element analysis (FEA) code under control of a co-simulation middleware software environment.",
+  isbn      = "978-3-319-06898-5",
+  doi       = "10.1007/978-3-319-06898-5_4"
 }

paper/paper.md

@@ -138,7 +138,7 @@ As well as these domain-specific tools, `matscipy` contains general utility func
   This allows compact code for evaluating properties that depend on pairs, such as pair-distribution function or interatomic potential energies and forces. Most of the tools described in the following rely on this neighbour list format.
   The neighbour list is becoming widely used for post-processing and structural analysis of the trajectories resulting from molecular dynamics simulations, and even to accelerate next-generation message passing neural networks such as MACE [@Batatia2022mace;@Batatia2022Design].
 
-![Neighbour list computation time comparison between ASE and Matscipy implementations.\label{fig:nl_time}](nl_time.svg)
+![Execution time of the computation of the neighbour list in ASE and Matscipy. These results were obtained on a single core of an Intel i7-1260P processor on the ASE master branch (git hash 52a8e783).\label{fig:nl_time}](nl_time.svg)
 
 - **Atomic strain.** Continuum mechanics is formulated in terms of strains, which characterizes the fractional shape changes of small volumes. Strains are typically only well defined if averaged over sufficiently large volumes, and extracting strain fields from atomic-scale calculations is notoriously difficult. `matscipy` implements calculations of strain by observing changes in local atomic neighbourhoods across trajectories. It fits a per-atom displacement gradient that minimizes the error in displacement between two configurations as described by @Falk1998. The error resulting from this fit quantifies the non-affine contribution ot the overall displacement and is known as $D^2_\text{min}$. We used this analysis to quantify local strain in the deformation of crystals [@Gola2019;@Gola2020] and glasses [@Jana2019].
 
@@ -160,7 +160,7 @@ The module `matscipy.numerical` additionally provides routines for the numerical
 
 - **Quantum mechanics/molecular mechanics.** The module `matscipy.calculators.mcfm` implements a generalised force-mixing potential [@Bernstein2009] with support for multiple concurrent QM clusters, named MultiClusterForceMixing (MCFM). It has been applied to model failure of graphene-nanotube composites [@Golebiowski2018;@Golebiowski2020].
 
-- **Committee models.**. The module `matscipy.calculators.committee` provides support for committees of interatomic potentials with the same functional form but differing parameters, in order to allow the effect of the uncertainty in parameters on model predictions to be estimated. This is typically used with machine learning interatomic potentials (MLIPs). The implementation follows the approach of [@Musil2019] where the ensemble of models is generated by training models on different subsets of a common overall training database.
+- **Committee models.** The module `matscipy.calculators.committee` provides support for committees of interatomic potentials with the same functional form but differing parameters, in order to allow the effect of the uncertainty in parameters on model predictions to be estimated. This is typically used with machine learning interatomic potentials (MLIPs). The implementation follows the approach of [@Musil2019] where the ensemble of models is generated by training models on different subsets of a common overall training database.
 
 # Acknowledgements