D1.6: Classical MD E-CAM Modules V

In this report for Deliverable 1.6 of E-CAM, 9 software modules in classical dynamics are presented. These modules represent improvements and new features in response to the needs of users. 

These modules represent a wide range of topics in classical molecular dynamics, including developments in software for rare event sampling, for developing neural network potentials, for modeling polymer dynamics, for the optimized design and in-silico validation of fusion proteins especially sensors, and for general analysis of MD trajectories. Modules have been developed that interact with OpenPathSampling, n2p2, ODE, and LAMMPS, as well as a new, independent package called Dask-Traj. Many of these modules are designed to leverage the performance and scalability of other tools. In particular, a number of these module interface with or extend LAMMPS, which has excellent scalability.

Additionally, there has been a focus in these module on high throughput (task-based) computing as a paradigm to enhance scalability, in particular using the Python package Dask.

The 9 modules presented here are:

1. SimStore: OPS new storage subsystem (core)

2. SimStore: Storable functions

3. dask-traj

4. minDist2segments_KKT_for_SRP

5. velocities_resolve_EVC_for_LAMMPS

6. n2p2 - Improved link to HPCMD software

7. n2p2 - Polynomial Symmetry Functions

8. lammps_pyinterfaceExt

9. PIUtils

Each module is thoroughly tested, includes in-code documentation as well as external documentation, frequently in the formof Jupyter notebook examples. 

Section 1 of this report gives a brief description of E-CAM modules and the role of this deliverable in the broader goals of E-CAM Work Package 1 (WP1). Section 2 provides background material on the contexts shared between multiple modules: rare events and applications path sampling, simulation setup tools, and neural network potentials. In section 3, we describe each of the modules and provide links to their documentation. Section 4 describes performance aspects of these modules, and section 6 summarizes the deliverable and describes the outlook for future development of modules within WP1, including the increasing importance of transverse actions across simulation communities and work packages.

As this deliverable is the last one of its series, a section on the overall impact of the results achieved within the Workpackage WP1 Classical Molecular Dynamics was also included in section 5.