Equivalence analysis between quasi-coarse-grained and atomistic simulations

Dong-Dong Jiang, Jian-Li Shao, Pei WangComput. Phys. Commun 2025 · DOI

TL;DR

This paper argues that current quasi-coarse-grained dynamics (QCGD) methods, while seemingly offering mesoscale insights, are fundamentally just scaled-up versions of atomistic simulations. The authors demonstrate that QCGD reproduces atomistic results without capturing true mesoscale phenomena, suggesting that claims of QCGD revealing "scale effects" might be misleading.

What Problem Does This Paper Address?

Simulating materials at the atomic level (atomistic simulations, or MD) provides incredible detail but is computationally prohibitive for larger systems and longer timescales. To bridge this gap, mesoscale simulation methods like quasi-coarse-grained dynamics (QCGD) have emerged. QCGD aims to represent groups of atoms with single "representative particles," allowing for larger system sizes and faster simulations. However, a critical question remains: do these QCGD methods truly capture emergent mesoscale phenomena, or are they simply a computationally cheaper way to reproduce atomistic behavior? This paper specifically investigates whether QCGD can accurately model "scale effects" – how material properties change with size – in crystalline materials.

Methods

The core of this paper is a theoretical and numerical analysis of the equivalence between atomistic molecular dynamics (MD) and quasi-coarse-grained dynamics (QCGD).

  • QCGD Formulation: The authors focus on QCGD methods that scale interatomic potentials directly. In this approach, representative particles (R-particles) are introduced, each representing a cluster of atoms. The QCGD system maintains the same structure and orientation as the atomistic system but with a larger lattice constant, defined by a "distance scaling parameter" ($L_{cg}$). The potential energy function for R-particles is derived by directly scaling the atomistic potential.
  • Theoretical Equivalence Proof: The paper mathematically demonstrates that under these scaling rules, the forces and subsequent trajectories of R-particles in QCGD are directly proportional to the forces and trajectories of their corresponding atoms in MD. This is shown by deriving relationships between the forces and accelerations in both systems, highlighting that the QCGD evolution is essentially a scaled version of the MD evolution.
  • Coarse-Grained Molecular Dynamics (CGMD): The paper also briefly discusses CGMD, where coarse-grained particles directly replace atoms in a region, with their mass being the sum of the replaced atoms. It concludes that CGMD also exhibits a direct correspondence with MD, lacking fundamental disparity from QCGD in this regard.
  • Example Verification: The authors use uniaxial compression of polycrystalline copper and shock loading of polycrystalline aluminum as case studies to empirically support their theoretical findings.

Datasets

Not discussed in the paper. The simulations are based on established interatomic potentials for the materials studied.

Key Insights and Ideas

  • QCGD as a Scaled MD: The most significant finding is the demonstration that QCGD, as formulated by scaling interatomic potentials, is not an independent mesoscale method but rather a direct, scaled-up representation of atomistic MD. The spatiotemporal evolution of QCGD systems is shown to be mathematically equivalent to that of MD systems.
  • Misleading Scale Effects: The paper argues that observed "scale effects" in QCGD simulations (e.g., in shock loading or compression) might be artifacts of the scaling process itself, rather than genuine emergent mesoscale phenomena. The "different grain sizes" or "different shock depths" in QCGD are shown to be merely different scaling factors applied to the same underlying atomistic physics.
  • Loss of True Mesoscale Physics: The direct amplification of MD results means that QCGD might miss crucial information lost during the coarse-graining process, preventing it from capturing true mesoscale physical phenomena like microstructural evolution.
  • Implications for Research: This work suggests that researchers need to be cautious when interpreting QCGD results, especially claims about scale-dependent behavior. Verifying consistency with atomistic simulations is necessary, but not sufficient; the ability to accurately predict mesoscale phenomena must also be rigorously assessed.

Key Results

  • Theoretical Equivalence: Mathematical derivations show a direct correspondence between the spatiotemporal evolution of QCGD and MD systems, implying that QCGD reproduces MD results at different scales.
  • Uniaxial Compression: Simulations of copper under uniaxial compression showed nearly identical stress-strain curves for MD and QCGD systems with different scaling factors ($L_{cg}=10$ and $L_{cg}=100$), even with significantly different strain rates. Minor discrepancies were attributed to floating-point arithmetic and the chaotic nature of simulations.
  • Shock Loading: Simulations of polycrystalline aluminum under shock loading indicated that QCGD results were not significantly different from MD results. The paper argues that observed differences in grain size effects were manifestations of different scaling factors rather than true physical differences. When shock depth was normalized by model size, all QCGD simulations aligned perfectly with MD.

Limitations and Future Work

  • Limitations: The paper's primary conclusion is that QCGD, as currently implemented by scaling potentials, is fundamentally limited in its ability to capture true mesoscale phenomena and scale effects. The direct amplification of MD results means that inherent physical information is lost, and emergent behaviors are not captured.
  • Future Work: The authors suggest that to overcome these limitations, future mesoscale simulation methods might need to:
  • Compensate for lost physical information by incorporating intrinsic physical quantities onto coarse-grained particles.
  • Introduce corrective terms, such as dissipation and random forces, into the dynamics of the coarse-grained model.
  • Meticulously evaluate the capacity of mesoscale methods to accurately capture mesoscale physical phenomena, beyond just verifying consistency with atomistic simulations.

Summary updated

BibTeX
@article{jiang2025equivalence,
  author = {Dong-Dong Jiang and Jian-Li Shao and Pei Wang},
  title = {{Equivalence analysis between quasi-coarse-grained and atomistic simulations}},
  journal = {Comput. Phys. Commun.},
  year = {2025},
  doi = {10.1016/j.cpc.2025.109780},
}
Figure