Qian Ouyang
← Selected research

Polymer properties / Molecular dynamics

Preliminary polymer density screening with MACE

Following three polymer systems from molecular simulation to density estimates, sampling checks, and comparison with available reference values.

Preliminary screening17 September 2026
Density estimates for PE, PMP, and PIB compared with dataset reference values, alongside fluctuations over 25 picoseconds. Mean differences are minus 2.59 percent, plus 3.25 percent, and minus 0.80 percent.
Three preliminary MACE density pilots at 300 K. Reference values are reported as experimental by the source dataset; their phase, temperature, molecular weight, and uncertainty are incompletely specified. Sampling intervals do not include independent-packing or experimental uncertainty.
Open full-size figure Download figure data (CSV)

Research question

What can short MLIP-based polymer simulations establish about density, and how does reference-data matching affect the interpretation?

Approach

  • Simulated polyethylene (PE), poly(4-methyl-1-pentene) (PMP), and polyisobutylene (PIB) using MACE-MH-1, the omol head, and float32 precision.
  • Used NPT conditions with targets of 300 K and 1.01325 bar, one initial packing per polymer, and 25 ps of target sampling.
  • Compared simulation means and temporal sampling intervals with dataset reference values reported as experimental. The reference table is external data, not measurements made in this work.

What the measurements show

  • Recorded simulation means were 0.9120 g/cm³ for PE, 0.8590 g/cm³ for PMP, and 0.9047 g/cm³ for PIB.
  • Differences from the available reference table were −2.59%, +3.25%, and −0.80%, respectively.
  • The three pilots passed the workflow's numerical and sampling checks, while remaining preliminary screening results rather than production-qualified experimental validation.

This is a snapshot of three completed pilots as of 17 September 2026, not a statement that a larger polymer collection has been fully evaluated.

Scope & limitations

  • One initial packing per polymer does not measure variation across independent preparations. Temporal intervals cover within-trajectory sampling only.
  • Reference phase, temperature, molecular weight, and measurement uncertainty are incompletely specified, preventing a fully matched experimental comparison.
  • The underlying records remain ineligible for production-level scientific reuse. These three differences must not be promoted as general predictive accuracy.

Data attribution

PolyGraphMT dataset and repository Context for the density reference table. Individual measurement provenance and conditions are not established by the table used here.

Continue exploring

GPU resource trade-offs in MACE molecular dynamics

Read the case study