Table 2 Configuration dataset and training scheme of EMFF-2025 model

Structure	Iterations^a	Scaling factor^b	Temperature (K)	Sampling no. (AIMD/DP-GEN)^c
RDX HMX CL-20	-	0.92, 0.96, 1.00, 1.04, 1.08	300–4000 300–4000 300–4000	5000/4000 5000/4000 5000/4000
TNT	1–5	1.00	300–4000	0/800
ADN	5–9	1.00	300–4000	0/400
FOX-7	10–11	1.00	300–4000	0/400
TKX-50	12–13	1.00	300–4000	0/400
DNBF	14–16	1.00	300–4000	0/300
BTF	17–19	1.00	300–4000	0/300
TATB	20–21	1.00	300–4000	0/200
TAGN	22–23	1.00	300–4000	0/200
NG	24–25	1.00	300–4000	0/200
PETN	26–27	1.00	300–4000	0/500
DTTO/iso-DTTO ^d	28–30	1.00	300–4000	0/200
NTO	31–33	1.00	300–4000	0/200
TEX	34–35	1.00	300–4000	0/200
BTTN	36–38	1.00	300–4000	0/500
NC	39–40	1.00	300–4000	0/500
TNB	41–42	1.00	300–4000	0/800
HNS	43–45	1.00	300–4000	0/800

^aIn the DP-GEN iterative process, the first iteration for each HEM involves sampling over the 300–4000 K temperature range, while the subsequent iterations are used for training the NNP model.
^bStructural coordinates were transformed according to the scaling factors in the x, y, and z directions.
^cAIMD simulations and DP-GEN processes within the foundational DFT dataset were sampled across various temperatures (300–4000 K).
^diso-DTTO refers to isomeric forms of DTTO, where the molecular structure differs in the arrangement of atoms or groups but retains the same molecular formula.

Quick links

Search