CO adsorption sites on interstellar water ices explored with machine learning potentials - Binding energy distributions and snowline

Open Access

Table C.1

Training set composition and test set accuracy for the machine learning potential.

Type	Formula	Training (66%)	Energies [meV/atom]		Forces [meV/Å]
			MAE	RMSE	MAE	RMSE
W₃₂	MD	820	0.318	0.416	20.284	28.484
W₇₀	MD	122	0.358	0.426	28.793	39.812
W₄₉	MD	414	0.252	0.304	24.148	33.663
W₂₂−CO	MD / metadynamics	3090	0.390	0.508	22.972	32.734
W₃₇−CO	MD / metadynamics	2195	0.322	0.401	24.762	34.070
W₅₀−CO	active learning	1540	0.589	0.770	24.348	33.548
W₆₀−CO	active learning	140	0.268	0.319	27.271	36.718

TOTAL		8321	0.393	0.524	23.970	33.384

Notes. The training set comprised 66% of the total data, while the test set and validation set each contained 16%.

Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.

Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.

Initial download of the metrics may take a while.