We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Neural-network force field backed nested sampling: Study of the silicon <mml:math xmlns:mml="http://www.w3.org/1998/Math/MathML"><mml:mrow><mml:mi>p</mml:mi><mml:mtext>−</mml:mtext><mml:mi>T</mml:mi></mml:mrow></mml:math> phase diagram
Summary
Researchers combined a neural-network interatomic potential for silicon with nested sampling to predict silicon's low-pressure phase diagram, evaluating the method's accuracy against experimental data. The model accurately reproduced the melting line and identified thermodynamically stable structures, demonstrating that neural-network force fields paired with nested sampling can efficiently predict phase diagrams while assessing exchange-correlation functional accuracy.
Phase diagrams map out the thermodynamically stable conditions for different phases. Their predictive atomistic simulations demands integration of statistical mechanics and quantum mechanics, making computational cost a challenge. Our work successfully combines a neural-network model for silicon's potential energy with nested-sampling to predict its low-pressure phase diagram accurately. Trained on diverse silicon structures, the model aligns remarkably well with experiments, accurately reproducing the melting line and identifying stable structures. The fusion of neural networks and nested sampling opens the door to not only predict phase diagrams but also assess the accuracy of the underlying exchange-correlation functionals.
Sign in to start a discussion.