An official website of the United States government
We study data-driven representations for three- dimensional triangle meshes, which are one of the prevalent objects used to represent 3D geometry. Recent works have developed models that exploit the intrinsic geometry of manifolds and graphs, namely the Graph Neural Networks (GNNs) and its spectral variants, which learn from the local metric tensor via the Laplacian operator. Despite offering excellent sample complexity and built-in invariances, intrinsic geometry alone is invariant to isometric deformations, making it unsuitable for many applications. To overcome this limitation, we propose several upgrades to GNNs to leverage extrinsic differential geometry properties of three-dimensional surfaces, increasing its modeling power. In particular, we propose to exploit the Dirac operator, whose spectrum detects principal curva- ture directions — this is in stark contrast with the classical Laplace operator, which directly measures mean curvature. We coin the resulting models Surface Networks (SN). We prove that these models define shape representations that are stable to deformation and to discretization, and we demonstrate the efficiency and versatility of SNs on two challenging tasks: temporal prediction of mesh deformations under non-linear dynamics and generative models using a variational autoencoder framework with encoders/decoders given by SNs.
more »
« less
Optimum surface-passivation schemes for near-surface spin defects in silicon carbide
- Award ID(s):
- 1738076
- PAR ID:
- 10509250
- Publisher / Repository:
- American Physical Society
- Date Published:
- Journal Name:
- Physical Review Materials
- Volume:
- 8
- Issue:
- 5
- ISSN:
- 2475-9953; PRMHAR
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
Free surface flows driven by boundary undulations are observed in many biological phenomena, including the feeding and locomotion of water snails. To simulate the feeding strategy of apple snails, we develop a centimetric robotic undulator that drives a thin viscous film of liquid with the wave speed$$V_w$$. Our experimental results demonstrate that the behaviour of the net fluid flux$$Q$$strongly depends on the Reynolds number$$Re$$. Specifically, in the limit of vanishing$$Re$$, we observe that$$Q$$varies non-monotonically with$$V_w$$, which has been successfully rationalised by Pandeyet al.(Nat. Commun., vol. 14, no. 1, 2023, p. 7735) with the lubrication model. By contrast, in the regime of finite inertia ($${Re} \sim O(1)$$), the fluid flux continues to increase with$$V_w$$and completely deviates from the prediction of lubrication theory. To explain the inertia-enhanced pumping rate, we build a thin-film, two-dimensional model via the asymptotic expansion in which we linearise the effects of inertia. Our model results match the experimental data with no fitting parameters and also show the connection to the corresponding free surface shapes$$h_2$$. Going beyond the experimental data, we derive analytical expressions of$$Q$$and$$h_2$$, which allow us to decouple the effects of inertia, gravity, viscosity and surface tension on free surface pumping over a wide range of parameter space.more » « less
