-
GlobalTomo: A global dataset for physics-ML seismic wavefield modeling and FWI
Authors:
Shiqian Li,
Zhi Li,
Zhancun Mu,
Shiji Xin,
Zhixiang Dai,
Kuangdai Leng,
Ruihua Zhang,
Xiaodong Song,
Yixin Zhu
Abstract:
Global seismic tomography, taking advantage of seismic waves from natural earthquakes, provides essential insights into the earth's internal dynamics. Advanced Full-waveform Inversion (FWI) techniques, whose aim is to meticulously interpret every detail in seismograms, confront formidable computational demands in forward modeling and adjoint simulations on a global scale. Recent advancements in Ma…
▽ More
Global seismic tomography, taking advantage of seismic waves from natural earthquakes, provides essential insights into the earth's internal dynamics. Advanced Full-waveform Inversion (FWI) techniques, whose aim is to meticulously interpret every detail in seismograms, confront formidable computational demands in forward modeling and adjoint simulations on a global scale. Recent advancements in Machine Learning (ML) offer a transformative potential for accelerating the computational efficiency of FWI and extending its applicability to larger scales. This work presents the first 3D global synthetic dataset tailored for seismic wavefield modeling and full-waveform tomography, referred to as the GlobalTomo dataset. This dataset is uniquely comprehensive, incorporating explicit wave physics and robust geophysical parameterization at realistic global scales, generated through state-of-the-art forward simulations optimized for 3D global wavefield calculations. Through extensive analysis and the establishment of ML baselines, we illustrate that ML approaches are particularly suitable for global FWI, overcoming its limitations with rapid forward modeling and flexible inversion strategies. This work represents a cross-disciplinary effort to enhance our understanding of the earth's interior through physics-ML modeling.
△ Less
Submitted 19 May, 2025; v1 submitted 26 June, 2024;
originally announced June 2024.
-
Zero Coordinate Shift: Whetted Automatic Differentiation for Physics-informed Operator Learning
Authors:
Kuangdai Leng,
Mallikarjun Shankar,
Jeyan Thiyagalingam
Abstract:
Automatic differentiation (AD) is a critical step in physics-informed machine learning, required for computing the high-order derivatives of network output w.r.t. coordinates of collocation points. In this paper, we present a novel and lightweight algorithm to conduct AD for physics-informed operator learning, which we call the trick of Zero Coordinate Shift (ZCS). Instead of making all sampled co…
▽ More
Automatic differentiation (AD) is a critical step in physics-informed machine learning, required for computing the high-order derivatives of network output w.r.t. coordinates of collocation points. In this paper, we present a novel and lightweight algorithm to conduct AD for physics-informed operator learning, which we call the trick of Zero Coordinate Shift (ZCS). Instead of making all sampled coordinates as leaf variables, ZCS introduces only one scalar-valued leaf variable for each spatial or temporal dimension, simplifying the wanted derivatives from "many-roots-many-leaves" to "one-root-many-leaves" whereby reverse-mode AD becomes directly utilisable. It has led to an outstanding performance leap by avoiding the duplication of the computational graph along the dimension of functions (physical parameters). ZCS is easy to implement with current deep learning libraries; our own implementation is achieved by extending the DeepXDE package. We carry out a comprehensive benchmark analysis and several case studies, training physics-informed DeepONets to solve partial differential equations (PDEs) without data. The results show that ZCS has persistently reduced GPU memory consumption and wall time for training by an order of magnitude, and such reduction factor scales with the number of functions. As a low-level optimisation technique, ZCS imposes no restrictions on data, physics (PDE) or network architecture and does not compromise training results from any aspect.
△ Less
Submitted 14 March, 2024; v1 submitted 1 November, 2023;
originally announced November 2023.
-
Multilayer GZ/YSZ Thermal Barrier Coating from Suspension and Solution Precursor Plasma Spray
Authors:
Kah Leng,
Acacio Rincon Romero,
Tanvir Hussain
Abstract:
Gas turbines rely on thermal barrier coating (TBC) to thermally insulate the nickel-based superalloys underneath during operation; however, current TBCs, yttria stabilised zirconia (YSZ), limit the operating temperature and hence efficiency. At an operating temperature above 1200°C, YSZ is susceptible to failure due to phase instabilities and CMAS (Calcia-Magnesia-Alumina-Silica) attack. Gadoliniu…
▽ More
Gas turbines rely on thermal barrier coating (TBC) to thermally insulate the nickel-based superalloys underneath during operation; however, current TBCs, yttria stabilised zirconia (YSZ), limit the operating temperature and hence efficiency. At an operating temperature above 1200°C, YSZ is susceptible to failure due to phase instabilities and CMAS (Calcia-Magnesia-Alumina-Silica) attack. Gadolinium zirconates (GZ) could overcome the drawback of YSZ, complementing each other with the multi-layer approach. This study introduces a novel approach utilising axial suspension plasma spray (ASPS) and axial solution precursor plasma spray (ASPPS) to produce a double-layer and a triple-layer TBCs with improved CMAS resistance. The former comprised suspension plasma sprayed GZ and YSZ layers while the latter had an additional dense layer deposited through a solution precursor to minimise the columnar gaps that pre-existed in the SPS GZ layer, thus resisting CMAS infiltration. Both coatings performed similarly in furnace cycling test (FCT) and burner rig testing (BRT). In the CMAS test, triple-layer coating showed better CMAS resistivity, as evidenced by the limited CMAS infiltration observed on the surface.
△ Less
Submitted 5 June, 2023;
originally announced June 2023.
-
On the Compatibility between Neural Networks and Partial Differential Equations for Physics-informed Learning
Authors:
Kuangdai Leng,
Jeyan Thiyagalingam
Abstract:
We shed light on a pitfall and an opportunity in physics-informed neural networks (PINNs). We prove that a multilayer perceptron (MLP) only with ReLU (Rectified Linear Unit) or ReLU-like Lipschitz activation functions will always lead to a vanished Hessian. Such a network-imposed constraint contradicts any second- or higher-order partial differential equations (PDEs). Therefore, a ReLU-based MLP c…
▽ More
We shed light on a pitfall and an opportunity in physics-informed neural networks (PINNs). We prove that a multilayer perceptron (MLP) only with ReLU (Rectified Linear Unit) or ReLU-like Lipschitz activation functions will always lead to a vanished Hessian. Such a network-imposed constraint contradicts any second- or higher-order partial differential equations (PDEs). Therefore, a ReLU-based MLP cannot form a permissible function space for the approximation of their solutions. Inspired by this pitfall, we prove that a linear PDE up to the $n$-th order can be strictly satisfied by an MLP with $C^n$ activation functions when the weights of its output layer lie on a certain hyperplane, as called the out-layer-hyperplane. An MLP equipped with the out-layer-hyperplane becomes "physics-enforced", no longer requiring a loss function for the PDE itself (but only those for the initial and boundary conditions). Such a hyperplane exists not only for MLPs but for any network architecture tailed by a fully-connected hidden layer. To our knowledge, this should be the first PINN architecture that enforces point-wise correctness of PDEs. We show a closed-form expression of the out-layer-hyperplane for second-order linear PDEs, which can be generalised to higher-order nonlinear PDEs.
△ Less
Submitted 1 April, 2023; v1 submitted 30 November, 2022;
originally announced December 2022.
-
Local energy landscape drives long exciton diffusion in 2D halide perovskite semiconductors
Authors:
Alan Baldwin,
Géraud Delport,
Kai Leng,
Rosemonde Chahbazian,
Krzysztof Galkowski,
Kian Ping Loh,
Samuel D. Stranks
Abstract:
Halide perovskites have emerged as disruptive semiconductors for applications including photovoltaics and light emitting devices, with modular optoelectronic properties realisable through composition and dimensionality tuning. Layered Ruddlesden-Popper perovskites of the form BA2MAn-1PbnI3n+1, where n is the number of lead-halide and methylammonium (MA) sheets spaced by longer butylammonium (BA) c…
▽ More
Halide perovskites have emerged as disruptive semiconductors for applications including photovoltaics and light emitting devices, with modular optoelectronic properties realisable through composition and dimensionality tuning. Layered Ruddlesden-Popper perovskites of the form BA2MAn-1PbnI3n+1, where n is the number of lead-halide and methylammonium (MA) sheets spaced by longer butylammonium (BA) cations, are particularly interesting due to their unique two-dimensional character and charge carrier dynamics dominated by strongly bound excitons. However, long-range energy transport through exciton diffusion in these materials is not understood or realised. Here, we employ local time-resolved luminescence mapping techniques to visualise exciton transport in high-quality exfoliated flakes of the BA2MAn-1PbnI3n+1 perovskite family. We uncover two distinct transport regimes, depending on the temperature range studied. At temperatures above 100 K, diffusion is mediated by thermally activated hopping processes between localised states. At lower temperatures, a non-uniform energetic landscape emerges in which exciton transport is dominated by energy funnelling processes to lower energy states, leading to long range transport over hundreds of nanometres even in the absence of exciton-phonon coupling and in the presence of local optoelectronic heterogeneity. Efficient, long-range and switchable excitonic funnelling offers exciting possibilities of controlled directional long-range transport in these 2D materials for new device applications.
△ Less
Submitted 29 October, 2020;
originally announced October 2020.
