David S Greenberg

Dr. David S Greenberg

orcid

Research Focus & Metrics

Academic Career Length

33Years (1994-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
21 August 2020 in Journal of Open Source Software
Verified
Journal Article
Open access
PDF file
ATAlvaro Tejero-Cantero
Alvaro Tejero-Cantero
JBJan Boelts
Jan Boelts
MDMichael Deistler
Michael Deistler
JLJan-Matthis Lueckmann
Jan-Matthis Lueckmann
CDConor Durkan
Conor Durkan
JMJakob Macke
Jakob Macke

28 June 2021 in Journal of Advances in Modeling Earth Systems
Verified
Journal Article
Open access
PDF file
MNMarcel Nonnenmacher
Marcel Nonnenmacher
DGDavid S Greenberg
David S Greenberg
To understand and predict large, complex, and chaotic systems, Earth scientists build simulators from physical laws. Simulators generalize better to new scenarios, require fewer tunable parameters, and are more interpretable than nonphysical deep learning, but procedures for obtaining their derivatives with respect to their inputs are often unavailable. These missing derivatives limit the application of many important tools for forecasting, model tuning, sensitivity analysis, or subgrid‐scale parametrization. Here, we propose to overcome this limitation with deep emulator networks that learn to calculate the missing derivatives. By training directly on simulation data without analyzing source code or equations, this approach supports simulators in any programming language on any hardware without specialized routines for each case. To demonstrate the effectiveness of our approach, we train emulators on complete or partial system states of the chaotic Lorenz‐96 simulator and evaluate the accuracy of their dynamics and derivatives as a function of integration time and training data set size. We further demonstrate that emulator‐derived derivatives enable accurate 4D‐Var data assimilation and closed‐loop training of parametrizations. These results provide a basis for further combining the parsimony and generality of physical models with the power and flexibility of machine learning.

31 May 2025 in Journal of Advances in Modeling Earth Systems
Verified
Journal Article
Open access
SSShivani Sharma
Shivani Sharma
DGDavid S Greenberg
David S Greenberg
Cloud microphysics has important consequences for climate and weather phenomena, and inaccurate representations can limit forecast accuracy. While atmospheric models increasingly resolve storms and clouds, the accuracy of the underlying microphysics remains limited by computationally expedient bulk moment schemes based on simplifying assumptions. Droplet‐based Lagrangian schemes are more accurate but are underutilized due to their large computational overhead. Machine learning (ML) based schemes can bridge this gap by learning from vast droplet‐based simulation data sets, but have so far struggled to match the accuracy and stability of bulk moment schemes. To address this challenge, we developed SuperdropNet, an ML‐based emulator of the Lagrangian superdroplet simulations. To improve accuracy and stability, we employ multi‐step autoregressive prediction during training, impose physical constraints, and carefully control stochasticity in the training data. Superdropnet predicted hydrometeor states and cloud‐to‐rain transition times more accurately than previous ML emulators, and matched or outperformed bulk moment schemes in many cases. We further carried out detailed analyses to reveal how multistep autoregressive training improves performance, and how the performance of SuperdropNet and other microphysical schemes hydrometeors' mass, number and size distribution. Together our results suggest that ML models can effectively emulate cloud microphysics, in a manner consistent with droplet‐based simulations.

14 July 2023 in Machine Learning: Science and Technology
Verified
Journal Article
Open access
TSTobias Schanz
Tobias Schanz
KMKlas Ove Möller
Klas Ove Möller
SRSaskia Rühl
Saskia Rühl
DGDavid S Greenberg
David S Greenberg
Advances in in situ marine life imaging have significantly increased the size and quality of available datasets, but automatic image analysis has not kept pace. Machine learning has shown promise for image processing, but its effectiveness is limited by several open challenges: the requirement for large expert-labeled training datasets, disagreement among experts, under-representation of various species and unreliable or overconfident predictions. To overcome these obstacles for automated underwater imaging, we combine and test recent developments in deep classifier networks and self-supervised feature learning. We use unlabeled images for pretraining deep neural networks to extract task-relevant image features, allowing learning algorithms to cope with scarcity in expert labels, and carefully evaluate performance in subsequent label-based tasks. Performance on rare classes is improved by applying data rebalancing together with a Bayesian correction to avoid biasing inferred in situ class frequencies. A divergence-based loss allows training on multiple, conflicting labels for the same image, leading to better estimates of uncertainty which we quantify with a novel accuracy measure. Together, these techniques can reduce the required label counts ∼100-fold while maintaining the accuracy of standard supervised training, shorten training time, cope with expert disagreement and reduce overconfidence.

22 October 2024 in Mathematics
Verified
Journal Article
Open access
PDF file
KDKubilay Timur Demir
Kubilay Timur Demir
KLKai Logemann
Kai Logemann
DGDavid S Greenberg
David S Greenberg
Physics-informed neural networks (PINNs) have recently emerged as a promising alternative to traditional numerical methods for solving partial differential equations (PDEs) in fluid dynamics. By using PDE-derived loss functions and auto-differentiation, PINNs can recover solutions without requiring costly simulation data, spatial gridding, or time discretization. However, PINNs often exhibit slow or incomplete convergence, depending on the architecture, optimization algorithms, and complexity of the PDEs. To address these difficulties, a variety of novel and repurposed techniques have been introduced to improve convergence. Despite these efforts, their effectiveness is difficult to assess due to the wide range of problems and network architectures. As a novel test case for PINNs, we propose one-dimensional shallow water equations with closed boundaries, where the solutions exhibit repeated boundary wave reflections. After carefully constructing a reference solution, we evaluate the performance of PINNs across different architectures, optimizers, and special training techniques. Despite the simplicity of the problem for classical methods, PINNs only achieve accurate results after prohibitively long training times. While some techniques provide modest improvements in stability and accuracy, this problem remains an open challenge for PINNs, suggesting that it could serve as a valuable testbed for future research on PINN training techniques and optimization strategies.

1 September 1994 in Computers & Chemistry
Verified
Journal Article
Closed access
DGDavid S Greenberg
David S Greenberg
SISorin Istrail
Sorin Istrail