Reasoning4PDEs
Towards Reasoning for PDE Foundation Models: A Reward-Model-Driven Inference-Time-Scaling Algorithm
Mansingh, Siddharth↗
Engineering topics
Mohan, Arvind
Publications and source records attributed to Mohan, Arvind.
Development of the Senseiver for efficient field reconstruction from sparse observations
Abstract The reconstruction of complex time-evolving fields from sensor observations is a grand challenge. Frequently, sensors have extremely sparse coverage and low-resource computing capacity for measuring highly nonlinear phenomena. While numerical simulations can model some of these phenomena using partial differential equations, the reconstruction problem is ill-posed. Data-driven-strategies provide crucial disambiguation, but these suffer in cases with small amounts of data, and struggle to handle large domains. Here we present the Senseiver, an attention-based framework that excels in reconstructing complex spatial fields from few observations with low overhead. The Senseiver reconstructs n -dimensional fields by encoding arbitrarily sized sparse sets of inputs into a latent space using cross-attention, producing uniform-sized outputs regardless of the number of observations. This allows efficient inference by decoding only a sparse set of output observations, while a dense set of observations is needed to train. This framework enables training of data with complex boundary conditions and extremely large fine-scale simulations. We build on the Perceiver IO by enabling training models with fewer parameters, which facilitates field deployment, and a training framework that allows a flexible number of sensors as input, which is critical for real-world applications. We show that the Senseiver advances the state-of-the-art of field reconstruction in many applications.
97 MATHEMATICS AND COMPUTING↗
Quasi‐dynamic breathing model of the lung incorporating viscoelasticity of the lung tissue
Abstract We advanced a novel model to calculate viscoelastic lung compliance and airflow resistance in presence of mucus, accounting for the quasi‐linear viscoelastic stress–strain response of the parenchyma (alveoli) tissue. We adapted a continuum‐based numerical modeling approach for the lung, integrating the fluid mechanics of the airflow within individual generations of the bronchi and alveoli. The model accounts for elasticity of the deformable bronchioles, resistance to airflow due to the presence of mucus within the bronchioles, and subsequent mucus flow. Simulated quasi‐dynamic inhalation and expiration cycles were used to characterize the net compliance and resistance of the lung, considering the rheology of the mucus and viscoelastic properties of the parenchyma tissue. The structure and material properties of the lung were identified to have an important contribution to the lung compliance and airflow resistance. The secondary objective of this work was to assess whether a higher frequency and smaller volume of harmonic air flow rate compared to a normal ventilator breathing cycle enhanced mucus outflow. Results predict, lower mucus viscosity and higher excitation frequency of breathing are favorable for the flow of mucus up the bronchi tree, towards the trachea.
60 APPLIED LIFE SCIENCES↗
A Multi-Scale Inference, Estimation, and Prediction Engine for Earth System Modeling
We posit that AI methods can be leveraged to significantly enhance the predictive skill of forward Earth system modeling (ESM) activities. A hybrid framework incorporating traditional ESM modeling, inference methods, and AI techniques could make better use of both measured and computed information as well as computational resources by targeting inference tasks at program priorities, such as the predictability of precipitation extremes. This runtime pathway to closing the simulation/analysis-data/model improvement loop will streamline the traditional offline pathway to model improvement, which is based on domain science expertise, while suggesting guidance for further observations and measurements.
58 GEOSCIENCES↗
Multiscale Reduced Order Modeling and Parameter Estimation for Climate Sciences
Several problems in earth system modeling are dependent on highly multiscale phenomena, such as turbulence, where computational modeling is challenging and expensive. This issue is exacerbated in atmospheric and oceanic domains, due to inherent high-dimensionality of the problem. One approach to this problem has been reduced order modeling (ROM); which aims to represent the key physics of the phenomena as a low-dimensional system. AI methods have huge potential in building accurate, stable ROMs and parameter estimation for these ROMs, as it requires extracting nonlinearities and patterns from simulation and/or observational data. Developing physics-based AI approaches specialized for the complexities of multiscale data, along with strategies to account for uncertainties, will revolutionize rapid modeling, analysis and decision making for earth system problems of practical interest.
58 GEOSCIENCES↗