yi-fan chen

Followers

Following

Public Views

Interests

Uploads

Papers by yi-fan chen

Accelerating large-eddy simulations of clouds with Tensor Processing Units

Authorea (Authorea), Feb 9, 2023

Clouds, especially low clouds, are crucial for regulating Earth's energy balance and mediating th... more Clouds, especially low clouds, are crucial for regulating Earth's energy balance and mediating the response of the climate system to changes in greenhouse gas concentrations. Despite their importance for climate, they remain relatively poorly understood and are inaccurately represented in climate models. A principal reason is that the high computational expense of simulating them with large-eddy simulations (LES) has inhibited broad and systematic numerical experimentation and the generation of large datasets for training parametrization schemes for climate models. Here we demonstrate LES of low clouds on Tensor Processing Units (TPUs), application-specific integrated circuits that were originally developed for machine learning applications. We show that TPUs in conjunction with tailored software implementations can be used to simulate computationally challenging stratocumulus clouds in conditions observed during the Dynamics and Chemistry of Marine Stratocumulus (DYCOMS) field study. The TPU-based LES code successfully reproduces clouds during DYCOMS and opens up the large computational resources available on TPUs to cloud simulations. The code enables unprecedented weak and strong scaling of LES, making it possible, for example, to simulate stratocumulus with $10\times$ speedup over real-time evolution in domains with a $34.7 \mathrm{km} \times 53.8 \mathrm{km}$ horizontal cross section. The results open up new avenues for computational experiments and for substantially enlarging the sample of LES available to train parameterizations of low clouds.

Download

Scalable Realistic Recommendation Datasets through Fractal Expansions

ArXiv, 2019

Recommender System research suffers currently from a disconnect between the size of academic data... more Recommender System research suffers currently from a disconnect between the size of academic data sets and the scale of industrial production systems. In order to bridge that gap we propose to generate more massive user/item interaction data sets by expanding pre-existing public data sets. User/item incidence matrices record interactions between users and items on a given platform as a large sparse matrix whose rows correspond to users and whose columns correspond to items. Our technique expands such matrices to larger numbers of rows (users), columns (items) and non zero values (interactions) while preserving key higher order statistical properties. We adapt the Kronecker Graph Theory to user/item incidence matrices and show that the corresponding fractal expansions preserve the fat-tailed distributions of user engagements, item popularity and singular value spectra of user/item interaction matrices. Preserving such properties is key to building large realistic synthetic data sets ...

Download

An Efficient Simulation-Based Travel Demand Calibration Algorithm for Large-Scale Metropolitan Traffic Models

arXiv (Cornell University), Sep 23, 2021

Metropolitan scale vehicular traffic modeling is used by a variety of private and public sector u... more Metropolitan scale vehicular traffic modeling is used by a variety of private and public sector urban mobility stakeholders to inform the design and operations of road networks. High-resolution stochastic traffic simulators are increasingly used to describe detailed demand-supply interactions. The design of efficient calibration techniques remains a major challenge. This paper considers a class of high-dimensional calibration problems known as origin-destination (OD) calibration. We formulate the problem as a continuous simulation-based optimization problem. Our proposed algorithm builds upon recent metamodel methods that tackle the simulation-based problem by solving a sequence of approximate analytical optimization problems, which rely on the use of analytical network models. In this paper, we formulate a network model defined as a system of linear equations, the dimension of which scales linearly with the number of roads with field data and independently of the dimension of the route choice set. This makes the approach suitable for large-scale metropolitan networks. The approach has enhanced efficiency compared with past metamodel formulations that are based on systems of nonlinear, rather than linear, equations. It also has enhanced efficiency compared to traditional calibration methods that resort to simulation-based estimates of traffic assignment matrices, while the proposed approach uses analytical approximations of these matrices. We benchmark the approach considering a peak period Salt Lake City case study and calibrate based on field vehicular count data. The new formulation yields solutions with good performance and is suitable for large-scale road networks.

Download

Policy Learning and Evaluation with Randomized Quasi-Monte Carlo

arXiv (Cornell University), Feb 15, 2022

Reinforcement learning constantly deals with hard integrals, for example when computing expectati... more Reinforcement learning constantly deals with hard integrals, for example when computing expectations in policy evaluation and policy iteration. These integrals are rarely analytically solvable and typically estimated with the Monte Carlo method, which induces high variance in policy values and gradients. In this work, we propose to replace Monte Carlo samples with low-discrepancy point sets. We combine policy gradient methods with Randomized Quasi-Monte Carlo, yielding variancereduced formulations of policy gradient and actor-critic algorithms. These formulations are effective for policy evaluation and policy improvement, as they outperform state-of-theart algorithms on standardized continuous control benchmarks. Our empirical analyses validate the intuition that replacing Monte Carlo with Quasi-Monte Carlo yields significantly more accurate gradient estimates.

Download

HyperPINN: Learning parameterized differential equations with physics-informed hypernetworks

arXiv (Cornell University), Oct 28, 2021

Many types of physics-informed neural network models have been proposed in recent years as approa... more Many types of physics-informed neural network models have been proposed in recent years as approaches for learning solutions to differential equations. When a particular task requires solving a differential equation at multiple parameterizations, this requires either retraining the model, or expanding its representation capacity to include the parameterization-both solution that increase its computational cost. We propose the HyperPINN, which uses hypernetworks to learn to generate neural networks that can solve a differential equation from a given parameterization. We demonstrate with experiments on both a PDE and an ODE that this type of model can lead to neural network solutions to differential equations that maintain a small size, even when learning a family of solutions over a parameter space. * Work done while at Google Research. Preprint. Under review.

Download

Scaling Up Collaborative Filtering Data Sets through Randomized Fractal Expansions

arXiv (Cornell University), Apr 8, 2019

Recommender system research suffers from a disconnect between the size of academic data sets and ... more Recommender system research suffers from a disconnect between the size of academic data sets and the scale of industrial production systems. In order to bridge that gap, we propose to generate large-scale user/item interaction data sets by expanding pre-existing public data sets. Our key contribution is a technique that expands user/item incidence matrices matrices to large numbers of rows (users), columns (items), and non-zero values (interactions). The proposed method adapts Kronecker Graph Theory to preserve key higher order statistical properties such as the fat-tailed distribution of user engagements, item popularity, and singular value spectra of user/item interaction matrices. Preserving such properties is key to building large realistic synthetic data sets which in turn can be employed reliably to benchmark recommender systems and the systems employed to train them. We further apply our stochastic expansion algorithm to the binarized MovieLens 20M data set, which comprises 20M interactions between 27K movies and 138K users. The resulting expanded data set has 1.2B ratings, 2.2M users, and 855K items, which can be scaled up or down.

Download

High performance Monte Carlo simulation of ising model on TPU clusters

Proceedings of the International Conference for High Performance Computing, Networking, Storage and Analysis, 2019

Large-scale deep learning benefits from an emerging class of AI accelerators. Some of these accel... more Large-scale deep learning benefits from an emerging class of AI accelerators. Some of these accelerators'designs are general enough for compute-intensive applications beyond AI and Cloud TPU is one such example. In this paper, we demonstrate a novel approach using TensorFlow on Cloud TPU to simulate the two-dimensional Ising Model. TensorFlow and Cloud TPU framework enable the simple and readable code to express the complicated distributed algorithm without compromising the performance. Our code implementation fits into a small Jupyter Notebook and fully utilizes Cloud TPU's efficient matrix operation and dedicated high speed inter-chip connection. The performance is highly competitive: it outperforms the best published benchmarks to our knowledge by 60% in single-core and 250% in multi-core with good linear scaling. When compared to Tesla V100 GPU, the single-core performance maintains a ∼10% gain. We also demonstrate that using low precision arithmetic-bfloat16-does not compromise the correctness of the simulation results.

Download

Randomized Fractal Expansions for Production-Scale Public Collaborative-Filtering Data Sets

Recommender system research suffers from a disconnect between the size of academic data sets and ... more Recommender system research suffers from a disconnect between the size of academic data sets and the scale of industrial production systems. In order to bridge that gap, we propose to generate large-scale user/item interaction data sets by expanding pre-existing public data sets. Our key contribution is a technique that expands user/item incidence matrices to large numbers of rows (users), columns (items), and non-zero values (interactions). The proposed method adapts Kronecker Graph Theory to preserve key higher order statistical properties such as the fat-tailed distribution of user engagements, item popularity, and singular value spectra of user/item interaction matrices. Preserving such properties is key to building large realistic synthetic data sets which can be employed reliably to benchmark recommender systems and the systems employed to train them. We further apply our stochastic expansion algorithm to the binarized MovieLens 20M data set, which comprises 20M interactions b...

Download

Distributed Data Processing for Large-Scale Simulations on Cloud

2021 IEEE International Joint EMC/SI/PI and EMC Europe Symposium, 2021

The computational challenges encountered in the large-scale simulations are accompanied by those ... more The computational challenges encountered in the large-scale simulations are accompanied by those from dataintensive computing. In this work, we proposed a distributed data pipeline for large-scale simulations by using libraries and frameworks available on Cloud services. The building blocks of the proposed data pipeline such as Apache Beam and Zarr are commonly used in the data science and machine learning community. Our contribution is to apply the data-science approaches to handle large-scale simulation data for the hardware design community. The data pipeline is designed with careful considerations for the characteristics of the simulation data in order to achieve high parallel efficiency. The performance of the data pipeline is analyzed with two examples. In the first example, the proposed data pipeline is used to process electric potential obtained with a Poisson solver. In the second example, the data pipeline is used to process thermal and fluid data obtained with a computational fluid dynamic solver. Both solvers are in-house developed and finite-difference based, running in parallel on Tensor Processing Unit (TPU) clusters and serving the purpose of data generation. It is worth mentioning that in this work, the focus is on data processing instead of data generation. The proposed data pipeline is designed in a general manner and is suitable for other types of data generators such as fullwave electromagnetic and multiphysics solvers. The performance analysis demonstrates good storage and computational efficiency of the proposed data pipeline. As a reference, it takes 5 hours and 14 mins to convert simulation data of size 7.8 TB into Zarr format and the maximum total parallelism is chosen as 10,000.

Download

$Research paper thumbnail of A right-handed isotropic medium with a negative refractive index$

A right-handed isotropic medium with a negative refractive index

Physical review letters, 2004

The sign of the refractive index of any medium is soley determined by the requirement that the pr... more The sign of the refractive index of any medium is soley determined by the requirement that the propagation of an electromagnetic wave obeys Einstein causality. Our analysis shows that this requirement predicts that the real part of the refractive index may be negative in an isotropic medium even if the electric permittivity and the magnetic permeability are both positive. Such a system may be a route to negative index media at optical frequencies. We also demonstrate that the refractive index may be positive in left-handed media that contain two molecular species where one is in its excited state.

Download

$Research paper thumbnail of Negative Refraction at Optical Frequencies in Nonmagnetic Two-Component Molecular Media$

Negative Refraction at Optical Frequencies in Nonmagnetic Two-Component Molecular Media

Physical Review Letters, 2005

Download

Criteria for the experimental observation of multidimensional optical solitons in saturable media

Physical Review E, 2004

Criteria for experimental observation of multi-dimensional optical solitons in media with saturab... more Criteria for experimental observation of multi-dimensional optical solitons in media with saturable refractive nonlinearities are developed. The criteria are applied to actual material parameters (characterizing the cubic self-focusing and quintic self-defocusing nonlinearities, two-photon loss, and optical-damage threshold) for various glasses. This way, we identify operation windows for soliton formation in these glasses. It is found that two-photon absorption sets stringent limits on the windows. We conclude that, while a well-defined window of parameters exists for two-dimensional solitons (spatial or spatiotemporal), for their three-dimensional spatiotemporal counterparts such a window does not exist, due to the nonlinear loss in glasses.

Download

Controllable Raman-like nonlinearities from nonstationary, cascaded quadratic processes

Journal of the Optical Society of America B, 2004

We show that useful non-instantaneous nonlinear phase shifts can be obtained from cascaded quadra... more We show that useful non-instantaneous nonlinear phase shifts can be obtained from cascaded quadratic processes in the presence of group velocity mismatch. The two-field nature of the process permits responses that can be effectively advanced or retarded in time with respect to one of the fields. There is an analogy to a generalized Raman-scattering effect, permitting both red and blue shifts of short pulses. We expect this capability to have many applications in short-pulse generation and propagation, such as the compensation of Ramaninduced effects and high-quality pulse compression, which we discuss.

Download

Chen, Fischer, and Wise Reply

Physical Review Letters, 2006

Chen, Fischer, and Wise Reply: Yi-Fan Chen * , Peer Fischer , and Frank W. Wise Department of... more

$Research paper thumbnail of Sign of the refractive index in a gain medium with negative permittivity and permeability$

Sign of the refractive index in a gain medium with negative permittivity and permeability

JOSA B, 2006

We show how the sign of the refractive index in any medium may be derived using a rigorous analys... more

yi-fan chen

Uploads

Papers by yi-fan chen

Log In