Showing 51–100 of 129 results for author: Guha, A

Outlier-Robust Optimal Transport

Authors: Debarghya Mukherjee, Aritra Guha, Justin Solomon, Yuekai Sun, Mikhail Yurochkin

Abstract: Optimal transport (OT) measures distances between distributions in a way that depends on the geometry of the sample space. In light of recent advances in computational OT, OT distances are widely used as loss functions in machine learning. Despite their prevalence and advantages, OT loss functions can be extremely sensitive to outliers. In fact, a single adversarially-picked outlier can increase t… ▽ More Optimal transport (OT) measures distances between distributions in a way that depends on the geometry of the sample space. In light of recent advances in computational OT, OT distances are widely used as loss functions in machine learning. Despite their prevalence and advantages, OT loss functions can be extremely sensitive to outliers. In fact, a single adversarially-picked outlier can increase the standard $W_2$-distance arbitrarily. To address this issue, we propose an outlier-robust formulation of OT. Our formulation is convex but challenging to scale at a first glance. Our main contribution is deriving an \emph{equivalent} formulation based on cost truncation that is easy to incorporate into modern algorithms for computational OT. We demonstrate the benefits of our formulation in mean estimation problems under the Huber contamination model in simulations and outlier detection tasks on real data. △ Less

Submitted 20 June, 2021; v1 submitted 14 December, 2020; originally announced December 2020.

Comments: Accepted in Proceedings of the 38th International Conference on Machine Learning, PMLR 139, 2021

MRAC-RL: A Framework for On-Line Policy Adaptation Under Parametric Model Uncertainty

Authors: Anubhav Guha, Anuradha Annaswamy

Abstract: Reinforcement learning (RL) algorithms have been successfully used to develop control policies for dynamical systems. For many such systems, these policies are trained in a simulated environment. Due to discrepancies between the simulated model and the true system dynamics, RL trained policies often fail to generalize and adapt appropriately when deployed in the real-world environment. Current res… ▽ More Reinforcement learning (RL) algorithms have been successfully used to develop control policies for dynamical systems. For many such systems, these policies are trained in a simulated environment. Due to discrepancies between the simulated model and the true system dynamics, RL trained policies often fail to generalize and adapt appropriately when deployed in the real-world environment. Current research in bridging this sim-to-real gap has largely focused on improvements in simulation design and on the development of improved and specialized RL algorithms for robust control policy generation. In this paper we apply principles from adaptive control and system identification to develop the model-reference adaptive control & reinforcement learning (MRAC-RL) framework. We propose a set of novel MRAC algorithms applicable to a broad range of linear and nonlinear systems, and derive the associated control laws. The MRAC-RL framework utilizes an inner-loop adaptive controller that allows a simulation-trained outer-loop policy to adapt and operate effectively in a test environment, even when parametric model uncertainty exists. We demonstrate that the MRAC-RL approach improves upon state-of-the-art RL algorithms in develo** control policies that can be applied to systems with modeling errors. △ Less

Submitted 20 November, 2020; originally announced November 2020.

Comments: Short version submitted to Learning for Dynamics & Control (L4DC) 2021 Conference

A New Bound on Odd Multicrossing Numbers of Knots and Links

Authors: Anshul Guha

Abstract: An $n$-crossing projection of a link $L$ is a projection of $L$ onto a plane such that $n$ points on $L$ are superimposed on top of each other at every crossing. We prove that for all $k \in \mathbb{N}$ and all links $L$, the inequality $$c_{2k+1}(L) \geq \frac{2g(L) + r(L)-1}{k^2}$$ holds, where $c_{2k+1}(L)$, $g(L)$, and $r(L)$ are the $(2k+1)$-crossing number, $3$-genus, and number of component… ▽ More An $n$-crossing projection of a link $L$ is a projection of $L$ onto a plane such that $n$ points on $L$ are superimposed on top of each other at every crossing. We prove that for all $k \in \mathbb{N}$ and all links $L$, the inequality $$c_{2k+1}(L) \geq \frac{2g(L) + r(L)-1}{k^2}$$ holds, where $c_{2k+1}(L)$, $g(L)$, and $r(L)$ are the $(2k+1)$-crossing number, $3$-genus, and number of components of $L$ respectively. This result is used to prove a new bound on the odd crossing numbers of torus knots and generalizes a result of Jablonowski. We also prove a new upper bound on the $5$-crossing numbers of the 2-torus knots and links. Furthermore, we improve the lower bounds on the $5$-crossing numbers of $79$ knots with $2$-crossing number $ \leq 12$. Finally, we improve the lower bounds on the $7$-crossing numbers of $5$ knots with $2$-crossing number $\leq 12$. △ Less

Submitted 29 October, 2020; originally announced October 2020.

Comments: 16 pages

Wasm/k: Delimited Continuations for WebAssembly

Authors: Donald Pinckney, Arjun Guha, Yuriy Brun

Abstract: WebAssembly is designed to be an alternative to JavaScript that is a safe, portable, and efficient compilation target for a variety of languages. The performance of high-level languages depends not only on the underlying performance of WebAssembly, but also on the quality of the generated WebAssembly code. In this paper, we identify several features of high-level languages that current approaches… ▽ More WebAssembly is designed to be an alternative to JavaScript that is a safe, portable, and efficient compilation target for a variety of languages. The performance of high-level languages depends not only on the underlying performance of WebAssembly, but also on the quality of the generated WebAssembly code. In this paper, we identify several features of high-level languages that current approaches can only compile to WebAssembly by generating complex and inefficient code. We argue that these problems could be addressed if WebAssembly natively supported first-class continuations. We then present Wasm/k, which extends WebAssembly with delimited continuations. Wasm/k introduces no new value types, and thus does not require significant changes to the WebAssembly type system (validation). Wasm/k is safe, even in the presence of foreign function calls (e.g., to and from JavaScript). Finally, Wasm/k is amenable to efficient implementation: we implement Wasm/k as a local change to Wasmtime, an existing WebAssembly JIT. We evaluate Wasm/k by implementing C/k, which adds delimited continuations to C/C++. C/k uses Emscripten and its implementation serves as a case study on how to use Wasm/k in a compiler that targets WebAssembly. We present several case studies using C/k, and show that on implementing green threads, it can outperform the state-of-the-art approach Asyncify with an 18% improvement in performance and a 30% improvement in code size. △ Less

Submitted 4 October, 2020; originally announced October 2020.

Journal ref: Proceedings of the ACM SIGPLAN International Symposium on Dynamic Languages (DLS 2020)

Accelerating Graph Sampling for Graph Machine Learning using GPUs

Authors: Abhinav Jangda, Sandeep Polisetty, Arjun Guha, Marco Serafini

Abstract: Representation learning algorithms automatically learn the features of data. Several representation learning algorithms for graph data, such as DeepWalk, node2vec, and GraphSAGE, sample the graph to produce mini-batches that are suitable for training a DNN. However, sampling time can be a significant fraction of training time, and existing systems do not efficiently parallelize sampling. Samplin… ▽ More Representation learning algorithms automatically learn the features of data. Several representation learning algorithms for graph data, such as DeepWalk, node2vec, and GraphSAGE, sample the graph to produce mini-batches that are suitable for training a DNN. However, sampling time can be a significant fraction of training time, and existing systems do not efficiently parallelize sampling. Sampling is an embarrassingly parallel problem and may appear to lend itself to GPU acceleration, but the irregularity of graphs makes it hard to use GPU resources effectively. This paper presents NextDoor, a system designed to effectively perform graph sampling on GPUs. NextDoor employs a new approach to graph sampling that we call transit-parallelism, which allows load balancing and caching of edges. NextDoor provides end-users with a high-level abstraction for writing a variety of graph sampling algorithms. We implement several graph sampling applications, and show that NextDoor runs them orders of magnitude faster than existing systems. △ Less

Submitted 10 May, 2021; v1 submitted 14 September, 2020; originally announced September 2020.

Comments: Published in EuroSys 2021

Multimodal Noisy Segmentation based fragmented burn scars identification in Amazon Rainforest

Authors: Satyam Mohla, Sidharth Mohla, Anupam Guha, Biplab Banerjee

Abstract: Detection of burn marks due to wildfires in inaccessible rain forests is important for various disaster management and ecological studies. The fragmented nature of arable landscapes and diverse crop** patterns often thwart the precise map** of burn scars. Recent advances in remote-sensing and availability of multimodal data offer a viable solution to this map** problem. However, the task to… ▽ More Detection of burn marks due to wildfires in inaccessible rain forests is important for various disaster management and ecological studies. The fragmented nature of arable landscapes and diverse crop** patterns often thwart the precise map** of burn scars. Recent advances in remote-sensing and availability of multimodal data offer a viable solution to this map** problem. However, the task to segment burn marks is difficult because of its indistinguishably with similar looking land patterns, severe fragmented nature of burn marks and partially labelled noisy datasets. In this work we present AmazonNET -- a convolutional based network that allows extracting of burn patters from multimodal remote sensing images. The network consists of UNet: a well-known encoder decoder type of architecture with skip connections commonly used in biomedical segmentation. The proposed framework utilises stacked RGB-NIR channels to segment burn scars from the pastures by training on a new weakly labelled noisy dataset from Amazonia. Our model illustrates superior performance by correctly identifying partially labelled burn scars and rejecting incorrectly labelled samples, demonstrating our approach as one of the first to effectively utilise deep learning based segmentation models in multimodal burn scar identification. △ Less

Submitted 9 September, 2020; originally announced September 2020.

Comments: 5 pages, 5 figures. Accepted at IEEE International Conference on Systems, Man and Cybernetics 2020. Earlier draft presented at Harvard CRCS AI for Social Good Workshop 2020

Journal ref: 2020 IEEE International Conference on Systems, Man, and Cybernetics (SMC)

Pairs of surface wave packets with zero-sum energy in the Hawking radiation analog

Authors: Anirban Guha, Eyal Heifetz, Akanksha Gupta

Abstract: Here we propose a minimal analog gravity setup and suggest how to select two surface gravity wave packets in order to mimic some key aspects of Hawking radiation from the horizon of non-rotating black holes. Our proposed setup, unlike the scattering problem conventionally studied, constitutes of a constant mean flow over a flat bathymetry, in which the two wave packets possess the same amount of w… ▽ More Here we propose a minimal analog gravity setup and suggest how to select two surface gravity wave packets in order to mimic some key aspects of Hawking radiation from the horizon of non-rotating black holes. Our proposed setup, unlike the scattering problem conventionally studied, constitutes of a constant mean flow over a flat bathymetry, in which the two wave packets possess the same amount of wave action but equal and opposite (sign) amount of energy, thereby mimicking virtual particles created out of near horizon vacuum fluctuations. Attention is given to the physical mechanism relating to the signs of the wave action and energy norm with the wave's intrinsic and total phase speeds. We construct narrow wave packets of equal wave action, the one with positive energy and group speed propagates against the mean flow and escapes from the black hole as Hawking radiation, while the other with negative energy and group speed is drifted by the mean flow and falls into it. Hawking's prediction of low frequency mode amplification is satisfied in our minimal model by construction. We find that the centroid wavenumbers and surface elevation amplitudes of the wave packets are related by simple analytical expressions. △ Less

Submitted 21 August, 2020; originally announced August 2020.

Journal ref: Phys. Rev. D 102, 104061 (2020)

Robot Action Selection Learning via Layered Dimension Informed Program Synthesis

Authors: Jarrett Holtz, Arjun Guha, Joydeep Biswas

Abstract: Action selection policies (ASPs), used to compose low-level robot skills into complex high-level tasks are commonly represented as neural networks (NNs) in the state of the art. Such a paradigm, while very effective, suffers from a few key problems: 1) NNs are opaque to the user and hence not amenable to verification, 2) they require significant amounts of training data, and 3) they are hard to re… ▽ More Action selection policies (ASPs), used to compose low-level robot skills into complex high-level tasks are commonly represented as neural networks (NNs) in the state of the art. Such a paradigm, while very effective, suffers from a few key problems: 1) NNs are opaque to the user and hence not amenable to verification, 2) they require significant amounts of training data, and 3) they are hard to repair when the domain changes. We present two key insights about ASPs for robotics. First, ASPs need to reason about physically meaningful quantities derived from the state of the world, and second, there exists a layered structure for composing these policies. Leveraging these insights, we introduce layered dimension-informed program synthesis (LDIPS) - by reasoning about the physical dimensions of state variables, and dimensional constraints on operators, LDIPS directly synthesizes ASPs in a human-interpretable domain-specific language that is amenable to program repair. We present empirical results to demonstrate that LDIPS 1) can synthesize effective ASPs for robot soccer and autonomous driving domains, 2) requires two orders of magnitude fewer training examples than a comparable NN representation, and 3) can repair the synthesized ASPs with only a small number of corrections when transferring from simulation to real robots. △ Less

Submitted 12 November, 2020; v1 submitted 10 August, 2020; originally announced August 2020.

Robust Unsupervised Learning of Temporal Dynamic Interactions

Authors: Aritra Guha, Rayleigh Lei, Jiacheng Zhu, XuanLong Nguyen, Ding Zhao

Abstract: Robust representation learning of temporal dynamic interactions is an important problem in robotic learning in general and automated unsupervised learning in particular. Temporal dynamic interactions can be described by (multiple) geometric trajectories in a suitable space over which unsupervised learning techniques may be applied to extract useful features from raw and high-dimensional data measu… ▽ More Robust representation learning of temporal dynamic interactions is an important problem in robotic learning in general and automated unsupervised learning in particular. Temporal dynamic interactions can be described by (multiple) geometric trajectories in a suitable space over which unsupervised learning techniques may be applied to extract useful features from raw and high-dimensional data measurements. Taking a geometric approach to robust representation learning for temporal dynamic interactions, it is necessary to develop suitable metrics and a systematic methodology for comparison and for assessing the stability of an unsupervised learning method with respect to its tuning parameters. Such metrics must account for the (geometric) constraints in the physical world as well as the uncertainty associated with the learned patterns. In this paper we introduce a model-free metric based on the Procrustes distance for robust representation learning of interactions, and an optimal transport based distance metric for comparing between distributions of interaction primitives. These distance metrics can serve as an objective for assessing the stability of an interaction learning algorithm. They are also used for comparing the outcomes produced by different algorithms. Moreover, they may also be adopted as an objective function to obtain clusters and representative interaction primitives. These concepts and techniques will be introduced, along with mathematical properties, while their usefulness will be demonstrated in unsupervised learning of vehicle-to-vechicle interactions extracted from the Safety Pilot database, the world's largest database for connected vehicles. △ Less

Submitted 17 June, 2020; originally announced June 2020.

Powering Electronic Devices from Salt Gradients in AA Battery-Sized Stacks of Hydrogel-Infused Paper

Authors: Anirvan Guha, Trevor J. Kalkus, Thomas B. H. Schroeder, Oliver G. Willis, Michael Mayer

Abstract: Strongly electric fish use gradients of ions within their bodies to generate stunning external electrical discharges; the most powerful of these organisms, the Atlantic torpedo ray, can produce pulses of over 1 kW from its electric organs. Despite extensive study of this phenomenon in nature, the development of artificial power generation schemes based on ion gradients for portable, wearable, or i… ▽ More Strongly electric fish use gradients of ions within their bodies to generate stunning external electrical discharges; the most powerful of these organisms, the Atlantic torpedo ray, can produce pulses of over 1 kW from its electric organs. Despite extensive study of this phenomenon in nature, the development of artificial power generation schemes based on ion gradients for portable, wearable, or implantable human use has remained out of reach. Previously, inspired by the electric eel, we developed an artificial electric organ that generated electricity from ion gradients within stacked hydrogels and, like the eel, was optimized to deliver large voltages that exceeded 100 V. Due to its high internal resistance, the current of this power source was, however, too low to power standard electronics. Here we introduce an artificial electric organ that takes inspiration from the unique morphologies of torpedo rays for maximal current output. This power source uses a hybrid material of hydrogel-infused paper to create, organize, and reconfigure stacks of thin, arbitrarily large gel films both in series and in parallel. The resulting increase in electrical power by almost two orders of magnitude compared to the original eel-inspired design makes it possible to power electronic devices and establishes that biology's mechanism of generating significant electrical power can now be realized from benign and soft materials in a portable size. △ Less

Submitted 28 May, 2020; originally announced May 2020.

Comments: 34 pages, 10 figures, 1 table

Wave interactions in neutrally stable shear layers: regular and singular modes, and non-modal growth

Authors: Eyal Heifetz, Anirban Guha, Jeff Carpenter

Abstract: In a recent letter, Carpenter and Guha (2019) compared the neutral modes of a smooth two dimensional shear profile without an inflection point to the modes of its corresponding piecewise-linear profile. The regular mode in the smooth profile was identified as the one least sensitive to the numerical resolution, while the singular modes displayed high sensitivity. Here we provide a physical interpr… ▽ More In a recent letter, Carpenter and Guha (2019) compared the neutral modes of a smooth two dimensional shear profile without an inflection point to the modes of its corresponding piecewise-linear profile. The regular mode in the smooth profile was identified as the one least sensitive to the numerical resolution, while the singular modes displayed high sensitivity. Here we provide a physical interpretation using a wave interaction approach for understanding the structure and behavior of both the regular and singular modes. The regular modes are the interfacial Rossby waves located at the concentrated mean vorticity gradient of the shear profile. In contrast, the singular modes result from a one way phase-locking interaction between singular vorticity disturbances, passively advected by the mean flow at different levels of the profile, and the interfacial Rossby waves. We show that this one way interaction can also lead to a sustained non-modal growth of the interfacial Rossby waves that cannot be captured by standard eigenvalue analysis. △ Less

Submitted 18 June, 2020; v1 submitted 20 April, 2020; originally announced April 2020.

Modified Stokes drift due to surface waves and corrugated sea-floor interactions with and without a mean current

Authors: Akanksha Gupta, Anirban Guha

Abstract: In this paper, we show that Stokes drift may be significantly affected when an incident intermediate or shallow water surface wave travels over a corrugated sea-floor. The underlying mechanism is Bragg resonance -- reflected waves generated via nonlinear resonant interactions between an incident wave and a rippled bottom. We theoretically explain the fundamental effect of two counter-propagating S… ▽ More In this paper, we show that Stokes drift may be significantly affected when an incident intermediate or shallow water surface wave travels over a corrugated sea-floor. The underlying mechanism is Bragg resonance -- reflected waves generated via nonlinear resonant interactions between an incident wave and a rippled bottom. We theoretically explain the fundamental effect of two counter-propagating Stokes waves on Stokes drift and then perform numerical simulations of Bragg resonance using High-order Spectral method. A monochromatic incident wave on interaction with a patch of bottom ripple yields a complex interference between the incident and reflected waves. When the velocity induced by the reflected waves exceeds that of the incident, particle trajectories reverse, leading to a backward drift. Lagrangian and Lagrangian-mean trajectories reveal that surface particles near the up-wave side of the patch are either trapped or reflected, implying that the rippled patch acts as a non-surface-invasive particle trap or reflector. On increasing the length and amplitude of the rippled patch; reflection, and thus the effectiveness of the patch, increases. The inclusion of realistic constant current shows noticeable differences between Lagrangian-mean trajectories with and without the rippled patch. Theoretical analysis reveals additional terms in the Stokes drift arising from the particular solution due to mean-current--bottom-ripple interactions, irrespective of whether Bragg resonance condition is met. Our analyses may be useful for designing artificial, corrugated sea-floor patches for mitigating microplastics and other forms of ocean pollution. We also expect that sea-floor corrugations, especially in the nearshore region, may significantly affect oceanic tracer transport. △ Less

Submitted 25 July, 2020; v1 submitted 9 February, 2020; originally announced February 2020.

Journal ref: Phys. Rev. Fluids 6, 024801 (2021)

SMT-based Robot Transition Repair

Authors: Jarrett Holtz, Arjun Guha, Joydeep Biswas

Abstract: State machines are a common model for robot behaviors. Transition functions often rely on parameterized conditions to model preconditions for the controllers, where the correct values of the parameters depend on factors relating to the environment or the specific robot. In the absence of specific calibration procedures a roboticist must painstakingly adjust the parameters through a series of trial… ▽ More State machines are a common model for robot behaviors. Transition functions often rely on parameterized conditions to model preconditions for the controllers, where the correct values of the parameters depend on factors relating to the environment or the specific robot. In the absence of specific calibration procedures a roboticist must painstakingly adjust the parameters through a series of trial and error experiments. In this process, identifying when the robot has taken an incorrect action, and what should be done is straightforward, but finding the right parameter values can be difficult. We present an alternative approach that we call, interactive SMT-based Robot Transition Repair. During execution we record an execution trace of the transition function, and we ask the roboticist to identify a few instances where the robot has transitioned incorrectly, and what the correct transition should have been. A user supplies these corrections based on the type of error to repair, and an automated analysis of the traces partially evaluates the transition function for each correction. This system of constraints is then formulated as a MaxSMT problem, where the solution is a minimal adjustment to the parameters that satisfies the maximum number of constraints. In order to identify a repair that accurately captures user intentions and generalizes to novel scenarios, solutions are explored by iteratively adding constraints to the MaxSMT problem to yield sets of alternative repairs. We test with state machines from multiple domains including robot soccer and autonomous driving, and we evaluate solver based repair with respect to solver choice and optimization hyperparameters. Our results demonstrate that SRTR can repair a variety of states machines and error types 1) quickly, 2) with small numbers of corrections, while 3) not overcorrecting state machines and harming generalized performance. △ Less

Submitted 9 January, 2020; originally announced January 2020.

Comments: In submission to AIJ. arXiv admin note: text overlap with arXiv:1802.01706

A Language-based Serverless Function Accelerator

Authors: Emily Herbert, Arjun Guha

Abstract: Serverless computing is an approach to cloud computing that allows programmers to run serverless functions in response to external events. Serverless functions are priced at sub-second granularity, support transparent elasticity, and relieve programmers from managing the operating system. Thus serverless functions allow programmers to focus on writing application code, and the cloud provider to ma… ▽ More Serverless computing is an approach to cloud computing that allows programmers to run serverless functions in response to external events. Serverless functions are priced at sub-second granularity, support transparent elasticity, and relieve programmers from managing the operating system. Thus serverless functions allow programmers to focus on writing application code, and the cloud provider to manage computing resources globally. Unfortunately, today's serverless platforms exhibit high latency, because it is difficult to maximize resource utilization while minimizing operating costs. This paper presents serverless function acceleration, which is an approach that transparently lowers the latency and resource utilization of a large class of serverless functions. We accomplish this using language-based sandboxing, whereas existing serverless platforms employ more expensive operating system sandboxing technologies, such as containers and virtual machines. OS-based sandboxing is compatible with more programs than language-based techniques. However, instead of ruling out any programs, we use language-based sandboxing when possible, and OS-based sandboxing if necessary. Moreover, we seamlessly transition between language-based and OS-based sandboxing by leveraging the fact that serverless functions must tolerate re-execution for fault tolerance. Therefore, when a serverless function attempts to perform an unsupported operation in the language-based sandbox, we can safely re-execute it in a container. We use a new approach to trace compilation to build source-level, interprocedural, execution trace trees for serverless functions written in JavaScript. We compile trace trees to a safe subset of Rust, validate the compiler output, and link it to a runtime system. We evaluate these techniques in our implementation, which we call Containerless. △ Less

Submitted 3 August, 2020; v1 submitted 5 November, 2019; originally announced November 2019.

Model-Based Warp Overlapped Tiling for Image Processing Programs on GPUs

Authors: Abhinav Jangda, Arjun Guha

Abstract: Domain-specific languages that execute image processing pipelineson GPUs, such as Halide and Forma, operate by 1) dividing the image into overlapped tiles, and 2) fusing loops to improve memory locality. However, current approaches have limitations: 1) they require intra thread block synchronization, which has a non-trivial cost, 2) they must choose between small tiles that require more overlapped… ▽ More Domain-specific languages that execute image processing pipelineson GPUs, such as Halide and Forma, operate by 1) dividing the image into overlapped tiles, and 2) fusing loops to improve memory locality. However, current approaches have limitations: 1) they require intra thread block synchronization, which has a non-trivial cost, 2) they must choose between small tiles that require more overlapped computations or large tiles that increase shared memory access (and lowers occupancy), and 3) their autoscheduling algorithms use simplified GPU models that can result in inefficient global memory accesses. We present a new approach for executing image processing pipelines on GPUs that addresses these limitations as follows. 1) We fuse loops to form overlapped tiles that fit in a single warp, which allows us to use lightweight warp synchronization. 2) We introduce hybrid tiling, which stores overlapped regions in a combination of thread-local registers and shared memory. Thus hybrid tiling either increases occupancy by decreasing shared memory usage or decreases overlap** computations using larger tiles. 3) We present an automatic loop fusion algorithm that considers several factors that affect the performance of GPU kernels. We implement these techniques in PolyMage-GPU, which is a new GPU backend for PolyMage. Our approach produces code that is faster than Halide's manual schedules: 1.65x faster on an NVIDIA GTX 1080Ti and 1.33 faster on an NVIDIA Tesla V100. △ Less

Submitted 8 September, 2020; v1 submitted 16 September, 2019; originally announced September 2019.

Making High-Performance Robots Safe and Easy to Use for an Introduction to Computing

Authors: Joseph Spitzer, Joydeep Biswas, Arjun Guha

Abstract: Robots are a popular platform for introducing computing and artificial intelligence to novice programmers. However, programming state-of-the-art robots is very challenging, and requires knowledge of concurrency, operation safety, and software engineering skills, which can take years to teach. In this paper, we present an approach to introducing computing that allows students to safely and easily p… ▽ More Robots are a popular platform for introducing computing and artificial intelligence to novice programmers. However, programming state-of-the-art robots is very challenging, and requires knowledge of concurrency, operation safety, and software engineering skills, which can take years to teach. In this paper, we present an approach to introducing computing that allows students to safely and easily program high-performance robots. We develop a platform for students to program RoboCup Small Size League robots using JavaScript. The platform 1) ensures physical safety at several levels of abstraction, 2) allows students to program robots using the JavaScript in the browser, without the need to install software, and 3) presents a simplified JavaScript semantics that shields students from confusing language features. We discuss our experience running a week-long workshop using this platform, and analyze over 3,000 student-written program revisions to provide empirical evidence that our approach does help students. △ Less

Submitted 21 November, 2019; v1 submitted 6 September, 2019; originally announced September 2019.

Comments: 8 pages, 7 figures, 4 tables

Automatic Failure Recovery for End-User Programs on Service Mobile Robots

Authors: Jenna Claire Hammond, Joydeep Biswas, Arjun Guha

Abstract: For service mobile robots to be most effective, it must be possible for non-experts and even end-users to program them to do new tasks. Regardless of the programming method (e.g., by demonstration or traditional programming), robot task programs are challenging to write, because they rely on multiple actions to succeed, including human-robot interactions. Unfortunately, interactions are prone to f… ▽ More For service mobile robots to be most effective, it must be possible for non-experts and even end-users to program them to do new tasks. Regardless of the programming method (e.g., by demonstration or traditional programming), robot task programs are challenging to write, because they rely on multiple actions to succeed, including human-robot interactions. Unfortunately, interactions are prone to fail, because a human may perform the wrong action (e.g., if the robot's request is not clear). Moreover, when the robot cannot directly observe the human action, it may not detect the failure until several steps after it occurs. Therefore, writing fault-tolerant robot tasks is beyond the ability of non-experts. This paper presents a principled approach to detect and recover from a broad class of failures that occur in end-user programs on service mobile robots. We present a two-tiered Robot Task Programming Language (RTPL): 1) an expert roboticist uses a specification language to write a probabilistic model of the robot's actions and interactions, and 2) a non-expert then writes an ordinary sequential program for a particular task. The RTPL runtime system executes the task program sequentially, while using the probabilistic model to build a Bayesian network that tracks possible, unobserved failures. If an error is observed, RTPL uses Bayesian inference to find the likely root cause of the error, and then attempts to re-execute a portion of the program for recovery. Our empirical results show that RTPL 1) allows complex tasks to be written concisely, 2) correctly identifies the root cause of failure, and 3) allows multiple tasks to recover from a variety of errors, without task-specific error-recovery code. △ Less

Submitted 6 September, 2019; originally announced September 2019.

Instability of a smooth shear layer through wave interactions

Authors: Jeff Carpenter, Anirban Guha

Abstract: Wave interaction theory can be used as a tool to understand and predict instability in a variety of homogeneous and stratified shear flows. It is however, most often limited to piecewise-linear profiles of the shear layer background velocity, in which stable vorticity wave modes can be easily identified, and their interaction quantified. This approach to understanding shear flow instability is ext… ▽ More Wave interaction theory can be used as a tool to understand and predict instability in a variety of homogeneous and stratified shear flows. It is however, most often limited to piecewise-linear profiles of the shear layer background velocity, in which stable vorticity wave modes can be easily identified, and their interaction quantified. This approach to understanding shear flow instability is extended herein to smooth shear layer profiles. We describe a method, by which the stable vorticity wave modes can be identified, and show that their interaction results in an excellent description of the stability properties of the smooth shear layer, thus demonstrating the presence of the wave interaction mechanism in smooth shear flows. △ Less

Submitted 25 July, 2019; originally announced July 2019.

Predicting vortex merging and ensuing turbulence characteristics in shear layers from initial conditions

Authors: Anirban Guha, Mona Rahmani

Abstract: Unstable shear layers in environmental and industrial flows roll up into a series of vortices, which often form complex nonlinear merging patterns like pairs and triplets. These patterns crucially determine the subsequent turbulence, mixing and scalar transport. We show that the late-time, highly nonlinear merging patterns are predictable from the linearized initial state. The initial asymmetry be… ▽ More Unstable shear layers in environmental and industrial flows roll up into a series of vortices, which often form complex nonlinear merging patterns like pairs and triplets. These patterns crucially determine the subsequent turbulence, mixing and scalar transport. We show that the late-time, highly nonlinear merging patterns are predictable from the linearized initial state. The initial asymmetry between consecutive wavelengths of the vertical velocity field provides an effective measure of the strength and pattern of vortex merging. The predictions of this measure are substantiated using direct numerical simulations. We also show that this measure has significant implications in determining the route to turbulence and the ensuing turbulence characteristics. △ Less

Submitted 28 August, 2019; v1 submitted 16 June, 2019; originally announced June 2019.

Dirichlet Simplex Nest and Geometric Inference

Authors: Mikhail Yurochkin, Aritra Guha, Yuekai Sun, XuanLong Nguyen

Abstract: We propose Dirichlet Simplex Nest, a class of probabilistic models suitable for a variety of data types, and develop fast and provably accurate inference algorithms by accounting for the model's convex geometry and low dimensional simplicial structure. By exploiting the connection to Voronoi tessellation and properties of Dirichlet distribution, the proposed inference algorithm is shown to achieve… ▽ More We propose Dirichlet Simplex Nest, a class of probabilistic models suitable for a variety of data types, and develop fast and provably accurate inference algorithms by accounting for the model's convex geometry and low dimensional simplicial structure. By exploiting the connection to Voronoi tessellation and properties of Dirichlet distribution, the proposed inference algorithm is shown to achieve consistency and strong error bound guarantees on a range of model settings and data distributions. The effectiveness of our model and the learning algorithm is demonstrated by simulations and by analyses of text and financial data. △ Less

Submitted 27 May, 2019; originally announced May 2019.

Comments: ICML 2019

The motion of buoyant point vortices

Authors: Jeff Carpenter, Anirban Guha

Abstract: A general formulation is presented for studying the motion of buoyant vortices in a homogeneous ambient fluid. It extends the well-known Hamiltonian framework for interacting homogeneous point vortices to include buoyancy effects acting on the vortices. This is then used to systematically examine the buoyant 1-, 2-, and 3-vortex problems. In doing so we find that 2 buoyant vortices may either evol… ▽ More A general formulation is presented for studying the motion of buoyant vortices in a homogeneous ambient fluid. It extends the well-known Hamiltonian framework for interacting homogeneous point vortices to include buoyancy effects acting on the vortices. This is then used to systematically examine the buoyant 1-, 2-, and 3-vortex problems. In doing so we find that 2 buoyant vortices may either evolve as a pair in bounded circular orbits, or as two independent unbounded vortices that drift apart, and a criteria is found to distinguish these cases. Special attention is given to the buoyant vortex couple, consisting of two vortices of equal and opposite circulation, and equal buoyancy anomaly. We show that a theoretical maximum height is generally possible for the rise (or fall) of such couples against buoyancy forces. Finally, the possibility and onset of chaotic motions in the buoyant 3-vortex problem is addressed. In contrast to the homogeneous 3-vortex problem, the buoyant vortex system shows evidence that chaos is present. We also demonstrate the chaotic advection of tracer parcels arising from the flow field induced by just 2 buoyant vortices. △ Less

Submitted 16 April, 2020; v1 submitted 6 May, 2019; originally announced May 2019.

Journal ref: Phys. Rev. Fluids 5, 064702 (2020)

On the opposing roles of the Boussinesq and non-Boussinesq baroclinic torques in surface gravity wave propagation

Authors: Eyal Heifetz, Ron Maor, Anirban Guha

Abstract: Here we suggest an alternative understanding of the surface gravity wave propagation mechanism based on the baroclinic torque, which operates to translate the interfacial vorticity anomalies at the air-water interface. We demonstrate how the non-Boussinesq term of the baroclinic torque acts against the Boussinesq one to hinder wave propagation. By standard vorticity inversion and mirror imaging, w… ▽ More Here we suggest an alternative understanding of the surface gravity wave propagation mechanism based on the baroclinic torque, which operates to translate the interfacial vorticity anomalies at the air-water interface. We demonstrate how the non-Boussinesq term of the baroclinic torque acts against the Boussinesq one to hinder wave propagation. By standard vorticity inversion and mirror imaging, we then show how the existence of the bottom boundary affects the two types of torque. Since the opposing non-Boussinesq torque results solely from the mirror image, it vanishes in the deep water limit and its magnitude is half of the Boussinesq torque in the shallow water limit. This reveals that Boussinesq approximation is valid in the deep water limit, even though the density contrast between air and water is large. The mechanistic roles, played by the Boussinesq and non-Boussinesq parts of the baroclinic torque, remain obscured in the standard derivation where the time-dependent Bernoulli equation is implemented instead of the interfacial vorticity equation. Finally, we note on passing that the Virial theorem for surface gravity waves can be obtained solely from considerations of the dynamics at the air-water interface. △ Less

Submitted 4 December, 2019; v1 submitted 13 March, 2019; originally announced March 2019.

Similarities between the structure functions of thermal convection and hydrodynamic turbulence

Authors: Shashwat Bhattacharya, Shubhadeep Sadhukhan, Anirban Guha, Mahendra K. Verma

Abstract: In this paper, we analyze the scaling of velocity structure functions of turbulent thermal convection. Using high-resolution numerical simulations, we show that the structure functions scale similar to those of hydrodynamic turbulence, with the scaling exponents in agreement with She and Leveque's predictions [Phys. Rev. Lett. 72, 336-339 (1994)]. The probability distribution functions of velocity… ▽ More In this paper, we analyze the scaling of velocity structure functions of turbulent thermal convection. Using high-resolution numerical simulations, we show that the structure functions scale similar to those of hydrodynamic turbulence, with the scaling exponents in agreement with She and Leveque's predictions [Phys. Rev. Lett. 72, 336-339 (1994)]. The probability distribution functions of velocity increments are non-Gaussian with wide tails in the dissipative scales and become close to Gaussian in the inertial range. The tails of the probability distribution follow a stretched exponential. We also show that in thermal convection, the energy flux in the inertial range is less than the viscous dissipation rate. This is unlike in hydrodynamic turbulence where the energy flux and the dissipation rate are equal. △ Less

Submitted 14 November, 2019; v1 submitted 3 March, 2019; originally announced March 2019.

Comments: 21 pages, 9 figures

Journal ref: Phys. Fluids 31, 115107 (2019)

Formal Foundations of Serverless Computing

Authors: Abhinav Jangda, Donald Pinckney, Yuriy Brun, Arjun Guha

Abstract: Serverless computing (also known as functions as a service) is a new cloud computing abstraction that makes it easier to write robust, large-scale web services. In serverless computing, programmers write what are called serverless functions, and the cloud platform transparently manages the operating system, resource allocation, load-balancing, and fault tolerance. When demand for the service spike… ▽ More Serverless computing (also known as functions as a service) is a new cloud computing abstraction that makes it easier to write robust, large-scale web services. In serverless computing, programmers write what are called serverless functions, and the cloud platform transparently manages the operating system, resource allocation, load-balancing, and fault tolerance. When demand for the service spikes, the platform automatically allocates additional hardware to the service and manages load-balancing; when demand falls, the platform silently deallocates idle resources; and when the platform detects a failure, it transparently retries affected requests. In 2014, Amazon Web Services introduced the first serverless platform, AWS Lambda, and similar abstractions are now available on all major cloud computing platforms. Unfortunately, the serverless computing abstraction exposes several low-level operational details that make it hard for programmers to write and reason about their code. This paper sheds light on this problem by presenting $λ_Λ$, an operational semantics of the essence of serverless computing. Despite being a small (half a page) core calculus, $λ_Λ$ models all the low-level details that serverless functions can observe. To show that $λ_Λ$ is useful, we present three applications. First, to ease reasoning about code, we present a simplified naive semantics of serverless execution and precisely characterize when the naive semantics and $λ_Λ$ coincide. Second, we augment $λ_Λ$ with a key-value store to allow reasoning about stateful serverless functions. Third, since a handful of serverless platforms support serverless function composition, we show how to extend $λ_Λ$ with a composition language. We have implemented this composition language and show that it outperforms prior work. △ Less

Submitted 4 October, 2020; v1 submitted 15 February, 2019; originally announced February 2019.

Journal ref: PACMPL, OOPSLA issue, vol. 3, October 2019, pp. 149:1-149:26

Not So Fast: Analyzing the Performance of WebAssembly vs. Native Code

Authors: Abhinav Jangda, Bobby Powers, Emery Berger, Arjun Guha

Abstract: All major web browsers now support WebAssembly, a low-level bytecode intended to serve as a compilation target for code written in languages like C and C++. A key goal of WebAssembly is performance parity with native code; previous work reports near parity, with many applications compiled to WebAssembly running on average 10% slower than native code. However, this evaluation was limited to a suite… ▽ More All major web browsers now support WebAssembly, a low-level bytecode intended to serve as a compilation target for code written in languages like C and C++. A key goal of WebAssembly is performance parity with native code; previous work reports near parity, with many applications compiled to WebAssembly running on average 10% slower than native code. However, this evaluation was limited to a suite of scientific kernels, each consisting of roughly 100 lines of code. Running more substantial applications was not possible because compiling code to WebAssembly is only part of the puzzle: standard Unix APIs are not available in the web browser environment. To address this challenge, we build Browsix-Wasm, a significant extension to Browsix that, for the first time, makes it possible to run unmodified WebAssembly-compiled Unix applications directly inside the browser. We then use Browsix-Wasm to conduct the first large-scale evaluation of the performance of WebAssembly vs. native. Across the SPEC CPU suite of benchmarks, we find a substantial performance gap: applications compiled to WebAssembly run slower by an average of 45% (Firefox) to 55% (Chrome), with peak slowdowns of 2.08x (Firefox) and 2.5x (Chrome). We identify the causes of this performance degradation, some of which are due to missing optimizations and code generation issues, while others are inherent to the WebAssembly platform. △ Less

Submitted 31 May, 2019; v1 submitted 25 January, 2019; originally announced January 2019.

Comments: Accepted (to appear) at USENIX Annual Technical Conference 2019

On posterior contraction of parameters and interpretability in Bayesian mixture modeling

Authors: Aritra Guha, Nhat Ho, XuanLong Nguyen

Abstract: We study posterior contraction behaviors for parameters of interest in the context of Bayesian mixture modeling, where the number of mixing components is unknown while the model itself may or may not be correctly specified. Two representative types of prior specification will be considered: one requires explicitly a prior distribution on the number of mixture components, while the other places a n… ▽ More We study posterior contraction behaviors for parameters of interest in the context of Bayesian mixture modeling, where the number of mixing components is unknown while the model itself may or may not be correctly specified. Two representative types of prior specification will be considered: one requires explicitly a prior distribution on the number of mixture components, while the other places a nonparametric prior on the space of mixing distributions. The former is shown to yield an optimal rate of posterior contraction on the model parameters under minimal conditions, while the latter can be utilized to consistently recover the unknown number of mixture components, with the help of a fast probabilistic post-processing procedure. We then turn the study of these Bayesian procedures to the realistic settings of model misspecification. It will be shown that the modeling choice of kernel density functions plays perhaps the most impactful roles in determining the posterior contraction rates in the misspecified situations. Drawing on concrete posterior contraction rates established in this paper we wish to highlight some aspects about the interesting tradeoffs between model expressiveness and interpretability that a statistical modeler must negotiate in the rich world of mixture modeling. △ Less

Submitted 15 January, 2019; originally announced January 2019.

Comments: 53 pages; 8 figures

On the normal form of synchronization and resonance between vorticity waves in shear flow instability

Authors: Eyal Heifetz, Anirban Guha

Abstract: A central mechanism of linearised two dimensional shear instability can be described in terms of a nonlinear, action-at-a-distance, phase-locking resonance between two vorticity waves which propagate counter to their local mean flow as well as counter to each other. Here we analyze the prototype of this interaction as an autonomous, nonlinear dynamical system. The wave interaction equations can be… ▽ More A central mechanism of linearised two dimensional shear instability can be described in terms of a nonlinear, action-at-a-distance, phase-locking resonance between two vorticity waves which propagate counter to their local mean flow as well as counter to each other. Here we analyze the prototype of this interaction as an autonomous, nonlinear dynamical system. The wave interaction equations can be written in a generalized Hamiltonian action-angle form. The pseudo-energy serves as the Hamiltonian of the system, the action coordinates are the contribution of the vorticity waves to the wave-action, and the angles are the phases of the vorticity waves. The term "generalized action-angle" emphasizes that the action of each wave is generally time dependent, which allows instability. The synchronization mechanism between the wave phases depends on the cosine of their relative phase, rather than the sine as in the Kuramoto model. The unstable normal modes of the linearised dynamics correspond to the stable fixed points of the dynamical system and vice versa. Furthermore, the normal form of the wave interaction dynamics reveals a new type of inhomogeneous bifurcation -- annihilation of a pair of stable and unstable fixed points yields the emergence of two neutral center fixed points of opposite circulation. △ Less

Submitted 25 June, 2019; v1 submitted 15 January, 2019; originally announced January 2019.

Journal ref: Phys. Rev. E 100, 043105 (2019)

Revisiting Bolgiano-Obukhov scaling for moderately stably stratified turbulence

Authors: Shadab Alam, Anirban Guha, Mahendra K. Verma

Abstract: According to the celebrated Bolgiano--Obukhov \citep{Bolgiano_1959,Obukhov_1959} phenomenology for moderately stably stratified turbulence, the energy spectrum in the inertial range shows a dual scaling; the kinetic energy follows (i) $\sim k^{-11/5}$ for $k < k_B$, and (ii) $\sim k^{-5/3}$ for $k > k_B$, where $k_B$ is Bolgiano wavenumber. The $k^{-5/3}$ scaling akin to passive scalar turbulence… ▽ More According to the celebrated Bolgiano--Obukhov \citep{Bolgiano_1959,Obukhov_1959} phenomenology for moderately stably stratified turbulence, the energy spectrum in the inertial range shows a dual scaling; the kinetic energy follows (i) $\sim k^{-11/5}$ for $k < k_B$, and (ii) $\sim k^{-5/3}$ for $k > k_B$, where $k_B$ is Bolgiano wavenumber. The $k^{-5/3}$ scaling akin to passive scalar turbulence is a direct consequence of the assumption that buoyancy is insignificant for $k>k_B$. We revisit this assumption, and using constancy of kinetic and potential energy fluxes and simple theoretical analysis, we find that $ k^{-5/3}$ spectrum is absent. This is because the velocity field at small scales is too weak to establish a constant kinetic energy flux as in passive scalar turbulence. A quantitative condition for the existence of the second regime is also derived in the paper. △ Less

Submitted 25 June, 2019; v1 submitted 30 November, 2018; originally announced November 2018.

An inverse technique for reconstructing ocean's density stratification from surface data

Authors: Subhajit Kar, Anirban Guha

Abstract: In this article, we propose an inverse technique that accurately reconstructs the ocean's density stratification profile simply from free surface elevation data. Satellite observations suggest that ocean surface contains the signature of internal tides, which are internal gravity waves generated by the barotropic tides. Since internal tides contain the information of ocean's density stratification… ▽ More In this article, we propose an inverse technique that accurately reconstructs the ocean's density stratification profile simply from free surface elevation data. Satellite observations suggest that ocean surface contains the signature of internal tides, which are internal gravity waves generated by the barotropic tides. Since internal tides contain the information of ocean's density stratification, the latter can in principle be reconstructed from the free surface signature. We consider a simple theoretical model that approximates a continuously stratified ocean as discrete layers of constant buoyancy frequency; this facilitates the derivation of a closed-form dispersion relation. First, we numerically simulate internal tide generation for toy ocean scenarios and subsequently perform Space-Time Fourier Transform (STFT) of the free surface, which yields internal tide spectra with wavenumbers corresponding to the tidal frequency. The density profile is reconstructed by substituting these wavenumbers into the dispersion relation. Finally, we consider a more realistic situation with rotation, bottom topography, shear and density profiles representative of the Strait of Gibraltar. Density reconstruction in the presence and absence of shear are respectively found to be $90.2\%$ and $94.2\%$ accurate. △ Less

Submitted 25 June, 2019; v1 submitted 27 October, 2018; originally announced October 2018.

Model-independent Astrophysical Constraints on Leptophilic Dark Matter in the Framework of Tsallis Statistics

Authors: Atanu Guha, P. S. Bhupal Dev, Prasanta Kumar Das

Abstract: We derive model-independent astrophysical constraints on leptophilic dark matter (DM), considering its thermal production in a supernova core and taking into account core temperature fluctuations within the framework of $q$-deformed Tsallis statistics. In an effective field theory approach, where the DM fermions interact with the Standard Model via dimension-six operators of either scalar-pseudosc… ▽ More We derive model-independent astrophysical constraints on leptophilic dark matter (DM), considering its thermal production in a supernova core and taking into account core temperature fluctuations within the framework of $q$-deformed Tsallis statistics. In an effective field theory approach, where the DM fermions interact with the Standard Model via dimension-six operators of either scalar-pseudoscalar, vector-axial vector, or tensor-axial tensor type, we obtain bounds on the effective cut-off scale $Λ$ from supernova cooling and free-streaming of DM from supernova core, and from thermal relic density considerations, depending on the DM mass and the $q$-deformation parameter. Using Raffelt's criterion on the energy loss rate from SN1987A, we obtain a lower bound on $\Lda \gtrsim 3$ (12) TeV corresponding to $q = 1.0~(1.1)$ and an average supernova core temperature of $T_{\rm SN}=30$ MeV. From the optical depth criterion on the free-streaming of DM fermions from the outer 10\% of the SN1987A core, the cooling bound is restricted to $\Lda \gtrsim 1$ TeV. Both cooling and free-streaming bounds are insensitive to the DM mass $m_χ$ for $m_χ\lesssim T_{\rm SN}$, whereas for $m_χ\gg T_{\rm SN}$, the bounds weaken significantly due to the Boltzmann-suppression of the DM number density. We also calculate the thermal relic density of the DM particles in this setup and find that it imposes an upper bound on $Λ^4/m_χ^2$, which together with the cooling/free-streaming bound significantly constrains light leptophilic DM. △ Less

Submitted 16 February, 2019; v1 submitted 30 September, 2018; originally announced October 2018.

Comments: 27 pages, 9 figures and 3 tables

Journal ref: JCAP 02(2019)032

Scalable inference of topic evolution via models for latent geometric structures

Authors: Mikhail Yurochkin, Zhiwei Fan, Aritra Guha, Paraschos Koutris, XuanLong Nguyen

Abstract: We develop new models and algorithms for learning the temporal dynamics of the topic polytopes and related geometric objects that arise in topic model based inference. Our model is nonparametric Bayesian and the corresponding inference algorithm is able to discover new topics as the time progresses. By exploiting the connection between the modeling of topic polytope evolution, Beta-Bernoulli proce… ▽ More We develop new models and algorithms for learning the temporal dynamics of the topic polytopes and related geometric objects that arise in topic model based inference. Our model is nonparametric Bayesian and the corresponding inference algorithm is able to discover new topics as the time progresses. By exploiting the connection between the modeling of topic polytope evolution, Beta-Bernoulli process and the Hungarian matching algorithm, our method is shown to be several orders of magnitude faster than existing topic modeling approaches, as demonstrated by experiments working with several million documents in under two dozens of minutes. △ Less

Submitted 1 November, 2019; v1 submitted 23 September, 2018; originally announced September 2018.

Comments: NeurIPS 2019

Explosive instability due to flow over a rippled bottom

Authors: Raunak Raj, Anirban Guha

Abstract: In this paper, we study Bragg resonance, i.e. the triad interaction between surface and/or interfacial waves with bottom ripple, in presence of background velocity. We show that when one of the constituent waves of the triad has negative energy, the amplitudes of all the waves grow exponentially. This is very different from classic Bragg resonance in which one wave decays to cause growth of the ot… ▽ More In this paper, we study Bragg resonance, i.e. the triad interaction between surface and/or interfacial waves with bottom ripple, in presence of background velocity. We show that when one of the constituent waves of the triad has negative energy, the amplitudes of all the waves grow exponentially. This is very different from classic Bragg resonance in which one wave decays to cause growth of the other. The instabilities we observe are `explosive', and are different from normal mode shear instabilities since our velocity profiles are linearly stable. Our work may explain the existence of large amplitude internal waves over periodic bottom ripples in presence of tidal flow observed in oceans and estuaries. △ Less

Submitted 20 September, 2018; originally announced September 2018.

On Bragg resonances and wave triad interactions in two-layered shear flows

Authors: Raunak Raj, Anirban Guha

Abstract: The standard resonance conditions for Bragg scattering as well as weakly nonlinear wave triads have been traditionally derived in the absence of any background velocity. In this paper, we have studied how these resonance conditions get modified when uniform, as well as various piecewise linear velocity profiles, are considered for two-layered shear flows. Background velocity can influence the reso… ▽ More The standard resonance conditions for Bragg scattering as well as weakly nonlinear wave triads have been traditionally derived in the absence of any background velocity. In this paper, we have studied how these resonance conditions get modified when uniform, as well as various piecewise linear velocity profiles, are considered for two-layered shear flows. Background velocity can influence the resonance conditions in two ways (i) by causing Doppler shifts, and (ii) by changing the intrinsic frequencies of the waves. For Bragg resonance, even a uniform velocity field changes the resonance condition. Velocity shear strongly influences the resonance conditions since, in addition to changing the intrinsic frequencies, it can cause unequal Doppler shifts between the surface, pycnocline, and the bottom. Using multiple scale analysis and Fredholm alternative, we analytically obtain the equations governing both the Bragg resonance and the wave triads. We have also extended the Higher Order Spectral method, a highly efficient computational tool usually used to study triad and Bragg resonance problems, to incorporate the effect of piecewise linear velocity profile. A significant aspect, both in theoretical and numerical fronts, has been extending the potential flow approximation, which is the basis of studying these kinds of problems, to incorporate piecewise constant background shear. △ Less

Submitted 15 February, 2019; v1 submitted 19 August, 2018; originally announced August 2018.

Energy transfer in resonant and near-resonant internal wave triads for weakly non-uniform stratifications. Part I: Unbounded domain

Authors: G. Saranraj, Anirban Guha

Abstract: In this paper, using multiple scale analysis we derive a generalized mathematical model for amplitude evolution, and for calculating the energy exchange in resonant and near-resonant global triads consisting of weakly nonlinear internal gravity wave packets in weakly non-uniform density stratifications in an unbounded domain in the presence of viscous and rotational effects. Such triad interaction… ▽ More In this paper, using multiple scale analysis we derive a generalized mathematical model for amplitude evolution, and for calculating the energy exchange in resonant and near-resonant global triads consisting of weakly nonlinear internal gravity wave packets in weakly non-uniform density stratifications in an unbounded domain in the presence of viscous and rotational effects. Such triad interactions are one of the mechanisms by which high wavenumber internal waves lead to ocean turbulence and mixing via parametric subharmonic instability. Non-uniform stratification introduces detuning - mismatch in the vertical wavenumber triad condition, which may strongly affect the energy transfer process. We investigate in detail how factors like wave-packets' width, group speeds, nonlinear coupling coefficients, detuning, and viscosity affect energy transfer in weakly varying stratification. We find limitations of the well-known 'pump-wave approximation' and derive a non-dimensional number, which can be evaluated from initial conditions, that can predict the maximum energy transferred from the primary wave during the later stages. Two additional non-dimensional numbers, based on various factors affecting energy transfer between near-resonant wave-packets have also been defined. Moreover, we identify the optimal background stratification in a medium of varying stratification for the primary wave to form a triad with no detuning so that the energy transfer is maximum. Finally, we show that even a small change in the background stratification can cause a significant difference in the energy transfer process between the wave-packets when they have the same order of magnitude of energy. △ Less

Submitted 26 May, 2020; v1 submitted 16 August, 2018; originally announced August 2018.

Spectral bounds for non-uniform hypergraphs using weighted clique expansion

Authors: Ashwin Guha, Ambedkar Dukkipati

Abstract: Hypergraphs are an invaluable tool to understand many hidden patterns in large data sets. Among many ways to represent hypergraph, one useful representation is that of weighted clique expansion. In this paper, we consider this representation for non-uniform hypergraphs. We generalize the spectral results for uniform hypergraphs to non-uniform hypergraphs and show that they extend in a natural way.… ▽ More Hypergraphs are an invaluable tool to understand many hidden patterns in large data sets. Among many ways to represent hypergraph, one useful representation is that of weighted clique expansion. In this paper, we consider this representation for non-uniform hypergraphs. We generalize the spectral results for uniform hypergraphs to non-uniform hypergraphs and show that they extend in a natural way. We provide a bound on the largest eigenvalue with respect to the average degree of neighbours of a vertex in a graph. We also prove an inequality on the boundary of a vertex set in terms of the largest and second smallest eigenvalue and use it to obtain bounds on some connectivity parameters of the hypergraph. △ Less

Submitted 14 August, 2018; originally announced August 2018.

Ocean bathymetry reconstruction from surface data using hydraulics theory

Authors: Subhajit Kar, Anirban Guha

Abstract: Here we propose a technique that successfully reconstructs ocean bathymetry from the free surface velocity and elevation data. This technique is based on the principles of open-channel hydraulics, according to which a sub-critical flow over a seamount creates a free surface dip. The proposed method recognizes that such free surface dip contains the signature of the bottom topography, hence inverts… ▽ More Here we propose a technique that successfully reconstructs ocean bathymetry from the free surface velocity and elevation data. This technique is based on the principles of open-channel hydraulics, according to which a sub-critical flow over a seamount creates a free surface dip. The proposed method recognizes that such free surface dip contains the signature of the bottom topography, hence inverts the free surface to reconstruct the topography accurately. We applied our inversion technique on re-analysis data, and reconstructed the Mediterranean and the Red sea bathymetries of $1/12\degree$ resolution with approximately $90$\% accuracy. △ Less

Submitted 15 November, 2018; v1 submitted 1 May, 2018; originally announced May 2018.

Comments: Accepted for publication in Physics of Fluids (Letters)

Constraints on light Dark Matter fermions from relic density consideration and Tsallis statistics

Authors: Atanu Guha, Prasanta Kumar Das

Abstract: The cold dark matter fermions with mass MeV scale, pair produced inside the supernova SN1987A core, can freely stream away from the supernovae and hence contributes to its energy loss rate. Similar type of DM fermions(having similar kind of coupling to the standard model photon), produced from some other sources earlier, could have contributed to the relic density of the Universe. Working in a the… ▽ More The cold dark matter fermions with mass MeV scale, pair produced inside the supernova SN1987A core, can freely stream away from the supernovae and hence contributes to its energy loss rate. Similar type of DM fermions(having similar kind of coupling to the standard model photon), produced from some other sources earlier, could have contributed to the relic density of the Universe. Working in a theory with an effective dark matter-photon coupling (inversely proportional to the scale $Λ$) in the formalism of Tsallis statistics, we find the dark matter contribution to the relic density and obtain a upper bound on $Λ$ using the experimental bound on the relic density for cold non-baryonic matter i.e. $Ωh^2 = 0.1186 \pm 0.0020 $. The upper bound obtained from the relic density is shown with the lower bound obtained from the Raffelt's criterion on the emissibity rate of the supernovae SN1987A energy loss $\dot{\varepsilon}(e^+ e^- \to χ\overlineχ) \le 10^{19}~\rm{erg~g^{-1}s^{-1}}$ and the optical depth criteria on the free streaming of the dark matter fermion (produced inside the supernovae core). As the deformation parameter $q$ changes from $1.0$ (undeformed scenario) to $1.1$(deformed scenario), the relic density bound on $Λ$ is found to vary from $ \sim 4.9 \times 10^7 $ TeV to $1.6 \times 10^8$ TeV for a fermion dark matter($χ$) of mass $m_χ= 30~\rm{MeV}$, which is almost $10$ times more than the lower bound obtained from the SN1987A energy loss rate and the optical depth criteria. \noindent {{\bf Keywords}: Dark matter, Relic density, Supernova cooling, Tsallis statistics, free-streaming, } } △ Less

Submitted 27 June, 2018; v1 submitted 12 March, 2018; originally announced March 2018.

Comments: 18 Pages, 10 figures

Report number: JHEP_111P_0318

Journal ref: J. High Energ. Phys. (2018) 2018: 139. Publisher Name : Springer Berlin Heidelberg; Online ISSN : 1029-8479

Putting in All the Stops: Execution Control for JavaScript

Authors: Samuel Baxter, Rachit Nigam, Joe Gibbs Politz, Shriram Krishnamurthi, Arjun Guha

Abstract: Scores of compilers produce JavaScript, enabling programmers to use many languages on the Web, reuse existing code, and even use Web IDEs. Unfortunately, most compilers inherit the browser's compromised execution model, so long-running programs freeze the browser tab, infinite loops crash IDEs, and so on. The few compilers that avoid these problems suffer poor performance and are difficult to engi… ▽ More Scores of compilers produce JavaScript, enabling programmers to use many languages on the Web, reuse existing code, and even use Web IDEs. Unfortunately, most compilers inherit the browser's compromised execution model, so long-running programs freeze the browser tab, infinite loops crash IDEs, and so on. The few compilers that avoid these problems suffer poor performance and are difficult to engineer. This paper presents Stopify, a source-to-source compiler that extends JavaScript with debugging abstractions and blocking operations, and easily integrates with existing compilers. We apply Stopify to 10 programming languages and develop a Web IDE that supports stop**, single-step**, breakpointing, and long-running computations. For nine languages, Stopify requires no or trivial compiler changes. For eight, our IDE is the first that provides these features. Two of our subject languages have compilers with similar features. Stopify's performance is competitive with these compilers and it makes them dramatically simpler. Stopify's abstractions rely on first-class continuations, which it provides by compiling JavaScript to JavaScript. We also identify sub-languages of JavaScript that compilers implicitly use, and exploit these to improve performance. Finally, Stopify needs to repeatedly interrupt and resume program execution. We use a sampling-based technique to estimate program speed that outperforms other systems. △ Less

Submitted 15 April, 2018; v1 submitted 8 February, 2018; originally announced February 2018.

Comments: In proceedings of ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI) 2018

Interactive Robot Transition Repair With SMT

Authors: Jarrett Holtz, Arjun Guha, Joydeep Biswas

Abstract: Complex robot behaviors are often structured as state machines, where states encapsulate actions and a transition function switches between states. Since transitions depend on physical parameters, when the environment changes, a roboticist has to painstakingly readjust the parameters to work in the new environment. We present interactive SMT-based Robot Transition Repair (SRTR): instead of manuall… ▽ More Complex robot behaviors are often structured as state machines, where states encapsulate actions and a transition function switches between states. Since transitions depend on physical parameters, when the environment changes, a roboticist has to painstakingly readjust the parameters to work in the new environment. We present interactive SMT-based Robot Transition Repair (SRTR): instead of manually adjusting parameters, we ask the roboticist to identify a few instances where the robot is in a wrong state and what the right state should be. A lightweight automated analysis of the transition function's source code then 1) identifies adjustable parameters, 2) converts the transition function into a system of logical constraints, and 3) formulates the constraints and user-supplied corrections as MaxSMT problem that yields new parameter values. Our evaluation shows that SRTR is effective on real robots and in simulation. We show that SRTR finds new parameters 1) quickly, 2) with only a few corrections, and 3) that the parameters generalize to new scenarios. We also show that a simple state machine corrected by SRTR can out-perform a more complex, expert-tuned state machine in the real world. △ Less

Submitted 5 May, 2018; v1 submitted 5 February, 2018; originally announced February 2018.

Comments: International Joint Conference on Artificial Intelligence (IJCAI), 2018

Large density stratification stabilizes Rayleigh-Taylor instability in presence of shear

Authors: Raunak Raj, Anirban Guha

Abstract: This letter investigates the effect of shear on Rayleigh-Taylor instability (RTI). Even simple uniform shear strongly influences the instability; longer waves are completely stabilized when density stratification is large (higher Atwood numbers). This apparently counter-intuitive result is due to the presence of Atwood number in the shear term. When the unstable density interface is embedded in a… ▽ More This letter investigates the effect of shear on Rayleigh-Taylor instability (RTI). Even simple uniform shear strongly influences the instability; longer waves are completely stabilized when density stratification is large (higher Atwood numbers). This apparently counter-intuitive result is due to the presence of Atwood number in the shear term. When the unstable density interface is embedded in a shear layer, shear is again found to stabilize the RTI. However, this configuration introduces additional unstable shear instability modes in the lower wavenumber regime. A new type of shear instability, whose growth rate increases with Atwood number, plays a dominant role, while Kelvin-Helmholtz instability (KHI), which was previously understood to be the only possible shear instability in this context, has little significance. Hence the billows observed in the nonlinear stages of RTI, which are usually attributed to KHI, may actually be the nonlinear manifestation of this new instability. △ Less

Submitted 22 October, 2017; originally announced October 2017.

Conic Scan-and-Cover algorithms for nonparametric topic modeling

Authors: Mikhail Yurochkin, Aritra Guha, XuanLong Nguyen

Abstract: We propose new algorithms for topic modeling when the number of topics is unknown. Our approach relies on an analysis of the concentration of mass and angular geometry of the topic simplex, a convex polytope constructed by taking the convex hull of vertices representing the latent topics. Our algorithms are shown in practice to have accuracy comparable to a Gibbs sampler in terms of topic estimati… ▽ More We propose new algorithms for topic modeling when the number of topics is unknown. Our approach relies on an analysis of the concentration of mass and angular geometry of the topic simplex, a convex polytope constructed by taking the convex hull of vertices representing the latent topics. Our algorithms are shown in practice to have accuracy comparable to a Gibbs sampler in terms of topic estimation, which requires the number of topics be given. Moreover, they are one of the fastest among several state of the art parametric techniques. Statistical consistency of our estimator is established under some conditions. △ Less

Submitted 9 October, 2017; originally announced October 2017.

On the inertial effects of density variation in stratified shear flows

Authors: Anirban Guha, Raunak Raj

Abstract: In this paper, we first revisit the celebrated Boussinesq approximation in stratified flows. Using scaling arguments we show that when the background shear is weak, the Boussinesq approximation yields either (i) $A_t\ll \mathcal{O}(1)$ or (ii) $Fr_c^2 \ll \mathcal{O}(1)$, where $A_t$ is the ratio of density variation to the mean density and $Fr_c$ is the ratio of the phase speed to the long wave s… ▽ More In this paper, we first revisit the celebrated Boussinesq approximation in stratified flows. Using scaling arguments we show that when the background shear is weak, the Boussinesq approximation yields either (i) $A_t\ll \mathcal{O}(1)$ or (ii) $Fr_c^2 \ll \mathcal{O}(1)$, where $A_t$ is the ratio of density variation to the mean density and $Fr_c$ is the ratio of the phase speed to the long wave speed. The second clause implies, contrary to the commonly accepted notion, that a flow with large density variations can also be Boussinesq. Indeed, we show that deepwater surface gravity waves are Boussinesq while shallow water surface gravity waves are not. However, in the presence of moderate/strong shear, Boussinesq approximation implies the conventionally accepted $A_t\ll \mathcal{O}(1)$. To understand the inertial effects of density variation, our second objective is to explore various non-Boussinesq shear flows and study different kinds of stably propagating waves that can be present at an interface between two fluids of different background densities and vorticities. Furthermore, three kinds of density interfaces - neutral, stable and unstable - embedded in a background shear layer, are investigated. Instabilities ensuing from these configurations, which includes Kelvin-Helmholtz, Holmboe, Rayleigh-Taylor and triangular-jet, are studied in terms of resonant wave interactions. The effects of density stratification and the shear on the stability of each of these flow configurations are explored. Some of the results, e.g. the destabilizing role of density stratification, stabilizing role of shear, etc. are apparently counter-intuitive, but physical explanations are possible if the instabilities are interpreted from wave interactions perspective. △ Less

Submitted 1 December, 2018; v1 submitted 19 September, 2017; originally announced September 2017.

Comments: Physics of Fluids

Tortoise: Interactive System Configuration Repair

Authors: Aaron Weiss, Arjun Guha, Yuriy Brun

Abstract: System configuration languages provide powerful abstractions that simplify managing large-scale, networked systems. Thousands of organizations now use configuration languages, such as Puppet. However, specifications written in configuration languages can have bugs and the shell remains the simplest way to debug a misconfigured system. Unfortunately, it is unsafe to use the shell to fix problems wh… ▽ More System configuration languages provide powerful abstractions that simplify managing large-scale, networked systems. Thousands of organizations now use configuration languages, such as Puppet. However, specifications written in configuration languages can have bugs and the shell remains the simplest way to debug a misconfigured system. Unfortunately, it is unsafe to use the shell to fix problems when a system configuration language is in use: a fix applied from the shell may cause the system to drift from the state specified by the configuration language. Thus, despite their advantages, configuration languages force system administrators to give up the simplicity and familiarity of the shell. This paper presents a synthesis-based technique that allows administrators to use configuration languages and the shell in harmony. Administrators can fix errors using the shell and the technique automatically repairs the higher-level specification written in the configuration language. The approach (1) produces repairs that are consistent with the fix made using the shell; (2) produces repairs that are maintainable by minimizing edits made to the original specification; (3) ranks and presents multiple repairs when relevant; and (4) supports all shells the administrator may wish to use. We implement our technique for Puppet, a widely used system configuration language, and evaluate it on a suite of benchmarks under 42 repair scenarios. The top-ranked repair is selected by humans 76% of the time and the human-equivalent repair is ranked 1.31 on average. △ Less

Submitted 15 September, 2017; originally announced September 2017.

Comments: Published version in proceedings of IEEE/ACM International Conference on Automated Software Engineering (ASE) 2017

Journal ref: in Proceedings of the 32nd IEEE/ACM International Conference on Automated Software Engineering (ASE), 2017, pp. 625-636

An Extensive Study of Bose-Einstein Condensation in Liquid Helium using Tsallis Statistics

Authors: Atanu Guha, Prasanta Kumar Das

Abstract: Realistic scenario can be represented by general canonical ensemble way better than the ideal one, with proper parameter sets involved. We study the Bose-Einstein condensation phenomena of liquid helium within the framework of Tsallis statistics. With a comparatively high value of the deformation parameter $q(\sim 1.4)$, the theoretically calculated value of the critical temperature($T_c$) of the… ▽ More Realistic scenario can be represented by general canonical ensemble way better than the ideal one, with proper parameter sets involved. We study the Bose-Einstein condensation phenomena of liquid helium within the framework of Tsallis statistics. With a comparatively high value of the deformation parameter $q(\sim 1.4)$, the theoretically calculated value of the critical temperature($T_c$) of the phase transition of liquid helium is found to agree with the experimentally determined value ($T_c = 2.17~\rm{K}$), although they differs from each other for $q=1$ (undeformed scenario). This throws a light on the understanding of the phenomenon and connects temperature fluctuation(non-equilibrium conditions) with the interactions between atoms qualitatively. More interactions between atoms give rise to more non-equilibrium conditions which is as expected. \noindent {{\bf Keywords}: Tsallis statistics, Bose-Einstein condensation, liquid helium.} △ Less

Submitted 4 February, 2018; v1 submitted 22 August, 2017; originally announced August 2017.

Comments: 21 pages, 6 figures, 4 appendices

Report number: PHYSA19100

Journal ref: Physica A: Statistical Mechanics and its Applications 497C (2018) pp. 272-284

$q$-deformed Einstein's Model to Describe Specific Heat of Solid

Authors: Atanu Guha, Prasanta Kumar Das

Abstract: Realistic phenomena can be described more appropriately using generalized canonical ensemble, with proper parameter sets involved. We have generalized the Einstein's theory for specific heat of solid in Tsallis statistics, where the temperature fluctuation is introduced into the theory via the fluctuation parameter $q$. At low temperature the Einstein's curve of the specific heat in the nonextensi… ▽ More Realistic phenomena can be described more appropriately using generalized canonical ensemble, with proper parameter sets involved. We have generalized the Einstein's theory for specific heat of solid in Tsallis statistics, where the temperature fluctuation is introduced into the theory via the fluctuation parameter $q$. At low temperature the Einstein's curve of the specific heat in the nonextensive Tsallis scenario exactly lies on the experimental data points. Consequently this $q$-modified Einstein's curve is found to be overlap** with the one predicted by Debye. Considering only the temperature fluctuation effect(even without considering more than one mode of vibration is being triggered) we found that the $C_V$ vs $T$ curve is as good as obtained by considering the different modes of vibration as suggested by Debye. Generalizing the Einstein's theory in Tsallis statistics we found that a unique value of the Einstein temperature $θ_E$ along with a temperature dependent deformation parameter $q(T)$, can well describe the phenomena of specific heat of solid i.e. the theory is equivalent to Debye's theory with a temperature dependent $θ_D$. △ Less

Submitted 26 December, 2017; v1 submitted 6 August, 2017; originally announced August 2017.

Comments: 20 pages, 7 figures, 3 appendices, Keywords: Tsallis statistics, temperature fluctuation, specific heat of solid, Einstein's theory, Debye's modification

Report number: PHYSA19000

Journal ref: Physica A: Statistical Mechanics and its Applications 495C (2018) pp. 18-29

Blackbody Radiation in $q$-deformed Statistics

Authors: Atanu Guha, Prasanta Kumar Das

Abstract: More general canonical ensemble which gives rise the generalized statistics or $q$-deformed statistics can represent the realistic scenario than the ideal one, with proper parameter sets involved. We study the Planck's law of blackbody radiation, Wein's and Rayleigh-Jeans radiation formulae from the point of view of $q$-deformed statistics. We find that the blackbody energy spectrum curve for a gi… ▽ More More general canonical ensemble which gives rise the generalized statistics or $q$-deformed statistics can represent the realistic scenario than the ideal one, with proper parameter sets involved. We study the Planck's law of blackbody radiation, Wein's and Rayleigh-Jeans radiation formulae from the point of view of $q$-deformed statistics. We find that the blackbody energy spectrum curve for a given temperature $T$ corresponding to different $q$ values differs from each other: the location of the peak(i.e. $ν_m$) of the energy distribution $u_ν$ (corresponding to different $q$ ) shifted towards higher $ν$ for higher $q$. From the $q$-deformed Wein's displacement law, we find that $λ_m T$ varies from $0.0029~\rm{m~K}$ to $0.0017~\rm{m~K}$ as the deformation parameter $q$ varies from $1.0$(undeformed) to $1.1$(deformed). △ Less

Submitted 30 June, 2017; originally announced June 2017.

Comments: 13 Pages, 5 figures

Nonlinear modeling of wave-topography interactions, shear instabilities and shear induced wave breaking using vortex method

Authors: Divyanshu Bhardwaj, Anirban Guha

Abstract: Theoretical studies on linear shear instabilities often use simple velocity and density profiles (e.g. constant, piecewise) for obtaining good qualitative and quantitative predictions of the initial disturbances. Furthermore, such simple profiles provide a minimal model for obtaining a mechanistic understanding of otherwise elusive shear instabilities. However, except a few specific cases, the eff… ▽ More Theoretical studies on linear shear instabilities often use simple velocity and density profiles (e.g. constant, piecewise) for obtaining good qualitative and quantitative predictions of the initial disturbances. Furthermore, such simple profiles provide a minimal model for obtaining a mechanistic understanding of otherwise elusive shear instabilities. However, except a few specific cases, the efficacy of simple profiles has remained limited to the linear stability paradigm. In this work we have proposed a general framework that can simulate the fully nonlinear evolution of a variety of stratified shear instabilities as well as wave-wave and wave-topography interaction problems having simple profiles. To this effect, we have modified the classical vortex method by extending the Birkhoff-Rott equation to multiple interfaces, and furthermore, have incorporated background shear across a density interface. The latter is more subtle, and originates from the understanding that Bernoulli's equation is not just limited to irrotational flows, but can be modified to make it applicable for piecewise velocity profiles. We have solved diverse problems that can be essentially reduced to the multiple interacting interfaces paradigm, e.g. spilling and plunging breakers, stratified shear instabilities like Holmboe and Taylor-Caulfield, jet flows, and even wave-topography interaction problem like Bragg resonance. Free-slip boundary being a vortex sheet, its effect can also be effectively captured using vortex method. We found that the minimal models capture key nonlinear features, e.g. wave breaking features like cusp formation and roll-ups, which are observed in experiments and/or extensive simulations with smooth, realistic profiles. △ Less

Submitted 27 September, 2017; v1 submitted 14 June, 2017; originally announced June 2017.

Effect of free surface on submerged stratified shear instabilities

Authors: Mihir H. Shete, Anirban Guha

Abstract: Here we have considered the effects of shallowness of the domain as well as the air-water free surface on the stratified shear instabilities of the fluid underneath. First, we numerically solve the non-Boussinesq Taylor-Goldstein equation for smooth velocity and density profiles of a model shear layer with a free surface. When the depth of the fluid is relatively shallow compared to the shear laye… ▽ More Here we have considered the effects of shallowness of the domain as well as the air-water free surface on the stratified shear instabilities of the fluid underneath. First, we numerically solve the non-Boussinesq Taylor-Goldstein equation for smooth velocity and density profiles of a model shear layer with a free surface. When the depth of the fluid is relatively shallow compared to the shear layer thickness, the surface gravity waves existing at the free surface come closer to the waves existing in the shear layer. This can lead to resonant wave interactions, making the flow unstable to more varieties of modal instabilities. To obtain a deeper understanding of the instability mechanisms, we have performed analytical studies with broken-line profiles. Furthermore, reduced order broken-line profiles have also been developed, based on which dispersion diagrams are constructed. Through these diagrams we have underpinned the resonantly interacting waves leading to each type of instability. Two new instabilities have been found; one of them, referred to as the surface gravity - interfacial gravity (SG-IG) mode, arises due to the interaction between a surface gravity wave and an interfacial gravity wave, and would be absent if there is no internal density stratification. The other one - the surface gravity - lower vorticity (SG-LV) mode, which arises due to the interaction between a surface gravity wave and the lower vorticity wave, surpasses Kelvin-Helmholtz instability to become the most unstable mode. Except for Holmboe instability, remarkable differences are observed in all other instabilities occurring in shallow domains when the air-water interface is replaced by a rigid lid. We infer that the rigid-lid approximation is valid for large vertical domains. We have also shown that if shear is absent at the free surface, our problem can be modeled using a Boussinesq type approximation. △ Less

Submitted 20 March, 2018; v1 submitted 9 April, 2017; originally announced April 2017.

Comments: 27 pages, 11 figures

Journal ref: J. Fluid Mech. 843 (2018) 98-125

A generalized action-angle representation of wave interaction in stratified shear flows

Authors: Eyal Heifetz, Anirban Guha

Abstract: In this paper we express the linearized dynamics of interacting interfacial waves in stratified shear flows in the compact form of action-angle Hamilton equations. The pseudo-energy serves as the Hamiltonian of the system, the action coordinates are the contribution of the interfacial waves to the wave-action, and the angles are their phases. The term "generalized action-angle" aims to emphasize t… ▽ More In this paper we express the linearized dynamics of interacting interfacial waves in stratified shear flows in the compact form of action-angle Hamilton equations. The pseudo-energy serves as the Hamiltonian of the system, the action coordinates are the contribution of the interfacial waves to the wave-action, and the angles are their phases. The term "generalized action-angle" aims to emphasize that the action of each wave is generally time dependent and this allows instability. An attempt is made to relate this formalism to the action at a distance resonance instability mechanism between counter-propagating vorticity waves via the global conservations of pseudo-energy and pseudo-momentum. △ Less

Submitted 3 August, 2017; v1 submitted 8 March, 2017; originally announced March 2017.

Analytic Connectivity in General Hypergraphs

Authors: Ashwin Guha, Muni Sreenivas Pydi, Biswajit Paria, Ambedkar Dukkipati

Abstract: In this paper we extend the known results of analytic connectivity to non-uniform hypergraphs. We prove a modified Cheeger's inequality and also give a bound on analytic connectivity with respect to the degree sequence and diameter of a hypergraph. In this paper we extend the known results of analytic connectivity to non-uniform hypergraphs. We prove a modified Cheeger's inequality and also give a bound on analytic connectivity with respect to the degree sequence and diameter of a hypergraph. △ Less

Submitted 17 January, 2017; originally announced January 2017.