Searcharxiv⌕ Search

arXiv subjects

Anirban Das

Publications and source records attributed to Anirban Das.

72 records · Page 4Linked to original sources
arXiv2020

Skedulix: Hybrid Cloud Scheduling for Cost-Efficient Execution of Serverless Applications

We present a framework for scheduling multifunction serverless applications over a hybrid public-private cloud. A set of serverless jobs is input as a batch, and the objective is to schedule function executions over the hybrid platform to minimize the cost of public cloud use, while completing all jobs by a specified deadline. As this scheduling problem is NP-Hard, we propose a greedy algorithm that dynamically determines both the order and placement of each function execution using predictive models of function execution time and network latencies. We present a prototype implementation of our framework that uses AWS Lambda and OpenFaaS, for the public and private cloud, respectively. We evaluate our prototype in live experiments using a mixture of compute and I/O heavy serverless applications. Our results show that our framework can achieve a speedup in batch processing of up to 1.92 times that of an approach that uses only the private cloud, at 40.5% the cost of an approach that uses only the public cloud.

cs.DC↗
arXiv2020

Performance Optimization for Edge-Cloud Serverless Platforms via Dynamic Task Placement

We present a framework for performance optimization in serverless edge-cloud platforms using dynamic task placement. We focus on applications for smart edge devices, for example, smart cameras or speakers, that need to perform processing tasks on input data in real to near-real time. Our framework allows the user to specify cost and latency requirements for each application task, and for each input, it determines whether to execute the task on the edge device or in the cloud. Further, for cloud executions, the framework identifies the container resource configuration needed to satisfy the performance goals. We have evaluated our framework in simulation using measurements collected from serverless applications in AWS Lambda and AWS Greengrass. In addition, we have implemented a prototype of our framework that runs in these same platforms. In experiments with our prototype, our models can predict average end-to-end latency with less than 6% error, and we obtain almost three orders of magnitude reduction in end-to-end latency compared to edge-only execution.

cs.DC↗
arXiv2020

Mixed WIMP-axion dark matter

We study the experimental constraints on a model of a two-component dark matter, consisting of the QCD axion, and a scalar particle, both contributing to the dark matter relic abundance of the Universe. The global Peccei-Quinn symmetry of the theory can be spontaneously broken down to a residual $\mathbb{Z}_2$ symmetry, thereby identifying this scalar as a stable weakly interacting massive particle, i.e., a dark matter candidate, in addition to the axion. We perform a comprehensive study of the model using the latest data from dark matter direct and indirect detection experiments, as well as new physics searches at the Large Hadron Collider. We find that although the model is mostly constrained by the dark matter detection experiments, it is still viable around a small region of the parameter space where the scalar dark matter is half as heavy as the Standard Model Higgs. In this allowed region, the bounds from these experiments are evaded due to a cancellation mechanism in the dark matter--Higgs coupling. The collider search results, however, are shown to impose weak bounds on the model.

hep-ph↗
arXiv2020

Diffusion and Consensus in a Weakly Coupled Network of Networks

We study diffusion and consensus dynamics in a Network of Networks model. In this model, there is a collection of sub-networks, connected to one another using a small number of links. We consider a setting where the links between networks have small weights, or are used less frequently than links within each sub-network. Using spectral perturbation theory, we analyze the diffusion rate and convergence rate of the investigated systems. Our analysis shows that the first order approximation of the diffusion and convergence rates is independent of the topologies of the individual graphs; the rates depend only on the number of nodes in each graph and the topology of the connecting edges. The second order analysis shows a relationship between the diffusion and convergence rates and the information centrality of the connecting nodes within each sub-network. We further highlight these theoretical results through numerical examples.

cs.SI↗
arXiv2019

Privacy is What We Care About: Experimental Investigation of Federated Learning on Edge Devices

Federated Learning enables training of a general model through edge devices without sending raw data to the cloud. Hence, this approach is attractive for digital health applications, where data is sourced through edge devices and users care about privacy. Here, we report on the feasibility to train deep neural networks on the Raspberry Pi4s as edge devices. A CNN, a LSTM and a MLP were successfully trained on the MNIST data-set. Further, federated learning is demonstrated experimentally on IID and non-IID samples in a parametric study, to benchmark the model convergence. The weight updates from the workers are shared with the cloud to train the general model through federated learning. With the CNN and the non-IID samples a test-accuracy of up to 85% could be achieved within a training time of 2 minutes, while exchanging less than $10$ MB data per device. In addition, we discuss federated learning from an use-case standpoint, elaborating on privacy risks and labeling requirements for the application of emotion detection from sound. Based on the experimental findings, we discuss possible research directions to improve model and system performance. Finally, we provide best practices for a practitioner, considering the implementation of federated learning.

cs.LG↗
arXiv2019

Ballistic Dark Matter oscillates above $Λ$CDM

Dark matter may have been relativistic and collisional until relatively late times and become cold and collisionless after a phase transition before the matter-radiation equality of the standard $Λ$CDM cosmology. We show that such a dark matter has large peculiar velocities due to acoustic oscillations before the phase transition, and evolves ballistically after the phase transition in the collisionless phase until the initial acoustic velocities are redshifted away. We show that this Ballistic Dark Matter (BDM) results in new non-trivial interesting features in the cosmological observables. In particular, the linear matter power spectrum exhibits acoustic oscillations on scales smaller than the Hubble scale at the time of phase transition, and for fast transitions the power at the acoustic peaks in the matter power spectrum exceeds that in a $Λ$CDM cosmology. If BDM only forms a part of the total dark matter, an odd vs. even acoustic peak asymmetry becomes prominent. We give an approximate analytical treatment of the linear perturbations in BDM, explaining these features. We also discuss the possibility to constrain BDM using cosmic microwave background and large scale structure data.

astro-ph.CO↗
arXiv2018

EdgeBench: Benchmarking Edge Computing Platforms

The emerging trend of edge computing has led several cloud providers to release their own platforms for performing computation at the 'edge' of the network. We compare two such platforms, Amazon AWS Greengrass and Microsoft Azure IoT Edge, using a new benchmark comprising a suite of performance metrics. We also compare the performance of the edge frameworks to cloud-only implementations available in their respective cloud ecosystems. Amazon AWS Greengrass and Azure IoT Edge use different underlying technologies, edge Lambda functions vs. containers, and so we also elaborate on platform features available to developers. Our study shows that both of these edge platforms provide comparable performance, which nevertheless differs in important ways for key types of workloads used in edge applications. Finally, we discuss several current issues and challenges we faced in deploying these platforms.

cs.NI↗
arXiv2018

Optimal control of Markov-modulated multiclass many-server queues

We study multiclass many-server queues for which the arrival, service and abandonment rates are all modulated by a common finite-state Markov process. We assume that the system operates in the "averaged" Halfin-Whitt regime, which means that it is critically loaded in the average sense, although not necessarily in each state of the Markov process. We show that under any static priority policy, the Markov-modulated diffusion-scaled queueing process is geometrically ergodic. This is accomplished by employing a solution to an associated Poisson equation in order to construct a suitable Lyapunov function. We establish a functional central limit theorem for the diffusion-scaled queueing process and show that the limiting process is a controlled diffusion with piecewise linear drift and constant covariance matrix. We address the infinite-horizon discounted and long-run average (ergodic) optimal control problems and establish asymptotic optimality.

math.PR↗
arXiv2018

Critical neuronal models with relaxed timescales separation

Power laws in nature are considered to be signatures of complexity. The theory of self-organized criticality (SOC) was proposed to explain their origins. A longstanding principle of SOC is the \emph{separation of timescales} axiom. It dictates that external input is delivered to the system at a much slower rate compared to the timescale of internal dynamics. The statistics of neural avalanches in the brain was demonstrated to follow a power law, indicating closeness to a critical state. Moreover, criticality was shown to be a beneficial state for various computations leading to the hypothesis, that the brain is a SOC system. However, for neuronal systems that are constantly bombarded by incoming signals, separation of timescales assumption is unnatural. Recently it was experimentally demonstrated that a proper correction of the avalanche detection algorithm to account for the increased drive during task performance leads to a change of the power-law exponent from $1.5$ to approximately $1.3$, but there is so far no theoretical explanation for this change. Here we investigate the importance of timescales separation, by partly abandoning it in various models. We achieve it by allowing for external input during the avalanche, without compromising the separation of avalanches. We develop an analytic treatment and provide numerical simulations of a simple neuronal model.

q-bio.NC↗
arXiv2018

New Dissipation Mechanisms from Multi-level Dark Matter Scattering

Multi-level dark matter with diagonal and off-diagonal interactions shows a rich phenomenology in its self-scattering. If the interactions are mediated by a particle that is less massive than the dark matter, Sommerfeld effect can lead to resonant enhancement of the scattering. For mediators lighter than the level separation, dark matter particles can upscatter to excited states and de-excite by emitting these mediators. We compute these cross-sections, both above and below the kinematic threshold, in a generic two-component dark matter model and identify the large inelastic cross-section as a result of maximal mixing between the two states. A new route for cooling of large dark matter halos and a new drag force between two colliding halos are identified and shown to arise purely from the inelastic scattering.

hep-ph↗
arXiv2017

Selection Rule for Enhanced Dark Matter Annihilation

We point out a selection rule for enhancement (suppression) of odd (even) partial waves of dark matter coannihilation or annihilation using Sommerfeld effect. Using this, the usually velocity-suppressed p-wave annihilation can dominate the annihilation signals in the present Universe. The selection mechanism is a manifestation of the exchange symmetry of identical incoming particles, and generic for multi-state DM with off-diagonal long-range interactions. As a consequence, the relic and late-time annihilation rates are parametrically different and a distinctive phenomenology, with large but strongly velocity-dependent annihilation rates, is predicted.

hep-ph↗
arXiv2017

New effects of non-standard self-interactions of neutrinos in a supernova

Neutrino self-interactions are known to lead to non-linear collective flavor oscillations in a core-collapse supernova. We point out new possible effects of non-standard self-interactions (NSSI) of neutrinos on flavor conversions in a two-flavor framework. We show that, for a single-energy neutrino-antineutrino ensemble, a flavor instability is generated even in normal hierarchy for large enough NSSI. Using a toy model for the neutrino spectra, we show that flavor-preserving NSSI lead to pinching of spectral swaps, while flavor-violating NSSI cause swaps to develop away from a spectral crossing or even in the absence of a spectral crossing. Consequently, NSSI could give rise to collective oscillations and spectral splits even during neutronization burst, for both hierarchies.

hep-ph↗
arXiv2017

Constructions involving interaction of Markov process's with a random environment

Recently Belopolskaya and Suhov studied Markov process's in a random environment, where the environment changes in likeness to a Markov process. Constructions were made to allow the process to "interact with the environment", this was done in such a manner that the combined process could be studied by studying the individual ones. A new construction is made which has similar properties. This construction can be carried out with more relaxed conditions than the ones previously done.

math.PR↗
arXiv2014

Thermodynamic Geometry and Phase Transitions of Dyonic Charged AdS Black Holes

We investigate phase transitions and critical phenomena of four dimensional dyonic charged AdS black holes in the framework of thermodynamic geometry. In a mixed canonical grand canonical ensemble with a fixed electric charge and varying magnetic charge these black holes exhibit liquid gas like first order phase transition culminating in a second order critical point similar to the Van der Waals gas. We show that the thermodynamic scalar curvature R for these black holes follow our proposed geometrical characterization of the R-crossing Method for the first order liquid gas like phase transition and exhibits a divergence at the second order critical point. The pattern of R crossing and divergence exactly corresponds to those of a Van der Waals gas described by us in an earlier work.

hep-th↗
arXiv2011

Entanglement and its evolution after a quench in the presence of an energy current

We study the Ising spin chain with a Dzyaloshinskii-Moriya interaction focusing on the static and dynamic properties of the entanglement entropy, following a quantum quench. We show that the effects of the additional anisotropic interaction on the phase diagram and on the dynamics of the system are captured by the properties of the entanglement entropy. In particular, the model provides a way to study the quench dynamics in a system with an energy current. We consider quenches starting from an initial excited state of the Ising spin chain, and we analyze the effects of different initial conditions.

cond-mat.stat-mech↗
arXiv2008

Entanglement monogamy of multipartite higher-dimensional quantum systems using convex-roof extended negativity

We propose replacing concurrence by convex-roof extended negativity (CREN) for studying monogamy of entanglement (MoE). We show that all proven MoE relations using concurrence can be rephrased in terms of CREN. Furthermore we show that higher-dimensional (qudit) extensions of MoE in terms of CREN are not disproven by any of the counterexamples used to disprove qudit extensions of MoE in terms of concurrence. We further test the CREN version of MoE for qudits by considering fully or partially coherent mixtures of a qudit W-class state with the vacuum and show that the CREN version of MoE for qudits is satisfied in this case as well. The CREN version of MoE for qudits is thus a strong conjecture with no obvious counterexamples.

quant-ph↗
arXiv2004

An alternate model for protective measurements of two-level systems

In this article we propose an alternate model for the so called {\it protective measurements}, more appropriately {\it adiabatic measurements} of a spin 1/2 system where the {\it apparatus} is also a quantum system with a {\em finite dimensional Hilbert space}. This circumvents several technical as well as conceptual issues that arise when dealing with an infinite dimensional Hilbert space as in the analysis of conventional Stern-Gerlach experiment. Here also it is demonstrated that the response of the detector is continuous and it {\it directly} measures {\em expectation values without altering the state of the system}(when the unknown original state is a {\it nondegenerate eigenstate of the system Hamiltonian}, in the limit of {\em ideal} adiabatic conditions. We have also computed the corrections arising out of the inevitable departures from ideal adiabaticity i.e the time of measurement being large but finite. To overcome the {\em conceptual} difficulties with a {\it quantum apparatus}, we have simulated a {\it classical apparatus} as a {\em large} assembly of spin-1/2 systems. We end this article with a conclusion and a discussion of some future issues.

quant-ph↗