-
Interplay of Traditional Methods and Machine Learning Algorithms for Tagging Boosted Objects
Authors:
Camellia Bose,
Amit Chakraborty,
Shreecheta Chowdhury,
Saunak Dutta
Abstract:
Interest in deep learning in collider physics has been growing in recent years, specifically in applying these methods in jet classification, anomaly detection, particle identification etc. Among those, jet classification using neural networks is one of the well-established areas. In this review, we discuss different tagging frameworks available to tag boosted objects, especially boosted Higgs bos…
▽ More
Interest in deep learning in collider physics has been growing in recent years, specifically in applying these methods in jet classification, anomaly detection, particle identification etc. Among those, jet classification using neural networks is one of the well-established areas. In this review, we discuss different tagging frameworks available to tag boosted objects, especially boosted Higgs boson and top quark, at the Large Hadron Collider (LHC). Our aim is to study the interplay of traditional jet substructure based methods with the state-of-the-art machine learning ones. In this methodology, we would gain some interpretability of those machine learning methods, and which in turn helps to propose hybrid taggers relevant for tagging of those boosted objects belonging to both Standard Model (SM) and physics beyond the SM.
△ Less
Submitted 2 August, 2024;
originally announced August 2024.
-
Effect of sweep angle on three-dimensional vortex dynamics over plunging wings
Authors:
Alex Cavanagh,
Chandan Bose,
Kiran Ramesh
Abstract:
The effects of sweep angle and reduced frequency on the leading-edge vortex (LEV) structure over flapping swept wings in the Reynolds number ($Re$) range of $\mathbf{O}(10^4)$ are yet to be completely understood. With increasing interest in designing bio-inspired micro-air-vehicles (MAVs), understanding LEV dynamics in such scenarios is imperative. This study investigates the effects of three diff…
▽ More
The effects of sweep angle and reduced frequency on the leading-edge vortex (LEV) structure over flapping swept wings in the Reynolds number ($Re$) range of $\mathbf{O}(10^4)$ are yet to be completely understood. With increasing interest in designing bio-inspired micro-air-vehicles (MAVs), understanding LEV dynamics in such scenarios is imperative. This study investigates the effects of three different sweep angles ($Λ= 0^\circ$, $30^\circ$ and $60^\circ$) on LEV dynamics through high-fidelity improved delayed detached eddy simulation (IDDES) to analyze the underlying flow physics. Plunge ramp kinematics at two different reduced frequencies ($k = 0.05$ and $0.4$) are studied to investigate the unsteady motion effects on LEV characteristics. The leading-edge suction parameter (LESP) concept is applied to determine LEV initiation, and the results are verified against flow field visualization for swept-wing geometries. The force partitioning method (FPM) is used to investigate the spanwise lift distribution resulting from the LEV. Distinct peaks in the lift coefficient occur for the high reduced frequency case due to the impulse-like plunging acceleration. This causes the LEV to detach from the leading edge more quickly and convect faster, significantly affecting the lift generated by the wing. As reduced frequency increases, the LEV breakdown mechanism switches from vortex bursting to LEV leg-induced instabilities. These results provide insights into the complex vortex structures surrounding swept wings at $Re = 20,000$, and the impact both sweep angle and reduced frequency have on the lift contribution of these flow features.
△ Less
Submitted 6 July, 2024;
originally announced July 2024.
-
Effect of a fixed downstream cylinder on the flow-induced vibration of an elastically-supported primary cylinder
Authors:
Junlei Wang,
Shenfang Li,
Daniil Yurchenko,
Hongjun Zhu,
Chandan Bose
Abstract:
This paper numerically investigates the influence of a fixed downstream control cylinder on the flow-induced vibration of an elastically-supported primary cylinder. These two cylinders are situated in a tandem arrangement with small dimensionless centre-to-centre spacing ($L/D$, $L$ is the intermediate spacing, and $D$ is the cylinder diameter). The present two-dimensional (2D) simulations are car…
▽ More
This paper numerically investigates the influence of a fixed downstream control cylinder on the flow-induced vibration of an elastically-supported primary cylinder. These two cylinders are situated in a tandem arrangement with small dimensionless centre-to-centre spacing ($L/D$, $L$ is the intermediate spacing, and $D$ is the cylinder diameter). The present two-dimensional (2D) simulations are carried out in the low Reynolds number ($Re$) regime. The primary focus of this study is to reveal the underlying flow physics behind the transition from vortex-induced vibration to galloping in the response of the primary cylinder due to the presence of another fixed downstream cylinder. Two distinct flow field regimes, namely steady flow and alternate attachment regimes, are observed for different $L/D$ and Re values. Depending on the evolution of the near-field flow structures, four different wake patterns - `2S', `2P', `2C', and `aperiodic' - are observed. The corresponding vibration response of the upstream cylinder is characterized as interference galloping and extended vortex-induced vibration. As the $L/D$ ratio increases, the lift enhancement due to flow-induced vibration is seen to be weakened. The detailed correlation between the force generation and the near-wake interactions is investigated. The present findings will augment the understanding of vibration reduction or flow-induced energy harvesting of tandem cylindrical structures.
△ Less
Submitted 14 May, 2024;
originally announced May 2024.
-
Effect of structural parameters on the synchronization characteristics in a stall-induced aeroelastic system
Authors:
Dheeraj Tripathi,
Chandan Bose,
Sirshendu Mondal,
J Venkatramani
Abstract:
This study focuses on discerning the role of structural parameters on the bifurcation characteristics and the underlying synchronization mechanism in an aeroelastic system undergoing nonlinear stall behaviour. To that end, wind tunnel experiments are performed on a NACA 0012 airfoil capable of undergoing bending (plunging) and torsional (pitching) oscillations under scenarios involving nonlinear a…
▽ More
This study focuses on discerning the role of structural parameters on the bifurcation characteristics and the underlying synchronization mechanism in an aeroelastic system undergoing nonlinear stall behaviour. To that end, wind tunnel experiments are performed on a NACA 0012 airfoil capable of undergoing bending (plunging) and torsional (pitching) oscillations under scenarios involving nonlinear aerodynamic loads, i.e., dynamic stall conditions. Flow conditions under both deterministic/sterile flows and fluctuating/stochastic flows are fostered. The structure possesses continuous or polynomial-type stiffness nonlinearities, and therefore, is an aeroelastic experiment involving both structural and aerodynamic nonlinearities. We discern the bifurcation routes for a range of key structural parameters such as frequency ratio, static imbalance, and the extent of structural nonlinearity. In addition to interesting and atypical routes to stall-induced instabilities, we systematically demonstrate the role of modal interactions - via a synchronization analysis - over the manifestation of these instabilities. To the best of the authors' knowledge, this is perhaps the first study to document the role of multiple structural parameters on a stall-induced aeroelastic system, and in turn, cast the physical mechanism behind these dynamical transitions from the vantage of synchronization.
△ Less
Submitted 22 April, 2024;
originally announced April 2024.
-
Controlling the chaotic wake of a flapping foil by tuning its chordwise flexibility
Authors:
Chhote Lal Shah,
Dipanjan Majumdar,
Chandan Bose,
Sunetra Sarkar
Abstract:
Effects of chord-wise flexibility as an instrument to control chaotic transitions in the wake of a flexible flapping foil have been studied here using an immersed boundary method-based in-house fluid-structure-interaction solver. The ability of the flapping foil at an optimum level of flexibility to inhibit chaotic transition, otherwise encountered in a similar but rigid configuration, has been hi…
▽ More
Effects of chord-wise flexibility as an instrument to control chaotic transitions in the wake of a flexible flapping foil have been studied here using an immersed boundary method-based in-house fluid-structure-interaction solver. The ability of the flapping foil at an optimum level of flexibility to inhibit chaotic transition, otherwise encountered in a similar but rigid configuration, has been highlighted. The rigid foil manifests chaotic transition through a quasi-periodic-intermittency route at high dynamic plunge velocities; whereas, increasing the level of flexibility gradually regularises the aperiodic behaviour through a variety of interesting wake patterns. If flexibility is increased beyond an optimum level, aperiodicity sets in again and robust chaos is restored at very high flexibility levels. The mechanisms of triggering the order-to-chaos transition are different between the rigid and the high flexibility cases. Along the route to order and back to chaos, the flexible foil exhibits different flow-field behaviours, including far-wake switching, primary \& secondary vortex streets, bifurcated wakes and interactive vortices between the bifurcated wakes. The underlying interaction mechanisms of the flow-field vortices responsible for the associated dynamical signatures of the wake have been closely tracked. This study further examines the optimum propulsive performance range of the flexible flapper and investigates its connection with the periodicity/regularity of the system.
△ Less
Submitted 26 March, 2024;
originally announced March 2024.
-
LRMP: Layer Replication with Mixed Precision for Spatial In-memory DNN Accelerators
Authors:
Abinand Nallathambi,
Christin David Bose,
Wilfried Haensch,
Anand Raghunathan
Abstract:
In-memory computing (IMC) with non-volatile memories (NVMs) has emerged as a promising approach to address the rapidly growing computational demands of Deep Neural Networks (DNNs). Mapping DNN layers spatially onto NVM-based IMC accelerators achieves high degrees of parallelism. However, two challenges that arise in this approach are the highly non-uniform distribution of layer processing times an…
▽ More
In-memory computing (IMC) with non-volatile memories (NVMs) has emerged as a promising approach to address the rapidly growing computational demands of Deep Neural Networks (DNNs). Mapping DNN layers spatially onto NVM-based IMC accelerators achieves high degrees of parallelism. However, two challenges that arise in this approach are the highly non-uniform distribution of layer processing times and high area requirements. We propose LRMP, a method to jointly apply layer replication and mixed precision quantization to improve the performance of DNNs when mapped to area-constrained NVM-based IMC accelerators. LRMP uses a combination of reinforcement learning and integer linear programming to search the replication-quantization design space using a model that is closely informed by the target hardware architecture. Across five DNN benchmarks, LRMP achieves 2.8-9$\times$ latency and 11.8-19$\times$ throughput improvement at iso-accuracy.
△ Less
Submitted 5 December, 2023;
originally announced December 2023.
-
Wake-induced response of vibro-impacting systems
Authors:
Rohit Chawla,
Aasifa Rounak,
Chandan Bose,
Vikram Pakrashi
Abstract:
The stability and bifurcation behaviour of a wake-induced vibro-impacting oscillator is studied. The effects of a discontinuity on the stability of the structure while it is undergoing phase-locked motions due to the surrounding fluid-structure interactions (FSI) are examined. The primary structure and the near wake dynamics are modelled as a harmonic oscillator and a Van der Pol oscillator, respe…
▽ More
The stability and bifurcation behaviour of a wake-induced vibro-impacting oscillator is studied. The effects of a discontinuity on the stability of the structure while it is undergoing phase-locked motions due to the surrounding fluid-structure interactions (FSI) are examined. The primary structure and the near wake dynamics are modelled as a harmonic oscillator and a Van der Pol oscillator, respectively, and are weakly coupled to each other via acceleration coupling. Qualitative changes in the dynamical behaviour of this system are investigated in the context of discontinuity-induced bifurcations (DIBs) that result from the interaction of fluid flow and non-smoothness from the primary structure. Phenomenological behaviours like the co-existence of attractors and period-adding cascades of limit cycles separated by chaotic orbits are observed. The existence of these phenomena is demonstrated via stability analysis using Floquet theory and the associated Lyapunov spectra. In addition, the behaviour of orbits in the local neighbourhood of the barrier is defined using a higher-order transverse discontinuity map. This mapping is implemented to obtain the respective Lyapunov exponents. Solutions obtained using this modified algorithm are demonstrated to accurately predict both stable and chaotic regimes, as observed from the corresponding bifurcation diagrams.
△ Less
Submitted 16 May, 2024; v1 submitted 7 May, 2023;
originally announced May 2023.
-
Gust Response of Free-Falling Permeable Plates
Authors:
Chandan Bose,
Ignazio Maria Viola
Abstract:
This paper investigates the effect of transverse gusts on the flight dynamics and descent velocity of two-dimensional free-falling permeable plates. Two-way coupled fluid-structure interaction simulations are carried out for a range of Galilei number ($Ga$) from 10 and 50, nondimensional mass ($m$) from 0.5 to 2, and a fixed Darcy number of $10^{-4}$. The present results show that the plate falls…
▽ More
This paper investigates the effect of transverse gusts on the flight dynamics and descent velocity of two-dimensional free-falling permeable plates. Two-way coupled fluid-structure interaction simulations are carried out for a range of Galilei number ($Ga$) from 10 and 50, nondimensional mass ($m$) from 0.5 to 2, and a fixed Darcy number of $10^{-4}$. The present results show that the plate falls steadily in quiescent flow at the lowest $Ga$ and $m$ values, whereas fluttering and tumbling are observed for increased $Ga$ and /or $m$. A transverse (horizontal) gust temporarily decreases the descent velocity of the plate in the transient regime and can also uplift the plate. The gust effect increases with the gust ratio, $Ga$ and $m$. The underlying uplifting mechanism is not directly related to the permeability, and it is thus likely to occur also for impermeable bodies. The present findings might provide insights to interpret the effect of turbulence on the terminal velocity of free-falling bodies and inform the design of insect-scale flyers passively transported by the wind.
△ Less
Submitted 18 April, 2023; v1 submitted 9 April, 2023;
originally announced April 2023.
-
Porous plates at incidence
Authors:
Chandan Bose,
Callum Bruce,
Ignazio Maria Viola
Abstract:
This paper investigates the effect of permeability on two-dimensional rectangular plates at incidences. The flow topology is investigated for Reynolds number ($Re$) values between 30 and 90, and the forces on the plate are discussed for $Re=30$, where the wake is found to be steady for any value of the Darcy number ($Da$) and the flow incidence ($α$). At $Re=30$, for a plate normal to the stream a…
▽ More
This paper investigates the effect of permeability on two-dimensional rectangular plates at incidences. The flow topology is investigated for Reynolds number ($Re$) values between 30 and 90, and the forces on the plate are discussed for $Re=30$, where the wake is found to be steady for any value of the Darcy number ($Da$) and the flow incidence ($α$). At $Re=30$, for a plate normal to the stream and vanishing $Da$, the wake shows a vortex dipole with a stagnation point on the plate surface. With increasing $Da$, the separation between the vortex dipole and the plate increases; the vortex dipole shortens and is eventually annihilated at a critical $Da$. For any value of $Da$ below the critical one, the vortex dipole disappears with decreasing $α$. However, at low $Da$, the two saddle-node pairs merge at the same $α$, annihilating the dipole; while at high $Da$, they merge at different $α$, resulting in a single recirculating region for intermediate incidences. The magnitudes of lift, drag, and torque decrease with $Da$. Nevertheless, there exists a range of $Da$ and $α$, where the magnitude of the plate-wise force component increases with $Da$, driven by the shear on the plate's pressure side. Finally, the analysis of the fluid impulse suggests that the lift and drag reduction with $Da$ are associated with the weakening of the leading and trailing edge shear layer, respectively. The present findings will be directly beneficial in understanding the role of permeability on small porous wings.
△ Less
Submitted 27 June, 2024; v1 submitted 23 March, 2023;
originally announced March 2023.
-
Boosted top tagging and its interpretation using Shapley values
Authors:
Biplob Bhattacherjee,
Camellia Bose,
Amit Chakraborty,
Rhitaja Sengupta
Abstract:
Top tagging has emerged to be a fast-evolving subject due to the top quark's significant role in probing physics beyond the standard model. For the reconstruction of top jets, machine learning models have shown a significant improvement in the tagging and classification performance compared to the previous methods. In this work, we build top taggers using $N$-Subjettiness ratios and several observ…
▽ More
Top tagging has emerged to be a fast-evolving subject due to the top quark's significant role in probing physics beyond the standard model. For the reconstruction of top jets, machine learning models have shown a significant improvement in the tagging and classification performance compared to the previous methods. In this work, we build top taggers using $N$-Subjettiness ratios and several observables of the Energy Correlation functions as input features to train the eXtreme Gradient BOOSTed decision tree (XGBOOST). It is observed that the performance of the taggers depends on how well the top jets are matched to their truth-level partons. Furthermore, we use SHapley Additive exPlanation (SHAP) framework to calculate the feature importance of the trained models. It helps us to estimate how much each feature of the data contributed to the model's prediction and what regions are of more importance for each input variable. Finally, we combine all the tagger variables to form a hybrid tagger and interpret the results using the Shapley values.
△ Less
Submitted 22 December, 2022;
originally announced December 2022.
-
Prospects for exotic $h\rightarrow 4 τ$ decays in single and di-Higgs boson production at the LHC and future hadron colliders
Authors:
Amit Adhikary,
Shankha Banerjee,
Rahool Kumar Barman,
Brian Batell,
Biplob Bhattacherjee,
Camellia Bose,
Zhuoni Qian,
Michael Spannowsky
Abstract:
We study the prospects for observing exotic decays of the Standard Model Higgs boson $h$ into light beyond the Standard Model scalars $a$ with mass $m_{a} \lesssim m_{h}/2$ in the single Higgs and Higgs pair production channels at the high luminosity run of the Large Hadron Collider (HL-LHC). Discovery prospects for single Higgs production in the gluon-gluon fusion and vector boson fusion modes wi…
▽ More
We study the prospects for observing exotic decays of the Standard Model Higgs boson $h$ into light beyond the Standard Model scalars $a$ with mass $m_{a} \lesssim m_{h}/2$ in the single Higgs and Higgs pair production channels at the high luminosity run of the Large Hadron Collider (HL-LHC). Discovery prospects for single Higgs production in the gluon-gluon fusion and vector boson fusion modes with the Higgs boson decaying via the exotic mode $h \to aa \to 4τ$ are analyzed at the HL-LHC. The projected sensitivity for exotic Higgs decays in the non-resonant Higgs pair production channel $pp \to hh \to (h \to b\bar{b})(h \to aa \to 4τ) \to 2b4τ$ at the HL-LHC and a future $\sqrt{s}=100~$TeV hadron collider (FCC-hh) are also estimated. Furthermore, we study HL-LHC's potential reach for the Higgs-strahlung process in the $2b4τ$ channel, taking into account the contamination from non-resonant Higgs pair production. Finally, the potential reach for resonant Higgs pair production in the $2b4τ$ channel %$pp \to H \to (h \to b\bar{b})(h \to aa \to 4τ)$ at the HL-LHC is also explored for several choices of $\{m_{H},m_{a}\}$. Our studies suggest that significant improvements over existing bounds are achievable in several production channels, motivating new dedicated searches for $h \rightarrow aa \rightarrow 4 τ$ at the HL-LHC and future colliders.
△ Less
Submitted 23 March, 2024; v1 submitted 14 November, 2022;
originally announced November 2022.
-
Fourier Neural Networks for Function Approximation
Authors:
R Subhash Chandra Bose,
Kakarla Yaswanth
Abstract:
The success of Neural networks in providing miraculous results when applied to a wide variety of tasks is astonishing. Insight in the working can be obtained by studying the universal approximation property of neural networks. It is proved extensively that neural networks are universal approximators. Further it is proved that deep Neural networks are better approximators. It is specifically proved…
▽ More
The success of Neural networks in providing miraculous results when applied to a wide variety of tasks is astonishing. Insight in the working can be obtained by studying the universal approximation property of neural networks. It is proved extensively that neural networks are universal approximators. Further it is proved that deep Neural networks are better approximators. It is specifically proved that for a narrow neural network to approximate a function which is otherwise implemented by a deep Neural network, the network take exponentially large number of neurons. In this work, we have implemented existing methodologies for a variety of synthetic functions and identified their deficiencies. Further, we examined that Fourier neural network is able to perform fairly good with only two layers in the neural network. A modified Fourier Neural network which has sinusoidal activation and two hidden layer is proposed and the results are tabulated.
△ Less
Submitted 21 October, 2021;
originally announced November 2021.
-
Classification of COVID-19 from CXR Images in a 15-class Scenario: an Attempt to Avoid Bias in the System
Authors:
Chinmoy Bose,
Anirvan Basu
Abstract:
As of June 2021, the World Health Organization (WHO) has reported 171.7 million confirmed cases including 3,698,621 deaths from COVID-19. Detecting COVID-19 and other lung diseases from Chest X-Ray (CXR) images can be very effective for emergency diagnosis and treatment as CXR is fast and cheap. The objective of this study is to develop a system capable of detecting COVID-19 along with 14 other lu…
▽ More
As of June 2021, the World Health Organization (WHO) has reported 171.7 million confirmed cases including 3,698,621 deaths from COVID-19. Detecting COVID-19 and other lung diseases from Chest X-Ray (CXR) images can be very effective for emergency diagnosis and treatment as CXR is fast and cheap. The objective of this study is to develop a system capable of detecting COVID-19 along with 14 other lung diseases from CXRs in a fair and unbiased manner. The proposed system consists of a CXR image selection technique and a deep learning based model to classify 15 diseases including COVID-19. The proposed CXR selection technique aims to retain the maximum variation uniformly and eliminate poor quality CXRs with the goal of reducing the training dataset size without compromising classifier accuracy. More importantly, it reduces the often hidden bias and unfairness in decision making. The proposed solution exhibits a promising COVID-19 detection scheme in a more realistic situation than most existing studies as it deals with 15 lung diseases together. We hope the proposed method will have wider adoption in medical image classification and other related fields.
△ Less
Submitted 25 September, 2021;
originally announced September 2021.
-
Random Composition of L-S-V Maps Sampled Over Large Parameter Ranges
Authors:
Christopher Bose,
Anthony Quas,
Matteo Tanzi
Abstract:
Liverani-Saussol-Vaienti (L-S-V) maps form a family of piecewise differentiable dynamical systems on $[0,1]$ depending on one parameter $ω\in\mathbb R^+$. These maps are everywhere expanding apart from a neutral fixed point. It is well known that depending on the amount of expansion close to the neutral point, they have either an absolutely continuous invariant probability measure and polynomial d…
▽ More
Liverani-Saussol-Vaienti (L-S-V) maps form a family of piecewise differentiable dynamical systems on $[0,1]$ depending on one parameter $ω\in\mathbb R^+$. These maps are everywhere expanding apart from a neutral fixed point. It is well known that depending on the amount of expansion close to the neutral point, they have either an absolutely continuous invariant probability measure and polynomial decay of correlations ($ω<1$), or a unique physical measure that is singular and concentrated at the neutral point ($ω>1$). In this paper, we study the composition of L-S-V maps whose parameters are randomly sampled from a range in $\mathbb R^+$, and where these two contrasting behaviours are mixed. We show that if the parameters $ω<1$ are sampled with positive probability, then the stationary measure of the random system is absolutely continuous; the annealed decay rate of correlations is close (or in some cases equal) to the fastest rate of decay among those of the sampled systems; and suitably rescaled Birkhoff averages converge to limit laws. In contrast to previous studies where $ω\in [0,1]$, we allow $ ω>1$ in our sampling distribution. We also show that one can obtain similar decay of correlation rates for $ω\in [0,\infty)$, when sampling is done with respect to a family of smooth, heavy-tailed distributions.
△ Less
Submitted 3 March, 2021; v1 submitted 10 July, 2020;
originally announced July 2020.
-
Multiperspective Conformance Analysis of Central Venous Catheter Installation Procedure
Authors:
R. P. Jagadeesh Chandra Bose
Abstract:
Training and practice play a key role in a medical students' attainment of surgical procedural skills. It is beyond doubt that good skills correlate with better clinical outcomes and improved healthcare. Timely, holistic, and effective feedback provide a significant impetus to students acquiring skills with precision. In this paper, we analyze the activities performed by students while learning th…
▽ More
Training and practice play a key role in a medical students' attainment of surgical procedural skills. It is beyond doubt that good skills correlate with better clinical outcomes and improved healthcare. Timely, holistic, and effective feedback provide a significant impetus to students acquiring skills with precision. In this paper, we analyze the activities performed by students while learning the central venous catheter installation procedure. We perform a holistic analysis, using trace alignment, declarative conformance checking, data visualization, and statistical analysis techniques, at different levels of abstraction on control-flow and time perspectives and provide insights at individual student level as well as across students. These insights can help students discover what they are doing right and where they are not and take corrective steps. Instructors can uncover common patterns and mistakes that students demonstrate and think of interventions in their teaching methodology.
△ Less
Submitted 14 August, 2019;
originally announced August 2019.
-
Identifying Implementation Bugs in Machine Learning based Image Classifiers using Metamorphic Testing
Authors:
Anurag Dwarakanath,
Manish Ahuja,
Samarth Sikand,
Raghotham M. Rao,
R. P. Jagadeesh Chandra Bose,
Neville Dubash,
Sanjay Podder
Abstract:
We have recently witnessed tremendous success of Machine Learning (ML) in practical applications. Computer vision, speech recognition and language translation have all seen a near human level performance. We expect, in the near future, most business applications will have some form of ML. However, testing such applications is extremely challenging and would be very expensive if we follow today's m…
▽ More
We have recently witnessed tremendous success of Machine Learning (ML) in practical applications. Computer vision, speech recognition and language translation have all seen a near human level performance. We expect, in the near future, most business applications will have some form of ML. However, testing such applications is extremely challenging and would be very expensive if we follow today's methodologies. In this work, we present an articulation of the challenges in testing ML based applications. We then present our solution approach, based on the concept of Metamorphic Testing, which aims to identify implementation bugs in ML based image classifiers. We have developed metamorphic relations for an application based on Support Vector Machine and a Deep Learning based application. Empirical validation showed that our approach was able to catch 71% of the implementation bugs in the ML applications.
△ Less
Submitted 16 August, 2018;
originally announced August 2018.
-
Influence of polarization and self-polarization charges on impurity binding energy in spherical quantum dot with parabolic confinement
Authors:
Samrat Sarkar,
Supratik Sarkar,
Chayanika Bose
Abstract:
We present a general formulation of the ground state binding energy of a shallow hydrogenic impurity in spherical quantum dot with parabolic confinement, considering the effects of polarization and self energy. The variational approach within the effective mass approximation is employed here. The binding energy of an on-center impurity is computed for a $\mathrm{GaAs}$/…
▽ More
We present a general formulation of the ground state binding energy of a shallow hydrogenic impurity in spherical quantum dot with parabolic confinement, considering the effects of polarization and self energy. The variational approach within the effective mass approximation is employed here. The binding energy of an on-center impurity is computed for a $\mathrm{GaAs}$/ $\mathrm{{Al}_{x}{Ga}_{1-x}As}$ quantum dot as a function of the dot size with the dot barrier as parameter. The influence of polarization and self energy are also treated separately. Results indicate that the binding energy increases due to the presence of polarization charge, while decreases due to the self energy of the carrier. An overall enhancement in impurity binding energy, especially for small dots is noted.
△ Less
Submitted 2 May, 2018;
originally announced May 2018.
-
Quenched decay of correlations for slowly mixing systems
Authors:
Wael Bahsoun,
Christopher Bose,
Marks Ruziboev
Abstract:
We study random towers that are suitable to analyse the statistics of slowly mixing random systems. We obtain upper bounds on the rate of quenched correlation decay in a general setting. We apply our results to the random family of Liverani-Saussol-Vaienti maps with parameters in $[α_0,α_1]\subset (0,1)$ chosen independently with respect to a distribution $ν$ on $[α_0,α_1]$ and show that the quenc…
▽ More
We study random towers that are suitable to analyse the statistics of slowly mixing random systems. We obtain upper bounds on the rate of quenched correlation decay in a general setting. We apply our results to the random family of Liverani-Saussol-Vaienti maps with parameters in $[α_0,α_1]\subset (0,1)$ chosen independently with respect to a distribution $ν$ on $[α_0,α_1]$ and show that the quenched decay of correlation is governed by the fastest mixing map in the family. In particular, we prove that for every $δ>0$, for almost every $ω\in [α_0,α_1]^\mathbb Z$, the upper bound $n^{1-\frac{1}{α_0}+δ}$ holds on the rate of decay of correlation for Hölder observables on the fibre over $ω$. For three different distributions $ν$ on $[α_0,α_1]$ (discrete, uniform, quadratic), we also derive sharp asymptotics on the measure of return-time intervals for the quenched dynamics, ranging from $n^{-\frac{1}{α_0}}$ to $(\log n)^{\frac{1}{α_0}}\cdot n^{-\frac{1}{α_0}}$ to $(\log n)^{\frac{2}{α_0}}\cdot n^{-\frac{1}{α_0}}$ respectively.
△ Less
Submitted 29 January, 2018; v1 submitted 13 June, 2017;
originally announced June 2017.
-
On irreducibility of Oseledets subspaces
Authors:
Christopher Bose,
Joseph Horan,
Anthony Quas
Abstract:
For a cocycle of invertible real $n$-by-$n$ matrices, the Multiplicative Ergodic Theorem gives an Oseledets subspace decomposition of $\mathbb{R}^n$; that is, above each point in the base space, $\mathbb{R}^n$ is written as a direct sum of equivariant subspaces, one for each Lyapunov exponent of the cocycle. It is natural to ask if these summands may be further decomposed into equivariant subspace…
▽ More
For a cocycle of invertible real $n$-by-$n$ matrices, the Multiplicative Ergodic Theorem gives an Oseledets subspace decomposition of $\mathbb{R}^n$; that is, above each point in the base space, $\mathbb{R}^n$ is written as a direct sum of equivariant subspaces, one for each Lyapunov exponent of the cocycle. It is natural to ask if these summands may be further decomposed into equivariant subspaces; that is, if the Oseledets subspaces are reducible. We prove a theorem yielding sufficient conditions for irreducibility of the trivial equivariant subspaces $\mathbb{R}^2$ and $\mathbb{C}^2$ for $O_2(\mathbb{R})$-valued cocycles and give explicit examples where the conditions are satisfied.
△ Less
Submitted 9 February, 2017; v1 submitted 7 June, 2016;
originally announced June 2016.
-
Mixing rates and limit theorems for random intermittent maps
Authors:
Wael Bahsoun,
Christopher Bose
Abstract:
We study random transformations built from intermittent maps on the unit interval that share a common neutral fixed point. We focus mainly on random selections of Pomeu-Manneville-type maps $T_α$ using the full parameter range $0< α< \infty$, in general. We derive a number of results around a common theme that illustrates in detail how the constituent map that is fastest mixing (i.e.\ smallest…
▽ More
We study random transformations built from intermittent maps on the unit interval that share a common neutral fixed point. We focus mainly on random selections of Pomeu-Manneville-type maps $T_α$ using the full parameter range $0< α< \infty$, in general. We derive a number of results around a common theme that illustrates in detail how the constituent map that is fastest mixing (i.e.\ smallest $α$) combined with details of the randomizing process, determines the asymptotic properties of the random transformation. Our key result (Theorem 1.1) establishes sharp estimates on the position of return time intervals for the \emph{quenched} dynamics. The main applications of this estimate are to \textit{limit laws} (in particular, CLT and stable laws, depending on the parameters chosen in the range $0 < α< 1$) for the associated skew product; these are detailed in Theorem 3.2. Since our estimates in Theorem 1.1 also hold for $1 \leq α< \infty$ we study a piecewise affine version of our random transformations, prove existence of an infinite ($σ-$finite) invariant measure and study the corresponding correlation asymptotics. To the best of our knowledge, this latter kind of result is completely new in the setting of random transformations.
△ Less
Submitted 10 August, 2016; v1 submitted 20 March, 2015;
originally announced March 2015.
-
First hyperbolic times for intermittent maps with unbounded derivative
Authors:
Chris Bose,
Rua Murray
Abstract:
We establish some statistical properties of the hyperbolic times for a class of nonuniformly expanding dynamical systems. The maps arise as factors of area preserving maps of the unit square via a geometric Baker's map type construction, exhibit intermittent dynamics, and have unbounded derivatives. The geometric approach captures various examples from the literature over the last thirty years. Th…
▽ More
We establish some statistical properties of the hyperbolic times for a class of nonuniformly expanding dynamical systems. The maps arise as factors of area preserving maps of the unit square via a geometric Baker's map type construction, exhibit intermittent dynamics, and have unbounded derivatives. The geometric approach captures various examples from the literature over the last thirty years. The statistics of these maps are controlled by the order of tangency that a certain "cut function" makes with the boundary of the square. Using a large deviations result of Melbourne and Nicol we obtain sharp estimates on the distribution of first hyperbolic times. As shown by Alves, Viana and others, knowledge of the tail of the distribution of first hyperbolic times leads to estimates on the rate of decay of correlations and derivation of a CLT. For our family of maps, we compare the estimates on correlation decay rate and CLT derived via hyperbolic times with those derived by a direct Young tower construction. The latter estimates are known to be sharp.
△ Less
Submitted 28 July, 2013;
originally announced July 2013.
-
Decay of correlation for random intermittent maps
Authors:
Wael Bahsoun,
Christopher Bose,
Yuejiao Duan
Abstract:
We study a class of random transformations built over finitely many intermittent maps sharing a common indifferent fixed point. Using a Young-tower technique, we show that the map with the fastest relaxation rate dominates the asymptotics. In particular, we prove that the rate of correlation decay for the annealed dynamics of the random map is the same as the sharp rate of correlation decay for th…
▽ More
We study a class of random transformations built over finitely many intermittent maps sharing a common indifferent fixed point. Using a Young-tower technique, we show that the map with the fastest relaxation rate dominates the asymptotics. In particular, we prove that the rate of correlation decay for the annealed dynamics of the random map is the same as the sharp rate of correlation decay for the map with the fastest relaxation rate.
△ Less
Submitted 28 February, 2014; v1 submitted 28 May, 2013;
originally announced May 2013.
-
Rigorous pointwise approximations for invariant densities of nonuniformly expanding maps
Authors:
Wael Bahsoun,
Christopher Bose,
Yuejiao Duan
Abstract:
We use an Ulam-type discretization scheme to provide pointwise approximations for invariant densities of interval maps with a neutral fixed point. We prove that the approximate invariant density converges pointwise to the true density at a rate $C^*\cdot\frac{\ln m}{m}$, where $C^*$ is a computable fixed constant and $m^{-1}$ is the mesh size of the discretization.
We use an Ulam-type discretization scheme to provide pointwise approximations for invariant densities of interval maps with a neutral fixed point. We prove that the approximate invariant density converges pointwise to the true density at a rate $C^*\cdot\frac{\ln m}{m}$, where $C^*$ is a computable fixed constant and $m^{-1}$ is the mesh size of the discretization.
△ Less
Submitted 19 July, 2013; v1 submitted 17 January, 2013;
originally announced January 2013.
-
Numerical approximation of conditionally invariant measures via Maximum Entropy
Authors:
Christopher Bose,
Rua Murray
Abstract:
It is well known that open dynamical systems can admit an uncountable number of (absolutely continuous) conditionally invariant measures (ACCIMs) for each prescribed escape rate. We propose and illustrate a convex optimisation based selection scheme (essentially maximum entropy) for gaining numerical access to some of these measures. The work is similar to the Maximum Entropy (MAXENT) approach for…
▽ More
It is well known that open dynamical systems can admit an uncountable number of (absolutely continuous) conditionally invariant measures (ACCIMs) for each prescribed escape rate. We propose and illustrate a convex optimisation based selection scheme (essentially maximum entropy) for gaining numerical access to some of these measures. The work is similar to the Maximum Entropy (MAXENT) approach for calculating absolutely continuous invariant measures of nonsingular dynamical systems, but contains some interesting new twists, including: (i) the natural escape rate is not known in advance, which can destroy convex structure in the problem; (ii) exploitation of convex duality to solve each approximation step induces important (but dynamically relevant and not at first apparent) localisation of support; (iii) significant potential for application to the approximation of other dynamically interesting objects (for example, invariant manifolds).
△ Less
Submitted 21 February, 2013; v1 submitted 31 October, 2012;
originally announced November 2012.
-
Polynomial decay of correlations in the generalized baker's transformation
Authors:
Christopher Bose,
Rua Murray
Abstract:
We introduce a family of area preserving generalized baker's transformations acting on the unit square and having sharp polynomial rates of mixing for Holder data. The construction is geometric, relying on the graph of a single variable "cut function". Each baker's map B is non-uniformly hyperbolic and while the exact mixing rate depends on B, all polynomial rates can be attained. The analysis of…
▽ More
We introduce a family of area preserving generalized baker's transformations acting on the unit square and having sharp polynomial rates of mixing for Holder data. The construction is geometric, relying on the graph of a single variable "cut function". Each baker's map B is non-uniformly hyperbolic and while the exact mixing rate depends on B, all polynomial rates can be attained. The analysis of mixing rates depends on building a suitable Young tower for an expanding factor. The mechanisms leading to a slow rate of correlation decay are especially transparent in our examples due to the simple geometry in the construction. For this reason we propose this class of maps as an excellent testing ground for new techniques for the analysis of decay of correlations in non-uniformly hyperbolic systems. Finally, some of our examples can be seen to be extensions of certain 1-D non-uniformly expanding maps that have appeared in the literature over the last twenty years thereby providing a unified treatment of these interesting and well-studied examples.
△ Less
Submitted 11 June, 2012;
originally announced June 2012.
-
Ulam's method for Lasota-Yorke maps with holes
Authors:
Christopher Bose,
Gary Froyland,
Cecilia González-Tokman,
Rua Murray
Abstract:
Ulam's method is a rigorous numerical scheme for approximating invariant densities of dynamical systems. The phase space is partitioned into connected sets and an inter-set transition matrix is computed from the dynamics; an approximate invariant density is read off as the leading left eigenvector of this matrix. When a hole in phase space is introduced, one instead searches for \emph{conditional}…
▽ More
Ulam's method is a rigorous numerical scheme for approximating invariant densities of dynamical systems. The phase space is partitioned into connected sets and an inter-set transition matrix is computed from the dynamics; an approximate invariant density is read off as the leading left eigenvector of this matrix. When a hole in phase space is introduced, one instead searches for \emph{conditional} invariant densities and their associated escape rates. For Lasota-Yorke maps with holes we prove that a simple adaptation of the standard Ulam scheme provides convergent sequences of escape rates (from the leading eigenvalue), conditional invariant densities (from the corresponding left eigenvector), and quasi-conformal measures (from the corresponding right eigenvector). We also immediately obtain a convergent sequence for the invariant measure supported on the survivor set. Our approach allows us to consider relatively large holes. We illustrate the approach with several families of examples, including a class of Lorenz maps.
△ Less
Submitted 3 February, 2013; v1 submitted 10 April, 2012;
originally announced April 2012.
-
Invariant densities and escape rates: Rigorous and computable approximations in the $L^{\infty}$-norm
Authors:
Wael Bahsoun,
Christopher Bose
Abstract:
In this article we study a piecewise linear discretization schemes for transfer operators (Perron-Frobenius operators) associated with interval maps. We show how these can be used to provide rigorous {\bf pointwise} approximations for invariant densities of Markov interval maps. We also derive the order of convergence of the approximate invariant density to the real one in the $L^{\infty}$-norm. T…
▽ More
In this article we study a piecewise linear discretization schemes for transfer operators (Perron-Frobenius operators) associated with interval maps. We show how these can be used to provide rigorous {\bf pointwise} approximations for invariant densities of Markov interval maps. We also derive the order of convergence of the approximate invariant density to the real one in the $L^{\infty}$-norm. The outcome of this paper complements rigorous results on $L^1$ approximations of invariant densities \cite{KMY} and recent results on the formulae of escape rates of open dynamical systems \cite{KL2}. We implement our computations on two examples (one rigorous and one non-rigorous) to illustrate the feasibility and efficiency of our schemes.
△ Less
Submitted 3 August, 2010;
originally announced August 2010.
-
Quasi-Invariant measures, escape rates and the effect of the hole
Authors:
Wael Bahsoun,
Christopher Bose
Abstract:
Let $T$ be a piecewise expanding interval map and $T_H$ be an abstract perturbation of $T$ into an interval map with a hole. Given a number $\ell$, $0<\ell<1$, we compute an upper-bound on the size of a hole needed for the existence of an absolutely continuous conditionally invariant measure (accim) with escape rate not greater than $-\ln(1-\ell)$. The two main ingredients of our approach are Ul…
▽ More
Let $T$ be a piecewise expanding interval map and $T_H$ be an abstract perturbation of $T$ into an interval map with a hole. Given a number $\ell$, $0<\ell<1$, we compute an upper-bound on the size of a hole needed for the existence of an absolutely continuous conditionally invariant measure (accim) with escape rate not greater than $-\ln(1-\ell)$. The two main ingredients of our approach are Ulam's method and an abstract perturbation result of Keller and Liverani.
△ Less
Submitted 30 November, 2009; v1 submitted 29 June, 2009;
originally announced June 2009.
