28 research outputs found
Identifying people based on pressure board measurements
No full textIdentifying people is important for various applications and context. In this thesis, we examine the potential to identify people based on how they step. Towards this end, we designed and implemented a system utilizing the Wii Balance Board. When a user steps on the board, we can obtain weight distribution and center of pressure. These can be used to train our model and based on it, we can identify a user. We tested our system using dataset obtained from 19 volunteers. The presented approach has accuracy of 97%. This indicates the approach could have potential and warrants further work.M.S.Includes bibliographical referencesby Samarth Lakhatariy
Discovering the hidden users of Scratch
No full textThesis: M. Eng., Massachusetts Institute of Technology, Department of Electrical Engineering and Computer Science, 2018.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (pages 52-54).Usage statistics ("telemetry data") have become an essential tool for understanding how complex systems are used and how to improve them. However, many of these systems are deployed in areas with limited internet connectivity which hampers the ability to collect telemetry data. In this thesis, we describe a telemetry data collection system built for the Scratch programming language to collect usage data regarding how Scratch is being used in areas with poor internet connections. We develop the system to allow users to opt-in to sharing their usage and project data with the Scratch research team at the MIT Media Lab. The data is stored locally on the user's machine until it is ready to be transmitted. Once network conditions are appropriate, the packets are transmitted to a server which verifies the contents of the packet and stores it in a data storage cluster. We aggregate the data and build a visualization dashboard to examine usage patterns, geolocation statistics, and project content for Scratch users all around the world.by Samarth Mohan.M. Eng
Management of a high mix production system with interdependent demands : simulation of proposed policies
Full text linkThesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Mechanical Engineering, 2009.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (p. 92-94).The finished goods inventory management in the accessories area of a material testing company is complex. There is interdependence between the demands of products and they can be sold both as part of systems and as individual after sales items. Besides, there is uncertainty in determining replenishment lead times. An optimization problem is formulated considering customer satisfaction, inventory holding costs and correlation between demands. To ascertain its validity, a discrete event simulation is executed over historical demand. Simulation also helps to check the solution robustness by executing the proposed inventory levels over statistically generated demands. The result provides the right mix of finished products which should be stored on the shelves. 90% reduction in lost sales and 35% in inventory value on hand have been projected. The results have been further implemented at the part level inventory.by Samarth Chugh.M.Eng
Physics-Informed Deep Learning for Computational Fluid Flow Analysis: Coupling of physics-informed neural networks and autoencoders for aerodynamic flow predictions on variable geometries
No full textThe main objective of this thesis was to explore the capabilities of neural networks in terms of representing governing differential equations, primarily in the purview of fluid/aero dynamic flows. The governing differential equations were accommodated within the loss functions for training the neural networks, thereby making them 'physics-informed'. Subsequently, this idea of physics-informed neural networks (PINNs) was extended to parameterized geometries generated with the help of commercial auto-encoders developed by the UK based company Monolith AI pvt. ltd. because neural networks have the capability to learn the desired PDEs over variable/parameterized geometries without the need to recompute the solution for every minor change in the input geometry, which proves out to be a huge advantage over classical numerical techniques. The advantages, limitations and scope for further research in the field of physics-informed deep learning have been discussed in the contents of the underlying thesis report
Long-range singlet proximity effect in ferromagnetic nanowires
Full text linkRecently a long-ranged superconductor/ferromagnet (S/F) proximity effect has been reported in Co crystalline nanowires [J. Wang, M. Singh, M. Tian, N. Kumar, B. Liu, C. Shi, J. K. Jain, N. Samarth, T. E. Mallouk, and M. H. W. Chan, Nat. Phys. 6, 389 (2010)]. Since the authors take care to avoid the existence of magnetic domains, the triplet character of the long-ranged proximity effect is improbable. Here we demonstrate that in the one-dimensional ballistic regime the standard singlet S/F proximity effect becomes long ranged. We provide an exact solution for the decay of the superconducting correlations near critical temperature (Tc) and for arbitrary impurities concentration. In particular, we find a specific regime, between the diffusive and ballistic ones, where the decay length is simply the electronic mean-free path. Finally possible experiments which could permit to elucidate the nature of the observed long-ranged proximity effect in Co nanowires are discussed.Kavli Institute of NanoscienceApplied Science
Bilayer crystals of trapped ions for quantum information processing
Full text linkTrapped ion systems are a leading platform for quantum information processing, but they are currently limited to 1D and 2D arrays, which imposes restrictions on both their scalability and their range of applications. Here, we propose a path to overcome this limitation by demonstrating that Penning traps can be used to realize remarkably clean bilayer crystals, wherein hundreds of ions self-organize into two well-defined layers. These bilayer crystals are made possible by the inclusion of an anharmonic trapping potential, which is readily implementable with current technology. We study the normal modes of this system and discover salient differences compared to the modes of single-plane crystals. The bilayer geometry and the unique properties of the normal modes open new opportunities, in particular in quantum sensing and quantum simulation, that are not straightforward in single-plane crystals. Furthermore, we illustrate that it may be possible to extend the ideas presented here to realize multilayer crystals with more than two layers. Our work increases the dimensionality of trapped ion systems by efficiently utilizing all three spatial dimensions and lays the foundation for a new generation of quantum information processing experiments with multilayer 3D crystals of trapped ions.21+10 pages, 12 figures, 3 supplementary videos; Updated to version submitted to journa
INFLUENCING THE PUBLIC AND EFFICIENCY IN BUREAUCRATIC PROVISION
No full textIn the present paper a public bureau can extract additional budgetary allocations, not only by misreporting its production cost to its oversight committee but also by 'influencing' the perceptions of the public at large. I juxtapose the bureau's ability to influence the public with its ability to misreport to the oversight committee, and find that influencing the public might support an efficient level of production by invoking a separating equilibrium. However, a pooling equilibrium involving either overproduction or underproduction of output cannot be ruled out. Accordingly, overproduction could occur even when the bureau cares only about extracting excess budget and not increasing output per se. Copyright 2009 The Author. Journal compilation 2009 Blackwell Publishing Asia Pty Ltd
On Theory and Experimentation with the Probabilistic Nature of Quantum Being
No full textOver the course of this thesis, the reader is brought up to speed with the current understanding of quantum physics, as well as shown some theoretical and experimental developments done as part of research work. Chapter 1 explains the fundamental concepts of quantum physics, both mathematically and physically, as well as the historical developments that led to it and those that follow. Chapter 2 provides insight into original research done, building on previous work, in order to describe the complex quantum system of a chemical reaction under a magnetic field and to mathematically solve it in order to produce accurate results that can be confirmed by the proposed experiments. Chapter 3 goes over an experiment conducted by the author in order to ascertain the quantum properties of entanglement as well as the steps needed to reproduce the experiment. A quantum eraser was successfully built and run to test the properties of entanglement and superposition. A delayed-choice version of the experiment is proposed but could not be successfully run. An appendix follows providing the author’s thoughts on certain aspects of the research industry in its current form
Impedance-engineered Josephson parametric amplifier with single-step lithography
No full textWe present an experimental demonstration of an impedance-engineered Josephson parametric amplifier (IEJPA) fabricated in a single-step lithography process. Impedance-engineering is implemented using a lumped-element series LC circuit. We use a simpler lithography process where the entire device—impedance transformer and Josephson parametric amplifier (JPA)—is patterned in a single electron beam lithography step, followed by a double-angle Dolan-bridge technique for Al–AlOx–Al deposition. We observe amplification with 18 dB gain over a wide 400 MHz bandwidth centered around 5.3 GHz with added noise approaching the quantum limit, and a saturation power of −114 dBm. To accurately explain our experimental results, we extend existing theories for IEJPAs to incorporate the full sine nonlinearity of both the JPA and the transformer. Our work provides a route to simpler realization of broadband JPAs and a theoretical foundation for a regime of JPA operation that has been less explored in literature
