College of Computer, Mathematical & Natural Sciences
http://hdl.handle.net/1903/12
2018-09-18T21:55:15ZPIC simulation data for "Effect of the reconnection electric field on electron distribution functions in the diffusion region of magnetotail reconnection"
http://hdl.handle.net/1903/21408
PIC simulation data for "Effect of the reconnection electric field on electron distribution functions in the diffusion region of magnetotail reconnection"
Bessho, Naoki; Chen, Li-Jen; Wang, Shan; Hesse, Michael
2-dimensional particle-in-cell simulation data for a paper, "Effect of the reconnection electric field on electron distribution functions in the diffusion region of magnetotail reconnection".
Normalization: position x and z by d_i,
B field by B_0,
E field by B_0v_A/c.
In file names for fields, ?? represents a kind of field, bx, ey, ez, or ay. XX represents time, 21 or 25, indicating t=21.1 or 25.3, respectively.
Field profile in z (??avXX.txt): field profile, averaged in the x direction (between -5+i and 5+i, 11 points, where i is the x grid of the X-line),
1st column: z position,
2nd column: field.
2-D field contour (??XX.txt):
1st column: z position,
2nd column: x position,
3rd column: field.
In file names for reduced electron VDFs: ffXX-yx-zYY.txt, where XX represents time, 21 or 25 (t=21.1 or 25.3), and YY represents the distance from z=0 (for example, z0.3 means the VDF at z=0.3. z0 represents the VDF at z=0 ).
VDF in vy-vx plane (ffXX-yx-zYY.txt):
1st column: vy,
2nd column: vx,
3rd column: VDF.
VDF in vy-vz plane (ffXX-yz-zYY.txt):
1st column: vy,
2nd column: vz,
3rd column: VDF.
2018-01-01T00:00:00ZMotivic Cohomology of Groups of Order p^3
http://hdl.handle.net/1903/21406
Motivic Cohomology of Groups of Order p^3
Black, Rebecca
In this thesis we compute the motivic cohomology ring (also known as Bloch's higher Chow groups) with finite coefficients for the two nonabelian groups of order $27$, thought of as affine algebraic groups over $\mathbb{C}$. Specifically, letting $\tau$ denote a generator of the motivic cohomology group $H^{0,1}(BG,\Z/3) \cong \Z/3$ where $G$ is one of these groups, we show that the motivic cohomology ring contains no $\tau$-torsion, and so can be computed as a weight filtration on the ordinary group cohomology. In the case of a prime $p > 3$, there are also two nonabelian groups of order $p^3$. We make progress toward computing the motivic cohomology in the general case as well by reducing the question to understanding the $\tau$-torsion on the motivic cohomology of a $p$-dimensional variety; we also compute the motivic cohomology of $BG$ for general $p$ modulo the $\tau$-torsion classes.
2018-01-01T00:00:00ZTemporal Context Modeling for Text Streams
http://hdl.handle.net/1903/21404
Temporal Context Modeling for Text Streams
Rao, Jinfeng
There is increasing recognition that time plays an essential role in many information seeking tasks. This dissertation explores temporal models on evolving streams of text and the role that such models play in improving information access. I consider two cases: a stream of social media posts by many users for tweet search and a stream of queries by an individual user for voice search. My work explores the relationship between temporal models and context models: for tweet search, the evolution of an event serves as the context of clustering relevant tweets; for voice search, the user's history of queries provides the context for helping understand her true information need.
First, I tackle the tweet search problem by modeling the temporal contexts of the underlying collection. The intuition is that an information need in Twitter usually correlates with a breaking news event, thus tweets posted during that event are more likely to be relevant. I explore techniques to model two different types of temporal signals: pseudo trend and query trend. The pseudo trend is estimated through the distribution of timestamps from an initial list of retrieved documents given a query, which I model through continuous hidden Markov approach as well as neural network-based methods for relevance ranking and sequence modeling. As an alternative, the query trend, is directly estimated from the temporal statistics of query terms, obviating the need for an initial retrieval. I propose two different approaches to exploit query trends: a linear feature-based ranking model and a regression-based model that recover the distribution of relevant documents directly from query trends. Extensive experiments on standard Twitter collections demonstrate the superior effectivenesses of my proposed techniques.
Second, I introduce the novel problem of voice search on an entertainment platform, where users interact with a voice-enabled remote controller through voice requests to search for TV programs. Such queries range from specific program navigation (i.e., watch a movie) to requests with vague intents and even queries that have nothing to do with watching TV. I present successively richer neural network architectures to tackle this challenge based on two key insights: The first is that session context can be exploited to disambiguate queries and recover from ASR errors, which I operationalize with hierarchical recurrent neural networks. The second insight is that query understanding requires evidence integration across multiple related tasks, which I identify as program prediction, intent classification, and query tagging. I present a novel multi-task neural architecture that jointly learns to accomplish all three tasks. The first model, already deployed in production, serves millions of queries daily with an improved customer experience. The multi-task learning model is evaluated on carefully-controlled laboratory experiments, which demonstrates further gains in effectiveness and increased system capabilities. This work now serves as the core technology in Comcast Xfinity X1 entertainment platform, which won an Emmy award in 2017 for the technical contribution in advancing television technologies.
This dissertation presents families of techniques for modeling temporal information as contexts to assist applications with streaming inputs, such as tweet search and voice search. My models not only establish the state-of-the-art effectivenesses on many related tasks, but also reveal insights of how various temporal patterns could impact real information-seeking processes.
2018-01-01T00:00:00ZEfficient Data-Oblivious Computation
http://hdl.handle.net/1903/21403
Efficient Data-Oblivious Computation
Nayak, Kartik Ravidas
The rapid increase in the amount of data stored by cloud servers has resulted in growing privacy concerns for users. First, although keeping data encrypted at all times is an attractive approach to privacy, encryption may preclude mining and learning useful patterns from data. Second, companies are unable to distribute proprietary programs to other parties without risking the loss of their private code when those programs are reverse engineered. A challenge underlying both those problems is that how data is accessed — even when that data is encrypted — can leak secret information.
Oblivious RAM is a well studied cryptographic primitive that can be used to solve the underlying challenge of hiding data-access patterns. In this dissertation, we improve Oblivious RAMs and oblivious algorithms asymptotically. We then show how to apply our novel oblivious algorithms to build systems that enable privacy-preserving computation on encrypted data and program obfuscation.
Specifically, the first part of this dissertation shows two efficient Oblivious RAM algorithms: 1) The first algorithm achieves sub-logarithmic bandwidth blowup while only incurring an inexpensive XOR computation for performing Private Information Retrieval operations, and 2) The second algorithm is the first perfectly-secure Oblivious Parallel RAM with $O(\log^3 N )$ bandwidth blowup, $O((\log m + \log \log N)\log N)$ depth blowup, and $O(1)$ space blowup when the PRAM has $m$ CPUs and stores $N$ blocks of data. The second part of this dissertation describes two systems — HOP and GraphSC — that address the problem of computing on private data and the distribution of proprietary programs. HOP is a system that achieves simulation-secure obfuscation of RAM programs assuming secure hardware. It is the first prototype implementation of a provably secure virtual black-box (VBB) obfuscation scheme in any model under any assumptions. GraphSC is a system that allows cloud servers to run a class of data-mining and machine-learning algorithms over users’ data without learning anything about that data. GraphSC brings efficient, parallel secure computation to programmers by allowing them to express computation tasks using the GraphLab abstraction. It is backed by the first non-trivial parallel oblivious algorithms that outperform generic Oblivious RAMs.
2018-01-01T00:00:00Z