arXiv subjects
Publications and source records attributed to Yaqi Zhang.
We construct all six-particle helicity amplitude integrands in color-dressed, maximally supersymmetric Yang-Mills theory at two loops in a single prescriptive basis of master integrands.
Many data insight questions can be viewed as searching in a large space of tables and finding important ones, where the notion of importance is defined in some adhoc user defined manner. This paper presents Holistic Cube Analysis (HoCA), a framework that augments the capabilities of relational queries for such problems. HoCA first augments the relational data model and introduces a new data type AbstractCube, defined as a function which maps a region-features pair to a relational table (a region is a tuple which specifies values of a set of dimensions). AbstractCube provides a logical form of data, and HoCA operators are cube-to-cube transformations. We describe two basic but fundamental HoCA operators, cube crawling and cube join (with many possible extensions). Cube crawling explores a region space, and outputs a cube that maps regions to signal vectors. Cube join, in turn, is critical for composition, allowing one to join information from different cubes for deeper analysis. Cube crawling introduces two novel programming features, (programmable) Region Analysis Models (RAMs) and Multi-Model Crawling. Crucially, RAM has a notion of population features, which allows one to go beyond only analyzing local features at a region, and program region-population analysis that compares region and population features, capturing a large class of importance notions. HoCA has a rich algorithmic space, such as optimizing crawling and join performance, and physical design of cubes. We have implemented and deployed HoCA at Google. Our early HoCA offering has attracted more than 30 teams building applications with it, across a diverse spectrum of fields including system monitoring, experimentation analysis, and business intelligence. For many applications, HoCA empowers novel and powerful analyses, such as instances of recurrent crawling, which are challenging to achieve otherwise.
Transformer is popular in recent 3D human pose estimation, which utilizes long-term modeling to lift 2D keypoints into the 3D space. However, current transformer-based methods do not fully exploit the prior knowledge of the human skeleton provided by the kinematic structure. In this paper, we propose a novel transformer-based model EvoPose to introduce the human body prior knowledge for 3D human pose estimation effectively. Specifically, a Structural Priors Representation (SPR) module represents human priors as structural features carrying rich body patterns, e.g. joint relationships. The structural features are interacted with 2D pose sequences and help the model to achieve more informative spatiotemporal features. Moreover, a Recursive Refinement (RR) module is applied to refine the 3D pose outputs by utilizing estimated results and further injects human priors simultaneously. Extensive experiments demonstrate the effectiveness of EvoPose which achieves a new state of the art on two most popular benchmarks, Human3.6M and MPI-INF-3DHP.
In this paper, we consider a least-squares (LS)-based distributed algorithm build on a sensor network to estimate an unknown parameter vector of a dynamical system, where each sensor in the network has partial information only but is allowed to communicate with its neighbors. Our main task is to generalize the well-known theoretical results on the traditional LS to the current distributed case by establishing both the upper bound of the accumulated regrets of the adaptive predictor and the convergence of the distributed LS estimator, with the following key features compared with the existing literature on distributed estimation: Firstly, our theory does not need the previously imposed independence, stationarity or Gaussian property on the system signals, and hence is applicable to stochastic systems with feedback. Secondly, the cooperative excitation condition introduced and used in this paper for the convergence of the distributed LS estimate is the weakest possible one, which shows that even if any individual sensor cannot estimate the unknown parameter by the traditional LS, the whole network can still fulfill the estimation task by the distributed LS.
We outline the concrete steps involved in building prescriptive master integrand bases for scattering amplitudes beyond the planar limit. We highlight the role of contour choices in such bases, and illustrate the full process by constructing a complete, triangle power-counting basis at two loops for six particles. We show how collinear contour choices can be used to divide integrand bases into separately finite and divergent subspaces, and how double-poles can be used to further subdivide these spaces according to (transcendental) weight. Complete details of the basis constructed for six particles is provided in the ancillary files for this work's submission to the arXiv.
The tight upper bound $\operatorname{pt}_+(G) \leq \left\lceil \frac{\left\vert \operatorname{V}(G) \right\vert - \operatorname{Z}_+(G)}{2} \right\rceil$ is established for the positive semidefinite propagation time of a graph in terms of its positive semidefinite zero forcing number. To prove this bound, two methods of transforming one positive semidefinite zero forcing set into another and algorithms implementing these methods are presented. Consequences of the bound, including a tight Nordhaus-Gaddum sum upper bound on positive semidefinite propagation time, are established.
This paper proposes Capstan: a scalable, parallel-patterns-based, reconfigurable dataflow accelerator (RDA) for sparse and dense tensor applications. Instead of designing for one application, we start with common sparse data formats, each of which supports multiple applications. Using a declarative programming model, Capstan supports application-independent sparse iteration and memory primitives that can be mapped to vectorized, high-performance hardware. We optimize random-access sparse memories with configurable out-of-order execution to increase SRAM random-access throughput from 32% to 80%. For a variety of sparse applications, Capstan with DDR4 memory is 18x faster than a multi-core CPU baseline, while Capstan with HBM2 memory is 16x faster than an Nvidia V100 GPU. For sparse applications that can be mapped to Plasticine, a recent dense RDA, Capstan is 7.6x to 365x faster and only 16% larger.
Recurrent Neural Network (RNN) applications form a major class of AI-powered, low-latency data center workloads. Most execution models for RNN acceleration break computation graphs into BLAS kernels, which lead to significant inter-kernel data movement and resource underutilization. We show that by supporting more general loop constructs that capture design parameters in accelerators, it is possible to improve resource utilization using cross-kernel optimization without sacrificing programmability. Such abstraction level enables a design space search that can lead to efficient usage of on-chip resources on a spatial architecture across a range of problem sizes. We evaluate our optimization strategy on such abstraction with DeepBench using a configurable spatial accelerator. We demonstrate that this implementation provides a geometric speedup of 30x in performance, 1.6x in area, and 2x in power efficiency compared to a Tesla V100 GPU, and a geometric speedup of 2x compared to Microsoft Brainwave implementation on a Stratix 10 FPGA.
We propose the algorithms for performing multiway joins using a new type of coarse grain reconfigurable hardware accelerator~-- ``Plasticine''~-- that, compared with other accelerators, emphasizes high compute capability and high on-chip communication bandwidth. Joining three or more relations in a single step, i.e. multiway join, is efficient when the join of any two relations yields too large an intermediate relation. We show at least 200X speedup for a sequence of binary hash joins execution on Plasticine over CPU. We further show that in some realistic cases, a Plasticine-like accelerator can make 3-way joins more efficient than a cascade of binary hash joins on the same hardware, by a factor of up to 45X.
In this paper, we use reduced precision checking (RPC) to detect errors in floating point arithmetic. Prior work explored RPC for addition and multiplication. In this work, we extend RPC to a complete floating point unit (FPU), including division and square root, and we present precise analyses of the errors undetectable with RPC that show bounds that are smaller than prior work. We implement RPC for a complete FPU in RTL and experimentally evaluate its error coverage and cost.