Interactive, High-Dimensional Flow Visualization Using Image Space-Based Approaches

目前，我们作为一个文明，正在经历一个信息大爆炸，即，数据爆炸根据Peter Lyman和Hal R.瓦里安在加州伯克利大学（多少信息？）2003年至2005年期间产生的新数据比以前产生的所有文件都要多，其中90%以上是数字形式。数字数据的数量远远超过了纸或电影。原始数据本身不一定很有价值。解释和提取重要数据的意义是一种能力。有效地管理、理解和利用当前产生的大量数据，即使不是21世纪世纪最大的挑战，也是其中之一。可视化是在我们不断被轰炸的海量数据中导航的关键。可视化需要兆字节、千兆字节或太字节的数据，并使用计算机器生成有意义的图像。我们可以用图片来总结数据中的潜在模式和重要现象。图片利用了非常强大的人类视觉系统，它提供了一个非常高的带宽连接到大脑。这个建议是关于推进一个特定的子领域的可视化，即流动可视化。流场可视化涉及的是矢量场数据，矢量场数据是一类既有方向又有幅值的特殊数据。此外，我们关注的不是任何类型的矢量场数据，而是计算流体动力学（CFD）模拟数据。计算流体力学是一门工程学科，其目标是以定量的方式预测流体的行为。例如，模拟可以用来研究洋流和天气模式的行为。使用CFD分析的工程师人数正在迅速增长。越来越多的工程师依靠计算机辅助建模、仿真和可视化工具来为众多设计和制造挑战找到最佳解决方案。计算机辅助工程的动机是强大的能力，建立原型，并提出优化使用一个基于软件的过程，这是比建设真实的，重型组件更快。此外，科学家和研究人员越来越依赖于仿真软件来研究自然现象，如洪水、火山和化学反应等，这些在真实的世界中根本不可能重现。尽管计算机辅助工程和仿真提供了许多优点，如更快速的原型设计、更快速的反馈循环和降低制造成本，和研究自然现象的能力，否则是不切实际的，这些软件工具带来了一些挑战。计算流体动力学（CFD）模拟产生非常大的复杂数据集，例如，以千兆字节或太字节为单位。分析、探索和呈现仿真结果所需的软件工具必须面对几个密切相关的挑战。这项研究提案将通过开发新的计算机算法、新的交互技术以及新的可视化来解决这些挑战，从而为CFD模拟数据提供有价值的见解。

Currently, we as a civilization, are experiencing an information big bang ,i.e., an explosion of data. According to a study conducted by Peter Lymanand Hal R. Varian at the University of California Berkeley ( How muchinformation ?) new data generated in the period from 2003 to 2005 is greaterthan that of all previously created documents, more than 90% of which isin digital form. The amount of digital data vastly exceeds that produced inpaper or film. Raw data itself is not necessarily very valuable. It is ourability to interpret and extract meaning from the data that is important.One of, if not the greatest, challenge of the 21st century is to effectivelymanage, understand, and make use of the enormous amounts of data currentlybeing generated. Visualization is the key to navigating through the vastamounts of data we are constantly bombarded with. Visualization takesmegabytes, gigabytes, or terabytes of data and generates meaningful imagesusing computing machinery. We can summarize underlying patterns and importantphenomenon in the data using pictures. Pictures harness the very powerfulhuman visual system which provides a very high bandwidth connection to thebrain.This proposal is concerned with advancing a specific sub-field ofvisualization, namely, flow visualization. Flow visualization is concernedwith depicted vector field data, a special category of data with bothmagnitude and direction. Moreover, we are concerned with the depiction ofnot just any type of vector field data, but rather, with Computational FluidDynamics (CFD) simulation data. CFD is an engineering discipline whose goalis to predict the behavior of flow in a quantitative fashion. For example,simulation can be used to study the behavior of ocean currents and weatherpatterns.The number of engineers who use CFD analysis is growing rapidly. More andmore engineers rely on computer aided modeling, simulation, and visualizationtools in order to find the best solutions for a multitude of design andmanufacturing challenges. Computer aided engineering is motivated stronglyby the ability to build prototypes and propose optimizations using asoftware-based process that is faster than building real, heavyweightcomponents. Furthermore, scientists and researchers rely ever more onsimulation software in order to study natural phenomena such as flooding,volcanoes, and chemical reactions etc. that are simply not feasible tore-create in the real world.Despite the advantages that computer aided engineering and simulationoffers, such as more rapid prototyping, a more rapid feedback loop, and areduction in manufacturing costs, and the ability to study natural phenomena that would otherwise be impractical, these software tools bring several challenges with them. Computational Fluid Dynamics (CFD) simulations result in very large, complex data sets, e.g., measured in gigabytes or terabytes. The software tools needed to analyze, explore, and present simulation results must face several closely related challenges. This research proposal will address these challenges in a novel way by developing new computer algorithms, new interaction techniques, and hence new visualizations that will provide valuable insight into CFDsimulation data.