What is CDE?
Computational Discovery and Engineering (CDE) comprises the development and innovative use of mathematical algorithms and models on high performance computers (HPC) to support basic scientific research, data interpretation, product development, and forecasting. CDE is an enabling discipline with computations widely accepted as the third mode of scientific discovery on par with theory and physical experimentation.
Below are listed a few areas engineering and computer science related fields where computational engineering has played, and is expected to continue to play, a pivotal role and for which graduates of our program will be well-equipped to work in. The list is a limited sample of the many areas of computational engineering in which the faculty associated with the proposed program are active. (Click on each field for more information.)
Aerospace Engineering
Computation plays a significant role in the three main areas of aerospace: gas dynamics, structures, and controls. Most notably, perhaps, is the area of Computational Fluid Dynamics (CFD) that has been developed extensively in the aerospace community. CFD has evolved to the point where unsteady, three-dimensional simulations of air flow around geometrically complex vehicles including multi-physics phenomena (e.g. turbulence, chemistry, surface ablation, radiation) can be performed. CFD analyses are particularly critical to some aerospace systems (such as very high speed vehicles) where laboratory and flight experiments are both technically challenging and extremely expensive.
Material Physics and Biophysics
The field of materials physics encompasses the mechanical, chemical, thermal and (to a lesser extent) the electrical behavior of materials (metals, polymers, various composites and biological materials) at length scales from the sub-atomic to the macroscopic level, and at time scales from femtoseconds to years. Numerical methods have been developed, and high performance computing now allows for predicting material properties and physical mechanisms at all spatial and temporal scales. One yet unsolved example is the computation of the detailed microstructure developing from chemo-mechanically coupled phase transformations in battery materials. In biophysics, a holy grail problem is the computation of the chemical and mechanical phenomena underlying cellular processes such as division, migration and endocytosis. Both these examples easily generate nonlinear systems of equations with billions of degrees of freedom
Biomedical Engineering
The field of Biomedical Engineering is broad in scope and diverse in its applications, and the use of advanced computational methods has become equally wide-ranging. Specific examples would include image processing, visualization, fluid mechanics and fluid-structure interactions, protein engineering and drug design, biomaterials modeling and design, and biomechanics. As with examples from other areas, computation allows us to connect the observables we measure with other system or material properties. Here we realize other advantages since biological data is inherently stochastic and often difficult to obtain non-invasively, computation provides a mechanism for understanding these systems at a basic level and thereby helping with design, analysis, diagnosis and treatment.
Materials Science
Computations have provided tremendous insights into material behavior through continuum level simulations, atomistic and molecular dynamics computations, and quantum-mechanical calculations using electronic structure calculations. While experiments can measure many observable properties, it is computations which fill the gap of determining the governing mechanisms that result in the observed properties. Predictive calculations can be used to design new materials with desirable properties. An illustration of the significance of computations in materials science and computational chemistry is the Nobel prize in Chemistry in 1999 which was awarded to Walter Kohn “for his development of the density-functional theory” and to John A. Pople “for his development of computational methods in quantum chemistry.”
Data Mining
In the last few years the growth of inexpensive storage, inexpensive sensors, and online human activity have enabled computer systems to collect astonishing volumes of data about real-world phenomena. This data has been used to obtain a number of remarkable results in practical computer systems (such as Google web search and IBM’s Watson), but can also be used in scientific applications. For example, a traditional CO2 modeling task might be primarily one of atmospheric simulation; but when combined with various forms of online data, we might combine the simulation with information about carbon-generating human activities reported via social media.
Parallel Computing
Essentially all computers involve parallelism of some form. Supercomputers can have millions of cores, laptops can have 100 cores in their graphics processing unit (GPU), and distributed sensor networks can involve thousands of sensors. To solve large problems, scientists and engineers must be able to exploit this parallelism. In some cases, the only way to perform a large complex simulation is via supercomputers running for long periods of time; in other cases parallel computers are needed to provide real-time solutions to problems such as weather prediction; and in yet other cases distributing computation along with sensing can be the only way to reduce data to a manageable size.
Nuclear Engineering
The field of Nuclear Engineering has realized considerable benefits from the very beginnings of high performance parallel computing. By their very nature, Nuclear Engineering applications involve complex, multiphysics and multiscale simulations. For example, one of the more computationally intensive problems in engineering during the past past 50 years has been the solution of the coupled temperature/fluid and neutron/photon/nuclide fields in a nuclear reactor during anticipated reactor accident conditions. During the last several decades, the fidelity of the reactor simulations has generally been increased to match the computational resources available on the most advanced HPC platforms available. And because of the continued importance of nuclear power to the U.S. energy future, the DOE has continued to invest considerable resources to investigate further improvements in the application of HPC to the reactor problem, as well as other nuclear engineering applications.
Electronic and Communication Engineering
Wireless communication systems and sensor networks are ubiquitous and their growth rate exponential. Modeling however is a ondition sine qua non to continue the quest for continued system miniaturization and integration while ensuring their proper functioning under tight power constraints. The modeling task extends from the analysis of new electronic components that operate at the nano-scale to their incorporation into ultra large-scale circuits containing billions of components, and subsequent integration into handheld devices and other supporting platforms. The simulation challenge is inherently multiscale in nature, spans several physical domains, and requires state-of-the-art tools that rapidly execute on readily available computers to allow system designers to convert design concepts into manufacturable systems in a single iteration.
Structural Engineering
Experimental research in structural engineering is expensive and dangerous because it entails loading large structural components and subassemblages to failure. Moreover, many loading configurations are just not feasible; just imagine how difficult it would be to blast a tall building to investigate its collapse resistance or shake it to determine its seismic resistance. In situations like this, computational structural simulation becomes indispensable, allowing researchers to explore situations that are not experimentally tractable. Simulations models may contain millions of elements, each representing key pieces of the structure in question, allowing researchers unprecedented insight into how resistance mechanisms originate and are mobilized.
Climate and Space Weather Modeling
The fields of climate and space weather science are interdisciplinary endeavors that combine geophysical fluid dynamics and physics, applied mathematics, scientific computing as well as big observational data sets. Climate and space physics models need to cover a wide range of spatial and temporal scales. Modern multi-scale numerical algorithms and high-performance computing techniques have therefore become the key to tackling the scientific challenges and broadening our understanding of climate change and space weather processes. The broad range of research topics includes global and regional climate modeling and change, the development of adaptive and variable-resolution computational grids for both climate and space physics models, the integration of satellite data into models, data assimilation, the representation of biosphere-atmosphere, ice-atmosphere and chemistry-climate interactions, as well as software and data management issues for atmospheric and space science models.
Industrial and Operations Engineering (IOE)
Research within IOE ranges from the highly theoretical development and analysis of new algorithms to the application of these algorithms to fields spanning healthcare, energy, telecommunications, transportation, manufacturing, logistics and distribution, and more. The ability to model a complex system, to develop the tools necessary to analyze this system, to design optimization techniques for improving this system, and to translate the results into practice all often require advanced computational efforts. Areas such as discrete event simulation, non-linear optimization, large-scale mixed inter programming, and stochastic optimization all yield challenging computational problems that require a blend of IOE-specific expertise and more general CDE knowledge.