Session1 10:35 - 12:15 on 1st June 2015

6th Workshop on Computational Optimization, Modelling & Simulation (COMS) Session 1

Chair: Leifur Leifsson

 261 Surrogate-Based Airfoil Design with Space Mapping and Adjoint Sensitivity [abstract]Abstract: This paper presents a space mapping algorithm for airfoil shape optimization enhanced with adjoint sensitivities. The surrogate-based algorithm utilizes low-cost derivative information obtained through adjoint sensitivities to improve the space mapping matching between a high-fidelity airfoil model, evaluated through expensive CFD simulations, and its fast surrogate. Here, the airfoil surrogate model is constructed though low-fidelity CFD simulations. As a result, the design process can be performed at a low computational cost in terms of the number of high-fidelity CFD simulations. The adjoint sensitivities are also exploited to speed up the surrogate optimization process. Our method is applied to a constrained drag minimization problem in two-dimensional inviscid transonic flow. The problem is solved for several low-fidelity model termination criteria. The results show that when compared with direct gradient-based optimization with adjoint sensitivities, the proposed approach requires 49-78% less computational cost while still obtaining a comparable airfoil design. Yonatan Tesfahunegn, Slawomir Koziel, Leifur Leifsson, Adrian Bekasiewicz 317 How to Speed up Optimization? Opposite-Center Learning and Its Application to Differential Evolution [abstract]Abstract: This paper introduces a new sampling technique called Opposite-Center Learning (OCL) intended for convergence speedup of meta-heuristic optimization algorithms. It comprises an extension of Opposition-Based Learning (OBL), a simple scheme that manages to boost numerous optimization methods by considering the opposite points of candidate solutions. In contrast to OBL, OCL has a theoretical foundation – the opposite center point is defined as the optimal choice in pair-wise sampling of the search space given a random starting point. A concise analytical background is provided. Computationally the opposite center point is approximated by a lightweight Monte Carlo scheme for arbitrary dimension. Empirical results up to dimension 20 confirm that OCL outperforms OBL and random sampling: the points generated by OCL have shorter expected distances to a uniformly distributed global optimum. To further test its practical performance, OCL is applied to differential evolution (DE). This novel scheme for continuous optimization named Opposite-Center DE (OCDE) employs OCL for population initialization and generation jumping. Numerical experiments on a set of benchmark functions for dimensions 10 and 30 reveal that OCDE on average improves the convergence rates by 38% and 27% compared to the original DE and the Opposition-based DE (ODE), respectively, while remaining fully robust. Most promising are the observations that the accelerations shown by OCDE and OCL increase with problem dimensionality. H. Xu, C.D. Erdbrink, V.V. Krzhizhanovskaya 281 Visualizing and Improving the Robustness of Phase Retrieval Algorithms [abstract]Abstract: Coherent x-ray diffractive imaging is a novel imaging technique that utilizes phase retrieval and nonlinear optimization methods to image matter at nanometer scales. We explore how the convergence properties of a popular phase retrieval algorithm, Fienup’s HIO, behave by introducing a reduced dimensionality problem allowing us to visualize convergence to local minima and the globally optimal solution. We then introduce generalizations of HIO that improve upon the original algorithm’s ability to converge to the globally optimal solution. Ashish Tripathi, Sven Leyffer, Todd Munson, Stefan Wild 257 Fast Optimization of Integrated Photonic Components Using Response Correction and Local Approximation Surrogates [abstract]Abstract: A methodology for a rapid design optimization of integrated photonic couplers is presented. The proposed technique exploits variable-fidelity electromagnetic (EM) simulation models, additive response correction for accommodating the discrepancies between the EM models of various fidelities, and local response surface approximations for a fine tuning of the final design. A specific example of a 1,555 nm coupler is considered with an optimum design obtained at a computational cost corresponding to about 24 high-fidelity EM simulations of the structure. Adrian Bekasiewicz, Slawomir Koziel, Leifur Leifsson 197 Model Selection for Discriminative Restricted Boltzmann Machines Through Meta-heuristic Techniques [abstract]Abstract: Discriminative learning of Restricted Boltzmann Machines has been recently introduced as an alternative to provide a self-contained approach for both unsupervised feature learning and classification purposes. However, one of the main problems faced by researchers interested in such approach concerns with a proper selection of its parameters, which play an important role in its final performance. In this paper, we introduced some meta-heuristic techniques for this purpose, as well as we showed they can be more accurate than a random search, that is commonly used by some works. Joao Paulo Papa, Gustavo Rosa, Aparecido Marana, Walter Scheirer and David Cox

International Workshop on Computational Flow and Transport: Modeling, Simulations and Algorithms (CFT) Session 1

Chair: Shuyu Sun

 388 Statistical Inversion of Absolute Permeability in Single Phase Darcy Flow [abstract]Abstract: In this paper, we formulate the permeability inverse problem in the Bayesian framework using total variation (TV) and $\ell_p$ regularization prior. We use the Markov Chain Monte Carlo (MCMC) method for sampling the posterior distribution to solve the ill-posed inverse problem. We present simulations to estimate the distribution for each pixel for the image reconstruction of the absolute permeability. Thilo Strauss, Xiaolin Fan, Shuyu Sun, Taufiquar Khan 32 An enhanced velocity multipoint flux mixed finite element method for Darcy flow on non-matching hexahedral grids [abstract]Abstract: This paper proposes a new enhanced velocity method to directly construct a flux-continuous velocity approximation with multipoint flux mixed finite element method on subdomains. This gives an efficient way to perform simulations on multiblock domains with non-matching hexahedral grids. We develop a reasonable assumption on geometry, discuss implementation issues, and give several numerical results with slightly compressible single phase flow. Benjamin Ganis, Mary Wheeler, Ivan Yotov 124 A compact numerical implementation for solving Stokes equations using matrix-vector operations [abstract]Abstract: In this work, a numerical scheme is implemented to solve Stokes equations based on cell-centered finite difference over staggered grid. In this scheme, all the difference operations have been vectored thereby eliminating loops. This is particularly important when using programming languages that require interpretations, e.g., Matlab and Python. Using this scheme, the execution time becomes significantly smaller compared with non-vectored operations and also become comparable with those languages that require no repeated interpretations like FORTRAN, C, etc. This technique has also been applied to Navier-Stokes equations under laminar flow conditions. Tao Zhang, Amgad Salama, Shuyu Sun, Hua Zhong 265 Numerical Models for the Simulation of Aeroacoustic Phenomena [abstract]Abstract: In the development of a numerical model for aeroacoustic problems, two main issues arise: which level of physical approximation to adopt and which numerical scheme is the most appropriate. It is possible to consider a hierarchy of physical aproximations, ranging from the wave equation, without or with convective effects, to the linearized Euler and Navier-Stokes equations, as well as a wide range of high-order numerical schemes, ranging from compact finite difference schemes to the discontinuous Galerkin method (DGM) for unstructured grids. For problems in complex geometries, significant hydrodynamic-acoustic interactions, coupling acoustic waves and vortical modes, may occur. For example in ducts with sudden changes of area where flow separation occurs in correspondence of sharp edges with a consequent generation of vorticity for viscous effects. To correctly model this coupling, the Navier-Stokes equations, linearized with respect to a representative mean flow, must be solved. The formulation based on Linearized Navier-Stokes (LNS) equations is suitable to deal with problems involving such hydrodynamic-acoustic interactions. The occurrence of geometrical complexities, such as sharp edges, where acoustic energy is transferred into the vortical modes for viscous effects, requires an highly accurate numerical scheme with non only reduced dispersive properties, to accurate model the wave propagation, but also providing a very low level of numerical dissipation on unstructured grids. The DGM is the most appropriate numerical scheme satisfying these requirements. The objective of the present work is to develop an efficient numerical solution of the LNS equations, based on a DGM on unstructured grids. To our knowledge, there is only one work dealing with the solution of the LNS for aeroacoustics where the equations are solved in the frequency domain. In this work we develop the method in the time domain. The non-dispersive and non-diffusive nature of acoustic waves propagating over long distances forces us to adopt highly accurate numerical methods. DGM is one of the most promising scheme due to its intrinsic stability and to its capability to treat unstructured grids. Both advantages make this method well suited for problems characterized by wave propagation phenomena in complex geometries. The main disadvantage of DGM is the high computational requirements because the discontinuous character of the method which adds extra nodes on the interfaces between cells respect to a standard continuous Galerkin Method (GM). Techniques of optimization of the DGM in the case of the Navier-Stokes equations, to reduce the computational effort, are currently object of intense research. At our knowledge, no similar effort is made in the context of the solution of the LNS equations. The LNS equations are derived and the DGM is presented. Preliminary results for the case of the scattering of plane waves traveling in a duct with a sudden area expansion and a comparison between LEE and LNS calculations of vortical modes, are presented. Renzo Arina

Dynamic Data Driven Applications Systems (DDDAS) Session 1

Chair: Craig Douglas

 215 Ensemble Learning for Dynamic Data Assimilation [abstract]Abstract: The organization of an ensemble of initial perturbations by a nonlinear dynamical system can produce highly non-Gaussian patterns, evidence of which is clearly observed in position-amplitude-scale features of coherent fluids. The true distribution of the ensemble is unknown, in part because models are in error and imperfect. A variety of distributions have been proposed in the context of Bayesian inference, including for example, mixture and kernel models. We contend that seeking posterior modes in non-Gaussian inference is fraught with heightened sensitivity to model error and demonstrate this fact by showing that a large component of the total variance remains unaccounted for as more modes emerge. Further, we show that in the presence of bias, this unaccounted variance slows convergence and produces distributions with lower information that require extensive auxiliary clean up procedures such as resampling. These procedures are difficult in large-scale problems where ensemble members may be generated through myriad schemes. We show that by treating the estimation problem entailed as a regression machine, multiple objectives can be incorporated in inference. The relative importance of these objectives can morph over time and can be dynamically adjusted by the data. In particular, we show that both variance reduction and nonlinear modes can be targeted using a stacked cascade generalization. We demonstrate this approach by constructing a new sequential filter called the Boosted Mixture Ensemble Filter and illustrating this on a lorenz system. Sai Ravela 504 A Method for Estimating Volcanic Hazards [abstract]Abstract: This paper presents one approach to determining the hazard threat to a locale due to a large volcanic avalanche. The methodology employed includes large-scale numerical simulations, field data reporting the volume and runout of flow events, and a detailed statistical analysis of uncertainties in the modeling and data. The probability of a catastrophic event impacting a locale is calculated, together with a estimate of the uncertainty in that calculation. By a careful use of simulations, a hazard map for an entire region can be determined. The calculation can be turned around quickly, and the methodology can be applied to other hazard scenarios. E Bruce Pitman and Abani Patra 55 Forecasting Volcanic Plume Hazards With Fast UQ [abstract]Abstract: This paper introduces a numerically-stable multiscale scheme to efficiently generate probabilistic hazard maps for volcanic ash transport using models of transport, dispersion and wind. The scheme relies on graph-based algorithms and low-rank approximations of the adjacency matrix of the graph. This procedure involves representing both the parameter space and physical space by a weighted graph. A combination of clustering and low rank approximation is then used to create a good approximation of the original graph. By performing a multiscale data sampling, a well-conditioned basis of a low rank Gaussian kernel matrix, is identified and used for out-of-sample extensions used in generating the hazard maps. Ramona Stefanescu, Abani Patra, M. I Bursik, E Bruce Pitman, Peter Webley, Matthew D. Jones 45 Forest fire propagation prediction based on overalapping DDDAS forecasts [abstract]Abstract: The effects of forest fires cause a widespread devastation throughout the world every year. A good prediction of fire behavior can help on coordination and management of human and material resources in the extinction of these emergencies. Given the high uncertainty of fire behavior and the difficulty of extracting information required to generate accurate predictions, one system able to adapt to fire dynamics considering the uncertainty of the data is necessary. In this work two different systems based on Dynamic Data Driven Application are applied and a new probabilistic method based on the combination of both approaches is presented. This new method uses the computational power provided by high performance computing systems to adapt the chances in these kind of dynamic environments. Tomás Artés, Adrián Cardil, Ana Cortés, Tomàs Margalef, Domingo Molina, Lucas Pelegrín, Joaquín Ramírez 533 Towards an Integrated Cyberinfrastructure for Scalable Data-Driven Monitoring, Dynamic Prediction and Resilience of Wildfires [abstract]Abstract: Wildfires are critical for ecosystems in many geographical regions. However, our current urbanized existence in these environments is inducing this ecological balance to evolve into a different dynamic leading to the biggest fires in history. Wildfire wind speeds and directions change in an instant, and first responders can only be effective if they take action as quickly as the conditions change. What is lacking in disaster management today is a system integration of real-time sensor networks, satellite imagery, near-real time data management tools, wildfire simulation tools, and connectivity to emergency command centers before, during and after a wildfire. As a first time example of such an integrated system, the WIFIRE project is building an end-to-end cyberinfrastructure for real-time and data-driven simulation, prediction and visualization of wildfire behavior. This paper summarizes the approach and early results of the WIFIRE project to integrate networked observations, e.g., heterogeneous satellite data and real-time remote sensor data with computational techniques in signal processing, visualization, modeling and data assimilation to provide a scalable, technological, and educational solution to monitor weather patterns to predict a wildfire’s Rate of Spread. Ilkay Altintas, Jessica Block, Raymond de Callafon, Daniel Crawl, Charles Cowart, Amarnath Gupta, Mai H. Nguyen, Hans-Werner Braun, Jurgen Schulze, Michael Gollner, Arnaud Trouve, Larry Smarr

Paradigms for Control in Social Systems (PCSS) Session 1

Chair: Justin Ruths

 755 Overview and Introduction [abstract]Abstract: TBD Derek Ruths 751 Jeff's Invited Talk [abstract]Abstract: TBD Jeff Shamma

Agent-Based Simulations, Adaptive Algorithms and Solvers (ABS-AAS) Session 1

Chair: Maciej Paszynski

 754 Agent-Based Simulations, Adaptive Algorithms and Solvers [abstract]Abstract: The aim of this workshop is to integrate results of different domains of computer science, computational science and mathematics. We invite papers oriented toward simulations, either hard simulations by means of finite element or finite difference methods, or soft simulations by means of evolutionary computations, particle swarm optimization and other. The workshop is most interested in simulations performed by using agent-oriented systems or by utilizing adaptive algorithms, but simulations performed by other kind of systems are also welcome. Agent-oriented system seems to be the attractive tool useful for numerous domains of applications. Adaptive algorithms allow significant decrease of the computational cost by utilizing computational resources on most important aspect of the problem. This year following the challenges of ICCS 2015 theme "Computational Science at the Gates of Nature" we invite submissions using techniques dealing with large simulations, e.g. agents based algorithms dealing with big data, model reduction techniques for large problems, fast solvers for large three dimensional simulations, etc. To give - rather flexible - guidance in the subject, the following, more detailed, topics are suggested. These of theoretical brand, like: (a) multi-agent systems in high-performance computing, (b) efficient adaptive algorithms for big problems, (c) low computational cost adaptive solvers, (d) agent-oriented approach to adaptive algorithms, (e) model reduction techniques for large problems, (f) mathematical modeling and asymptotic analysis of large problems, (g) finite element or finite difference methods for three dimensional or non-stationary problems, (h) mathematical modeling and asymptotic analysis. And those with stress on application sphere: (a) agents based algorithms dealing with big data, (b) application of adaptive algorithms in large simulation, (c) simulation and large multi-agent systems, (d) application of adaptive algorithms in three dimensional finite element and finite difference simulations, (e) application of multi-agent systems in computational modeling, (f) multi-agent systems in integration of different approaches. Maciej Paszynski, Robert Schaefer, Krzysztof Cetnarowicz, David Pardo and Victor Calo 631 Coupling Navier-Stokes and Cahn-Hilliard equations in a two-dimensional annular flow configuration [abstract]Abstract: In this work, we present a novel isogeometric analysis discretization for the Navier-Stokes-Cahn-Hilliard equation, which uses divergence-conforming spaces. Basis functions generated with this method can have higher-order continuity, and allow to directly discretize the higher-order operators present in the equation. The discretization is implemented in PetIGA-MF, a high-performance framework for discrete differential forms. We present solutions in a two-dimensional annulus, and model spinodal decomposition under shear flow. Philippe Vignal, Adel Sarmiento, Adriano Côrtes, Lisandro Dalcin, Victor Calo 656 High-Accuracy Adaptive Modeling of the Energy Distribution of a Meniscus-Shaped Cell Culture in a Petri Dish [abstract]Abstract: Cylindrical Petri dishes embedded in a rectangular waveguide and exposed to a polarized electromagnetic wave are often used to grow cell cultures. To guarantee the success of these cultures, it is necessary to enforce that the specific absorption rate distribution is sufficiently high and uniform over the Petri dish. Accurate numerical simulations are needed to design such systems. These simulations constitute a challenge due to the strong discontinuity of electromagnetic parameters of the materials involved, the relative low value of field within the dish cultures compared with the rest of the domain, and the presence of the meniscus shape developed at the liquid/solid interface. The latter greatly increases the level of complexity of the model in terms of geometry and the intensity of the gradients/singularities of the field solution. In here, we employ a three-dimensional (3D) $hp$-adaptive finite element method using isoparametric elements to obtain highly accurate simulations. We analyse the impact of the geometrical modeling of the meniscus shape cell culture in the $hp$-adaptivity. Numerical results concerning the convergence history of the error indicate the numerical difficulties arisen due to the presence of a meniscus-shaped object. At the same time, the resulting energy distribution shows that to consider such meniscus shape is essential to guarantee the success of the cell culture from the biological point of view. Ignacio Gomez-Revuelto, Luis Emilio Garcia-Castillo and David Pardo 162 Leveraging workflows and clouds for a multi-frontal solver for finite element meshes [abstract]Abstract: Scientific workflows in clouds have been successfully used for automation of large-scale computations, but so far they were applied to the loosely-coupled problems, where most workflow tasks can be processed independently in parallel and do not require high volume of communication. The multi-frontal solver algorithm for finite element meshes can be represented as a workflow, but the fine granularity of resulting tasks and the large communication to computation ratio makes it hard to execute it efficiently in loosely-coupled environments such as the Infrastructure-as-a-Service clouds. In this paper, we hypothesize that there exists a class of meshes that can be effectively decomposed into a workflow and mapped onto a cloud infrastructure. To show that, we have developed a workflow-based multi-frontal solver using the HyperFlow workflow engine, which comprises workflow generation from the elimination tree, analysis of the workflow structure, task aggregation based on estimated computation costs, and distributed execution using a~dedicated worker service that can be deployed in clouds or clusters. The results of our experiments using the workflows of over 10,000 tasks indicate that after task aggregation the resulting workflows of over 100 tasks can be efficiently executed and the overheads are not prohibitive. These results lead us to conclusions that our approach is feasible and gives prospects for providing a generic workflow-based solution using clouds for problems typically considered as requiring HPC infrastructure. Bartosz Balis, Kamil Figiela, Maciej Malawski, Konrad Jopek 571 Multi-pheromone ant colony optimization for socio-cognitive simulation purposes [abstract]Abstract: We present an application of Ant Colony Optimisation (ACO) to simulate socio-cognitive features of a population. We incorporated perspective taking ability to generate three different proportions of ant colonies: Control Sample, High Altercentricity Sample, and Low Altercentricity Sample. We simulated their performances on the Travelling Salesman Problem and compared them with the classic ACO. Results show that all three 'cognitively enabled' ant colonies require less time than the classic ACO. Also, though the best solution is found by the classic ACO, the Control Sample finds almost as good a solution but much faster. This study is offered as an example to illustrate an easy way of defining inter-individual interactions based on stigmergic features of the environment. Mateusz Sekara, Kowalski Michal, Aleksander Byrski, Bipin Indurkhya, Marek Kisiel-Dorohinicki, Dana Samson, Tom Lenaerts