Petros Koumoutsakos

Reducing the time complexity of the derandomized evolution strategy with covariance matrix adaptation (CMA-ES)

This paper presents a novel evolutionary optimization strategy based on the derandomized evolution strategy with covariance matrix adaptation (CMA-ES). This new approach is intended to reduce the number of generations required for convergence to the optimum. Reducing the number of generations, i.e., the time complexity of the algorithm, is important if a large population size is desired: (1) to reduce the effect of noise; (2) to improve global search properties; and (3) to implement the algorithm on (highly) parallel machines. Our method results in a highly parallel algorithm which scales favorably with large numbers of processors. This is accomplished by efficiently incorporating the available information from a large population, thus significantly reducing the number of generations needed to adapt the covariance matrix. The original version of the CMA-ES was designed to reliably adapt the covariance matrix in small populations but it cannot exploit large populations efficiently. Our modifications scale up the efficiency to population sizes of up to 10n, where n is the problem dimension. This method has been applied to a large number of test problems, demonstrating that in many cases the CMA-ES can be advanced from quadratic to linear time complexity.

TScratch: a novel and simple software tool for automated analysis of monolayer wound healing assays.

Cell migration plays a major role in development, physiology, and disease, and is frequently evaluated in vitro by the monolayer wound healing assay. The assay analysis, however, is a time-consuming task that is often performed manually. In order to accelerate this analysis, we have developed TScratch, a new, freely available image analysis technique and associated software tool that uses the fast discrete curvelet transform to automate the measurement of the area occupied by cells in the images. This tool helps to significantly reduce the time needed for analysis and enables objective and reproducible quantification of assays. The software also offers a graphical user interface which allows easy inspection of analysis results and, if desired, manual modification of analysis parameters. The automated analysis was validated by comparing its results with manual-analysis results for a range of different cell lines. The comparisons demonstrate a close agreement for the vast majority of images that were examined and indicate that the present computational tool can reproduce statistically significant results in experiments with well-known cell migration inhibitors and enhancers.

High-resolution simulations of the flow around an impulsively started cylinder using vortex methods

The development of a two-dimensional viscous incompressible flow generated from a circular cylinder impulsively started into rectilinear motion is studied computationally. An adaptative numerical scheme, based on vortex methods, is used to integrate the vorticity/velocity formulation of the Navier–Stokes equations for a wide range of Reynolds numbers (Re = 40 to 9500). A novel technique is implemented to resolve diffusion effects and enforce the no-slip boundary condition. The Biot–Savart law is employed to compute the velocities, thus eliminating the need for imposing the far-field boundary conditions. An efficient fast summation algorithm was implemented that allows a large number of computational elements, thus producing unprecedented high-resolution simulations. Results are compared to those from other theoretical, experimental and computational works and the relation between the unsteady vorticity field and the forces experienced by the body is discussed.

Dispersion corrections to density functionals for water aromatic interactions

We investigate recently published methods for extending density functional theory to the description of long-range dispersive interactions. In all schemes an empirical correction consisting of a C6r(-6) term is introduced that is damped at short range. The coefficient C6 is calculated either from average molecular or atomic polarizabilities. We calculate geometry-dependent interaction energy profiles for the water benzene cluster and compare the results with second-order Møller-Plesset calculations. Our results indicate that the use of the B3LYP functional in combination with an appropriate mixing rule and damping function is recommended for the interaction of water with aromatics.

Simulations of optimized anguilliform swimming

SUMMARY The hydrodynamics of anguilliform swimming motions was investigated using three-dimensional simulations of the fluid flow past a self-propelled body. The motion of the body is not specified a priori, but is instead obtained through an evolutionary algorithm used to optimize the swimming efficiency and the burst swimming speed. The results of the present simulations support the hypothesis that anguilliform swimmers modify their kinematics according to different objectives and provide a quantitative analysis of the swimming motion and the forces experienced by the body. The kinematics of burst swimming is characterized by the large amplitude of the tail undulations while the anterior part of the body remains straight. In contrast, during efficient swimming behavior significant lateral undulation occurs along the entire length of the body. In turn, during burst swimming, the majority of the thrust is generated at the tail, whereas in the efficient swimming mode, in addition to the tail, the middle of the body contributes significantly to the thrust. The burst swimming velocity is 42% higher and the propulsive efficiency is 15% lower than the respective values during efficient swimming. The wake, for both swimming modes, consists largely of a double row of vortex rings with an axis aligned with the swimming direction. The vortex rings are responsible for producing lateral jets of fluid, which has been documented in prior experimental studies. We note that the primary wake vortices are qualitatively similar in both swimming modes except that the wake vortex rings are stronger and relatively more elongated in the fast swimming mode. The present results provide quantitative information of three-dimensional fluid-body interactions that may complement related experimental studies. In addition they enable a detailed quantitative analysis, which may be difficult to obtain experimentally, of the different swimming modes linking the kinematics of the motion with the forces acting on the self-propelled body. Finally, the optimization procedure helps to identify, in a systematic fashion, links between swimming motion and biological function.

Accelerating evolutionary algorithms with Gaussian process fitness function models

We present an overview of evolutionary algorithms that use empirical models of the fitness function to accelerate convergence, distinguishing between evolution control and the surrogate approach. We describe the Gaussian process model and propose using it as an inexpensive fitness function surrogate. Implementation issues such as efficient and numerically stable computation, exploration versus exploitation, local modeling, multiple objectives and constraints, and failed evaluations are addressed. Our resulting Gaussian process optimization procedure clearly outperforms other evolutionary strategies on standard test functions as well as on a real-world problem: the optimization of stationary gas turbine compressor profiles.

Contrasting Actions of Selective Inhibitors of Angiopoietin-1 and Angiopoietin-2 on the Normalization of Tumor Blood Vessels

Angiopoietin-1 (Ang1) and angiopoietin-2 (Ang2) have complex actions in angiogenesis and vascular remodeling due to their effects on Tie2 receptor signaling. Ang2 blocks Ang1-mediated activation of Tie2 in endothelial cells under certain conditions but is a Tie2 receptor agonist in others. We examined the effects of selective inhibitors of Ang1 (mL4-3) or Ang2 (L1-7[N]), alone or in combination, on the vasculature of human Colo205 tumors in mice. The Ang2 inhibitor decreased the overall abundance of tumor blood vessels by reducing tumor growth and keeping vascular density constant. After inhibition of Ang2, tumor vessels had many features of normal blood vessels (normalization), as evidenced by junctional accumulation of vascular endothelial-cadherin, junctional adhesion molecule-A, and platelet/endothelial cell adhesion molecule-1 in endothelial cells, increased pericyte coverage, reduced endothelial sprouting, and remodeling into smaller, more uniform vessels. The Ang1 inhibitor by itself had little noticeable effect on the tumor vasculature. However, when administered with the Ang2 inhibitor, the Ang1 inhibitor prevented tumor vessel normalization, but not the reduction in tumor vascularity produced by the Ang2 inhibitor. These findings are consistent with a model whereby inhibition of Ang2 leads to normalization of tumor blood vessels by permitting the unopposed action of Ang1, but decreases tumor vascularity primarily by blocking Ang2 actions.

PPM: a highly efficient parallel particle-mesh library for the simulation of continuum systems

This paper presents a highly efficient parallel particle-mesh (PPM) library, based on a unifying particle formulation for the simulation of continuous systems. In this formulation, the grid-free character of particle methods is relaxed by the introduction of a mesh for the reinitialization of the particles, the computation of the field equations, and the discretization of differential operators. The present utilization of the mesh does not detract from the adaptivity, the efficient handling of complex geometries, the minimal dissipation, and the good stability properties of particle methods.The coexistence of meshes and particles, allows for the development of a consistent and adaptive numerical method, but it presents a set of challenging parallelization issues that have hindered in the past the broader use of particle methods. The present library solves the key parallelization issues involving particle-mesh interpolations and the balancing of processor particle loading, using a novel adaptive tree for mixed domain decompositions along with a coloring scheme for the particle-mesh interpolation.The high parallel efficiency of the library is demonstrated in a series of benchmark tests on distributed memory and on a shared-memory vector architecture. The modularity of the method is shown by a range of simulations, from compressible vortex rings using a novel formulation of smooth particle hydrodynamics, to simulations of diffusion in real biological cell organelles.The present library enables large scale simulations of diverse physical problems using adaptive particle methods and provides a computational tool that is a viable alternative to mesh-based methods.

A theoretical prediction of friction drag reduction in turbulent flow by superhydrophobic surfaces

We present a theoretical prediction for the drag reduction rate achieved by superhydrophobic surfaces in a turbulent channel flow. The predicted drag reduction rate is in good agreement with results obtained from direct numerical simulations at Reτ≃180 and 400. The present theory suggests that large drag reduction is possible also at Reynolds numbers of practical interest (Reτ∼105–106) by employing a hydrophobic surface, which induces a slip length on the order of ten wall units or more.

MorphoGraphX: A platform for quantifying morphogenesis in 4D

Morphogenesis emerges from complex multiscale interactions between genetic and mechanical processes. To understand these processes, the evolution of cell shape, proliferation and gene expression must be quantified. This quantification is usually performed either in full 3D, which is computationally expensive and technically challenging, or on 2D planar projections, which introduces geometrical artifacts on highly curved organs. Here we present MorphoGraphX (www.MorphoGraphX.org), a software that bridges this gap by working directly with curved surface images extracted from 3D data. In addition to traditional 3D image analysis, we have developed algorithms to operate on curved surfaces, such as cell segmentation, lineage tracking and fluorescence signal quantification. The softwares modular design makes it easy to include existing libraries, or to implement new algorithms. Cell geometries extracted with MorphoGraphX can be exported and used as templates for simulation models, providing a powerful platform to investigate the interactions between shape, genes and growth. DOI: http://dx.doi.org/10.7554/eLife.05864.001