We introduce the algorithm SHEBO (surrogate optimization of problems with hidden constraints and expensive black-box objectives), an efficient optimization algorithm that employs surrogate models to solve computationally expensive black-box simulation optimization problems that have hidden constraints. Hidden constraints are encountered when the objective function evaluation does not return a value for a parameter vector. These constraints are often encountered in optimization problems in which the objective function is computed by a black-box simulation code. SHEBO uses a combination of local and global search strategies together with an evaluability prediction function and a dynamically adjusted evaluability threshold to iteratively select new sample points. We compare the performance of our algorithm with that of the mesh-based algorithms mesh adaptive direct search (MADS, NOMAD [nonlinear optimization by mesh adaptive direct search] implementation) and implicit filtering and SNOBFIT (stable noisy optimization by branch and fit), which assigns artificial function values to points that violate the hidden constraints. Our numerical experiments for a large set of test problems with 2–30 dimensions and a 31-dimensional real-world application problem arising in combustion simulation show that SHEBO is an efficient solver that outperforms the other methods for many test problems.

Modern cosmological simulations have reached the trillion-element scale, rendering data storage and subsequent analysis formidable tasks. To address this circumstance, we present a new MPI-parallel approach for analysis of simulation data while the simulation runs, as an alternative to the traditional workflow consisting of periodically saving large data sets to disk for subsequent 'offline' analysis. We demonstrate this approach in the compressible gasdynamics/N-body code Nyx, a hybrid MPI + OpenMP code based on the BoxLib framework, used for large-scale cosmological simulations. We have enabled on-the-fly workflows in two different ways: one is a straightforward approach consisting of all MPI processes periodically halting the main simulation and analyzing each component of data that they own ('in situ'). The other consists of partitioning processes into disjoint MPI groups, with one performing the simulation and periodically sending data to the other 'sidecar' group, which post-processes it while the simulation continues ('in-transit'). The two groups execute their tasks asynchronously, stopping only to synchronize when a new set of simulation data needs to be analyzed. For both the in situ and in-transit approaches, we experiment with two different analysis suites with distinct performance behavior: one which finds dark matter halos in the simulation using merge trees to calculate the mass contained within iso-density contours, and another which calculates probability distribution functions and power spectra of various fields in the simulation. Both are common analysis tasks for cosmology, and both result in summary statistics significantly smaller than the original data set. We study the behavior of each type of analysis in each workflow in order to determine the optimal configuration for the different data analysis algorithms.

We present a graph-based strategy for representing the computational domain for embedded boundary discretizations of conservation-law PDE's. The representation allows recursive generation of coarse-grid geometry representations suitable for multigrid and adaptive mesh re nement calculations. Using this scheme, we implement a simple multigrid V-cycle relaxation algorithm to solve the linear elliptic equations arising from a block-structured adaptive discretization of the Poisson equation over an arbitrary two-dimensional domain. We demonstrate that the resulting solver is robust to a wide range of two-dimensional geometries, and performs as expected for multigrid-based schemes, exhibiting O N log N scaling with system size, N.

No abstract

No abstract

The classical plasma electron heat flux can greatly overestimate the physical heat flux when the electron mean-free-path becomes long compared to the electron temperature gradient scale length. For fluid modelling of plasmas, a common remedy is to apply an artificial "flux limiter" that keeps the heat flux at a physically reasonable value in this long mean-free-path regime. The ad-hoc limiter is not derived from first principles and it introduces a poorly understood free parameter into the model. We study the effect of this parameter in our divertor plasma models to understand how it influences the computed solution. We investigate regimes of both short and long mean-free-path and consistently find large parameter sensitivity. Thus, without additional experimental or theoretical guidance in the choice of flux-limit parameter, flux limiting does not appear to provide an acceptable basis for predictive plasma fluid modelling. BackgroundEarly experiments with laser-heated plasmas demonstrated that the classical expressions for thermal conductivities lead to significantly overestimated electron heat fluxes in regimes of long mean-free-path. This observation led to the flux-limit models which are employed to study and predict divertor behavior. Typically, flux limiters impose -.a maximum allowable heat flux of the form, ql = CYeVthenekTe (where Vthe = 4 -is the electron thermal velocity, and a, is an arbitrary constant such that qt remains less than the flux of thermal energy that could be carried by the electrons moving at their thermal speed). The limited heat flux is written as qe = (l/qs + l/ql)-', which is an arbitrary interpolation between the free-streaming and Spitzer values (here, qs is

No abstract

No abstract