Application of a hybrid MPI/OpenMP approach for parallel groundwater model calibration using multi-core computers

Long-term electrical resistivity monitoring of recharge-induced contaminant plume behavior

“…The utility of ERT datasets for such monitoring is dependent on the relationship between resistivity and nitrate concentration; at ORNL the ERT data were useful for identifying recharge-induced changes in nitrate concentrations that occurred in the 100-2000 mg/l range, but were insensitive to further resistivity decrease at depth due to the asymptotic resistivity behavior at high nitrate concentrations. Ongoing research is focused on expanding the study to explore recharge-induced dynamics at the leading edge of the plume using time-lapse ERT, on understanding how recharge may also impact other geochemical variations that could contribute to complex electrical signatures (i.e., Skold et al, 2011), and on using the information obtained by geophysics to parameterize and validate reactive transport models that are being performed to simulate long term plume behavior (i.e., Tang et al, 2010).…”

“…For this study, we use time-lapse surface and cross-well ERT data and associated wellbore geochemical data that have been co-collected over a 12-month timeframe, which includes a dry period, a wetting-up period, and a wet period. From a process perspective, our challenge is to determine if the time-lapse ERT data can be useful for quantifying spatiotemporal variations in recharge-related plume behavior, which can then be used to inform reactive transport modeling ongoing at the site (Tang et al, 2010). From a geophysical perspective, our challenge is to determine the sensitivity of the electrical response to changes in nitrate concentrations in the saturated zone with the additional complication of strongly dipping geological heterogeneity.…”

On parallelization of the loop over elements in FEAP

“…It has been applied to a range of problems, e.g. in recent years to: an explicit FE code for simulation of impact [19], solution of Navier-Stoke's equations [27], calibration of a model for ground water flow [26], in solver of a system of equations [6], integration over a volume in electromagnetic problems, [22], the Delaunay triangularization [13], and a topology optimization with stress constraints [20]. In these works, the speedup achieved due to implementation of OpenMP was about 3.5 for 4 threads, 10.6 for 12 threads and 11 for 16 threads, which confirms that OpenMP is a useful tool indeed.…”

Enabling and scaling biomolecular simulations of 100 million atoms on petascale machines with a multicore-optimized message-driven runtime

“…This MPI+pthreads type of hybrid programming requires the application developers to explicitly deal with multi-threading concurrency issues of threads. In general, compared with the popular hybrid programming approach including MPI+OpenMP [20,12,21] on SMP clusters, our SMP model is supported under the message-driven…”