Coupling of Parallelized DEGAS 2 and UEDGE Codes

The increased availability of massively parallel computers combined with the scalability of the Monte Carlo algorithm implies that the physical realism provided by Monte Carlo codes can be obtained with short computation times. The coupling of parallelized versions of the Monte Carlo neutral transport code <A HREF=http://w3.pppl.gov/degas2/>DEGAS 2</A> and the UEDGE fluid plasma transport code builds upon the success of the B2-EIRENE system and promises a practical tool for analyzing experimental results and predicting scrape-off layer conditions in future devices. DEGAS 2 now contains a complete set of physics, including elastic scattering. Benchmarks against EIRENE have been performed, yielding agreement to within 5 % and comparable single processor run times. The focus of the current work is on improving the coupling to UEDGE by increasing the parallel processing speed of both codes and reducing the impact of Monte Carlo noise on UEDGE iterations. Correlated sampling can be exploited to varying degrees to minimize run-to-run noise. A new, quasi-random, direct sampling scheme lowers the error in the initial source sampling from 1 / √N to 1 / N. Slab geometry examples will be presented and comparisons with coupled UEDGE-EIRENE runs will be made. Demonstration calculations of Alcator C-Mod experimental discharges will also be shown.