Ponholzer A.,Danube Hospital |
Popper N.,Drahtwarenhandlung Simulation Services |
Popper N.,Institute for Analysis and Scientific Computing |
Breitenecker F.,Institute for Analysis and Scientific Computing |
And 7 more authors.
Anticancer Research | Year: 2010
Purpose: Prostate-specific antigen (PSA) doubling-time (PSA-DT) is an important indicator of progression and survival in men with prostate cancer. Three major limitations regarding PSA-DT determination may lead to inconsistent results: the variety of mathematical methods currently applied, the non-standardized handling of input variables and the potential lack of accuracy due to PSA variability. The aim of this project was to develop a reproducible PSA-DT determination tool which simultaneously provides a PSA-DT error estimation. Materials and Methods: An internet-based PSA-DT calculation tool via nonlinear optimization implementing the least squares error method using the most recent three PSA values was developed. PSA-DT calculation error is estimated via randomly disturbed measurement data streams (n=65) based on a variable (5-25%) PSA variability. Results: According to a simulation in five men, PSA-DT was calculated to be between 1.7 and 15 month (mean: 6.3 month) and determined with another standard tool between 1.3 and 14.5 month (mean: 4.2 month). Conclusion: We present a defined, open and reproducible PSA-DT calculation and PSA-DT error estimation tool based on a standardized PSA data input. This tool is not better compared to other methods but is scientifically standardized and freely accessible via the following internet address: http://adam.drahtwarenhandlung.att/webapp/mg2008/chapter-prostata4/example-psa.
Rupp K.,Institute for Analysis and Scientific Computing
Procedia Computer Science | Year: 2012
High-level C++ proxies for the convenient manipulation of subvectors and submatrices on OpenCL-enabled devices are introduced. It is demonstrated that the programming convenience of standard host-based code can be retained using native C++ language features only, even if massively parallel computing architectures such as graphics processing units are employed. The required modifications of the underlying OpenCL kernels are discussed and a case study of an implementation of the QR-factorization is given. Benchmark results confirm that the convenience of purely CPU-based libraries can be preserved without sacrificing performance of OpenCL-enabled devices, particularly graphics processing units. © 2012 Published by Elsevier Ltd.
Huber M.,Institute for Analysis and Scientific Computing |
Pechstein A.,Institute for Technical Mechanics |
Schoberl J.,Institute for Analysis and Scientific Computing
Lecture Notes in Computational Science and Engineering | Year: 2013
We present hybrid finite element methods for the Helmholtz equation and the time harmonic Maxwell equations, which allow us to reduce the unknowns to degrees of freedom supported only on the element facets and to use efficient iterative solvers for the resulting system of equations. For solving this system, additive and multiplicative Schwarz preconditioners with local smoothers and a domain decomposition preconditioner with an exact subdomain solver are presented. Good convergence properties of these preconditioners are shown by numerical experiments. © Springer-Verlag Berlin Heidelberg 2013.