Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Cosmin Petra Project Name: stoch_prog Division: MCS Project title: Stochastic Programming and Application for Energy Systems Associated funding: DOE-ASCR: Mathematics of Petascale Data Program. Scalable Algorithms for Gaussian Processes. Mihai Anitescu, PI. Other Systems: We have obtained and used approx 2 million core hours on Intrepid. We also applied for a similar amount for the next year. Science: This project is an optimization framework that exploits the weather forecast information in the operation of energy systems. Our objective is to illustrate how the use of forecast information and high accuracy physics and computations can translate into lower operating costs. We will solve the stochastic integer unit commitment/energy dispatch problem with network transmission constraints and wind power generation. Project description: In the previous years, we obtained an unprecedented level of parallelization in solving stochastic optimization problems by exploiting the half-arrow shaped structure these problems have and using Elemental to reduce the computational bottleneck associated with the first stage dense Schur complement system. We also implemented a mixed approach MPI+shared memory (SMP) based on IBM's WSMP BG/P sparse solver to obtain intra-node shared memory parallelization. Our C++ code, PIPS, was able to solve a stochastic economic dispatch problem with close to 1 billion variables using 131072 cores on Intrepid with a strong scaling efficiency of more than 90%. A paper on these findings has been accepted to SC11. The final version can be found here: http://www.mcs.anl.gov/~anitescu/PUBLICATIONS/2011/miles-2011-pips-SC.pdf Currently, we only have the capability to solve problems with variables of a continuous nature. In electricity energy systems, however, it is also necessary to deal with binary (integer) variables representing switches and the “on/off” status of generators (a.k.a unit commitment) and transmission lines. In addition, in long-term planning studies, it is necessary to use binary variables to model the addition of transmission lines and generators to satisfy future demands. These optimization problems are called stochastic mixed-integer problems. While it is not uncommon to solve a relaxation of the integer problem that is cheaper to compute, the relaxed solutions are sub-optimal, generate extra-costs in planning the thermal units, reduce the benefits of using wind power and increase the carbon footprint. Furthermore, the relaxed solutions are usually non-integer and the common practice of rounding them to the closer integer causes a failure to satisfy demand and, inevitably , a higher price for energy since expensive electricity from reserves has to be pumped into the grid. Integer optimization problems have a combinatorial nature, and are often solved by recursively exploring the finite but exponentially large binary solution space using an approach called branch and bound. Parallel branch-and-bound has been well studied. An open-source framework, CHiPPS (https://projects.coin-or.org/CHiPPS), has been successfully run on a Blue Gene/L system. However, parallel branch-and-bound has not previously been used to solve problems so large that the continuous relaxation must be decomposed and solved in parallel. We plan to implement such a framework, which combines parallel branch-and-bound, based on existing codes such as CHiPPS, and our code PIPS, thereby obtaining two levels of parallelism. The requested hours on Fusion will be used to run small and medium scale computations with the parallel branch and bound framework we plan to implement. Project URL: http://www.mcs.anl.gov/~petra/pips.html Current FY Hours Used: undetermined amount New FY Requested allocation: 250000 Q1: 25000 Q2: 100000 Q3: 100000 Q4: 25000 Justification: Thank You, The LCRC Accounts System