Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Solving problems under uncertainty

Technical Report ·
OSTI ID:5371579
This final technical report covers mainly the progress made solving large scale linear programs and the recent successes in optimizing large-scale planning problems under uncertainty (stochastic programs). The general goal of the research was the development of effective methods for solving stochastic linear programs. Stochastic programming models are designed to deal with the uncertainty in the values of the parameters of complex scheduling, planning and control problems. Because solutions obtained from such models properly hedge against future uncertainties they are much to be preferred to solutions produced by deterministic models. The stochastic linear program, if it could be solved efficiently, is consequently a problem of greatest practical importance, especially in the fields of energy planning, industry and finance. Our recent work has made it possible to efficiently solve certain important classes of stochastic linear programs. Our approach combines large-scale system decomposition techniques and advanced sampling methods with the use of parallel processors. The use of adaptive importance sampling in tests conducted so far has so significantly reduced required sample sizes that it is now possible to solve some previously intractable problems on a personal computer. We have successfully solved several very large-scale energy planning and finance models; these problems are discussed in the technical appendix.
Research Organization:
Stanford Univ., CA (United States). Dept. of Operations Research
Sponsoring Organization:
DOE; DOD; EPRI; NSF; USDOE, Washington, DC (United States); Department of Defense, Washington, DC (United States); Electric Power Research Inst., Palo Alto, CA (United States); National Science Foundation, Washington, DC (United States)
DOE Contract Number:
FG03-87ER25028
OSTI ID:
5371579
Report Number(s):
DOE/ER/25028-T2; ON: DE91017940
Country of Publication:
United States
Language:
English