The Saskatoon Chapter of the

 Canadian Operational

Research Society

 

 

Winfried Grassmann

Computer Science Department

University of Saskatchewan

 

Thursday, April 13, 2000

Park Town Hotel

Poplar Room

Cocktails 5:30, Dinner 6:00, Presentation 7:00

 

 

 

Simulating Markovian Systems Containing Rare Events

 

Markovian systems are systems where all future probabilities depend on the present state only, and not on the past. This property is not only mathematically convenient, but it also reduces the amount of data needed for modelling. Unfortunately, classic numerical methods have a computational complexity that increases exponentially with the number of variables, which makes these methods impractical unless the number of variables is small.  Otherwise, simulation is required. Unfortunately, simulation becomes very expensive if some events are extremely rare, but of great economic importance, such as accidents.  Interest in such rare events has increased because of communication networks where the cell loss probabilities must often be held below 10E-9. The talk describes some methods how to deal with rare events of economic importance by combining classical numerical methods with simulation.

 

 

 

 

Tickets $17

For more information contact:

Winfried Grassmann

grassman@cs.usask.ca

966-4898

Kent Kostuk

k.kostuk@fcl.ca

244-3295