Multi objective Optimization: The String Example

In algorithms research, we formulate problems as optimizations. In reality, there are many objectives. If world were simple, we would optimize the right parameter. Else, we have to bound some of the parameters, and optimize wrt the others.

In a recent plenary talk at ESA 2010, Paolo Ferragina explores the multiple parameters in string indexing data structures: running time, memory hierarchy, compression, and now, it seems energy efficiency, and mobile. The talk connects the theoretical string data structure problems to practical performance and to modern data systems like Cassandra (facebook), Cosmos (Bing) and HyperTable (Bing).