Algorithms and Experiments (ALEX98)
Building bridges between theory and applications
Trento, Italy, February 9 - 11, 1998
Discussion topics
- What to measure -- (Also known as runningtime vs operation counts)
what is the best way to maximize both universality and precision in the
measurements of algorithm performance.
- Nature vs nurture : How much does the machine environement (platform,
operating system, compiler) contribute to performance, compared to t
he ``algorithmic, portable'' part?
- Would it be possible to build a more realistic model of computation ,
that could be used for both theoretical analysis and experimental work?
Something like MIX, but updated to deal with things like memory hierarchies.
Possibilities are Knuth's ``mems,'' or recent work analyzing cacheing
behavior of some data structures.
- When is an experimental paper publishable? How much scope should
it have, and how much reliability? Can we build a list of guidelines for
reviewers of papers, or are standards well understood by now?
- What do statisticians know that we should learn? What questions
do we have about analyzing data?
- Other topics are welcome: send them to the workshop email!
alex98@rtm.science.unitn.it
Back to ALEX98 home page.