dy@grad17.cs.duke.edu (Doreen Yen) (06/14/90)
It seems obvious that parallelism for functional programming can be
had with MAP and REDUCE. What are the refs. that actually say this?
Logic programmers find solutions with generate&test and improve
efficiency by pushing the testing into the generation (Shapiro,Coelho).
What is the equivalent idea for functional programmers? Is this in any
way related to lazy evaluation?
Doreen Yen UUCP: ..!{ihnp4,decvax}!duke!dy
CSNET: dy@duke
ARPA: dy@cs.duke.edu
"Have a Happy Year of the Horse"jgk@osc.COM (Joe Keane) (06/16/90)
In article <20193@duke.cs.duke.edu> dy@grad17.cs.duke.edu (Doreen Yen) writes: >It seems obvious that parallelism for functional programming can be >had with MAP and REDUCE. What are the refs. that actually say this? It's nothing special. If REDUCE can evaluate elements of the list in parallel, it's only because the reduction function (CONS for MAP) can evaluate its arguments in parallel. This is true exactly if the reduction function is strict in both arguments, which is always in a strict language.