|Re: Algorithm efficiently generating L(G), for a non-selfembedding CFG email@example.com (2001-03-31)|
|Re: Re: Algorithm efficiently generating L(G), for a non-selfembedding firstname.lastname@example.org (2001-04-04)|
|Re: Re: Algorithm efficiently generating L(G), for a non-selfembedding email@example.com (Manoj Plakal) (2001-04-10)|
|From:||Manoj Plakal <firstname.lastname@example.org>|
|Date:||10 Apr 2001 01:21:45 -0400|
|Organization:||University of Wisconsin-Madison|
|Posted-Date:||10 Apr 2001 01:21:45 EDT|
On Tuesday 03 April 2001 23:16, Gershon Miles\ wrote:
> Simply "walking the tree" will require recalculating many variables
> with the grammar.
Well, can't you add some memoization to the tree-walk
to avoid re-walking subtrees? And trade-off space with
time by varying the size of the memoization cache?
A related idea would be to do a bottom-up computation
of the tree and do something like dynamic programming
to avoid redundant recomputations.
Return to the
Search the comp.compilers archives again.