I'm new in this area and have been reading some paper but i have some problem that i hope you solve it.
i 'm reading lzp paper (by Charles Bloom) ...in implementation section of lzp1 and lzp2 it says
to calculate hash.
1.using 3 bytes as context (fixed context for lzp1 as paper said) we have pow(2,24) states but H has only pow(2,12) states
so we have collision here. is'nt it a problem to have collision?
2.do we achieve better compression ratio if just define a 16MB( pow(2,24) ) linear array(assume that there is
no problem in allocating 16 MB of memory.)?
3.where did this H formula come from? why shift right by 11 and then xor by C (why not shift by 12 for example)?
thx in advance
(BTW today is my birthday )