Apropos of...

Checking - there’s nothing faster than a binary search is there unless you are also worried about insert time which means back trees right? Ie a slow indexing process that creates a simple ordered list for fast binary search is the most efficient for reads, but not inserts (as you have to re-sort the file each time)?

Tradfi coder by day, hacker by night. Doesn’t know night from day. Freelance scribe. Dad cook. Dog butler.  Happy Grandpa. Draws badly.

Technically there's faster algorithms, but it's only in use case specific conditions. What are you trying to do?

BDFL of Quilibrium, Eng @ Farcaster, ex-Coinbase, always opinionated, never hydrated

/quilibrium

you could use a radix trie if you're always operating off the prefix, which does bring things back into binary search, or you could opt for a hypergraph embedding of shared segments such that progressing from a root marker you have a radix, or omitting the root you just do a hypergraph search

An inverted index of stemmed/lemmatized words so I can't use the integer key patterns of a forward index - although I could use prefix patterns now you've reminded me!