{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreidtki3mbkm2pdqqn4bomz7nypttbzcjzjxmdr7jog42i6pjwoznha",
    "uri": "at://did:plc:pi6woz4d47bkuws673w2il2r/app.bsky.feed.post/3mlrtzrznwud2"
  },
  "path": "/t/knownnat-indexed-vectors/14099#post_19",
  "publishedAt": "2026-05-14T02:36:32.000Z",
  "site": "https://discourse.haskell.org",
  "textContent": "I’m afraid switching to a data representation improved performance dramatically. Unfortunate, but unsurprising. Doing heavy math with these kinds of vectors, creating new vectors from old, requires keeping references around to all of these functions. I will experiment with memoization, but I think this spells doom for the approach I outlined in the original post.",
  "title": "KnownNat-indexed vectors"
}