Skip to content

2023 07 27 Meeting

affeldt-aist edited this page Jul 27, 2023 · 1 revision

Participants: Reynald, Cyril, Pierre, Zachary, Takafumi

  • naming convention for bigcup:
    • see comment https://github.com/math-comp/analysis/issues/988#issuecomment-1651515771
    • see also bigcup_set0 vs. bigcup0
    • \bigcup_(x in ~` P) F x, we write bigcup_setC
      • in general, in bigcup_xyz, xyz is about the in part
      • \bigcup_(A in P) ~` F A, we make a longer name, we may write something like bigcup_setCr
        • instead of, e.g., bigcupP_setC
        • also, try to swap the equality as in, e.g., setC_bigcup
  • Discussion about the pending issues about renaming
    • do not rename ball to open_ball because balls are not necessarily open
    • we actually need a stronger notion of pseudometric space where there is a distance between any two points
    • more generally, there is a bunch of renaming about closed balls
    • TODO: we should have a notion of open balls
    • about scale_cball from the PR on Vitali's lemma:
      • introduce a predicate is_ball so that we the predicate is satisfied we have have a center and a radius function
      • or introduce a ball record with a coercion to a set?
      • anyway, we'd like to unify the various notions of balls
      • we should be aiming for scale_ball to be a function from balls to balls
  • https://github.com/math-comp/analysis/pull/973 (Vitali's lemma)
    • change the lemma from B : I -> R * {posnum R} to two functions I -> R and I -> {posnum R}?
      • at least try not to expose the scale_cball by introducing alternative statements
    • as of now, many lemmas could be generalize to dimension n but they are still in dimention 1
  • news about perf of HB branch: thanks to latest elpi (1.17.0) and master branches of coq-elpi and HB:
    • total analysis: 26:31 (used to be 40 min) (2.6 GB (used to be 10 GB))
    • loading elpi code (HB): 02:59 (used to be 16 min)
    • running elpi code (HB): 08:55
    • not all elpi code is released so MathComp-Analysis 1 can maybe wait a bit
  • PR triaging: