2021 09 28 Meeting - math-comp/analysis GitHub Wiki

Participants: Reynald Affeldt, Cyril Cohen, Pierre Roux, Kazuhiko Sakaguchi, Zachary Stone, Pierre-Yves Strub, Laurent Thery

  • issue https://github.com/math-comp/analysis/issues/157 about the documentation of filters
    • about the 1st point: define what is "canonical"
    • about the 2nd point:
      • we have overloading: we first coerce using the element and then infer the filter
      • the goal is to be able to talk about a limit of a filter in R while covering both the finite and the infinite case, as in convergence in \bar R
      • TODO: provide an example to solve this issue
        • maybe not used yet in fact, put as undocumented until there is an actual use
      • LT: document the fact that it is a read-only notation
  • https://github.com/math-comp/analysis/pull/411 (is_interval large inequalities)
    • OK to merge
  • https://github.com/math-comp/analysis/pull/374 (dual real)
    • -oo + +oo = +oo
    • CC: notice a problem with scopes
    • TODO: CC to do the review after a fix by PR
    • put a link to the development that uses this feature as part of the documenting header?
  • CC: put Coq developments depending on MathComp-Analysis in nix and coq-packages to have the CI check the dependencies
  • https://github.com/math-comp/analysis/pull/438 (T0, T1)
    • TODO: reformat definitions and double-check statements
    • ZS: raise a more general question, use filter meet for these definitions?
      • CC: favor filter definition, equivalence with the same description in terms of sets
    • TODO: merge almost as is, just do reformatting
    • TODO: RA to review
  • https://github.com/math-comp/analysis/pull/311 (domain restriction)
    • CC have a completed branch and just need to update the changelog
    • NB: ZS unavailable for the next two weeks
    • the patch function should replace fer in the PRs on integration
  • https://github.com/math-comp/analysis/pull/410 (subspace)
    • CC has almost completed his review
    • this PR can be uses in the "continuous inverse PR" to replace continuity on a segment but there is no need to be delay the merge, this can be fixes later
    • ZS: see also https://github.com/math-comp/analysis/issues/408 to weaken the statement of the mean value theorem using continuity at the end points but on the subspace topology
  • https://github.com/math-comp/analysis/pull/385 (continuous inverse)
    • TODO: update changelog, put the right lemmas in the right places, create a realfun.v file, before merging the trigo PR
    • TODO: AR to commit
  • https://github.com/math-comp/analysis/pull/383 (trigo)
    • most things to go in realfun.v
    • TODO: get rid of names with prime suffixes
    • CC: it is only about R for the exponential? LT: yes
    • some things could be subsumed by power series (notation of radius of convergence, etc)
    • TODO: ping TS about that and report on Zulip
    • ZS: the Arzela-Ascoli PR contains some related elements
  • https://github.com/math-comp/analysis/pull/434 (density of rat)
    • rename archimedean
    • prove the relation with the dense predicate, it should be 5 lines
    • TODO: AR
  • https://github.com/math-comp/analysis/pull/435 (rat is countable)
    • redundancy with domain restriction
    • leq_card predicate should be better defined
    • should be merged after the domain restriction PR
  • inclusion in the Coq platform?
  • a quick work about getting rid of the type classes by CC
    • approach: copy the infrastructure for subalgebra (submodules, etc.) predicaates from MathComp
  • TODO: AR to release 0.3.11
  • https://github.com/math-comp/analysis/pull/398
    • lra to realType
      • it is not parametrical yet, it is based on a certificate, no relation with mczify
    • just work with the R in RStruct as of today