| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
- Replace `altP eqP` and `altP (_ =P _)` with `eqVneq`:
The improved `eqVneq` lemma (#351) is redesigned as a comparison predicate and
introduces a hypothesis in the form of `x != y` in the second case. Thus,
`case: (altP eqP)`, `case: (altP (x =P _))` and `case: (altP (x =P y))` idioms
can be replaced with `case: eqVneq`, `case: (eqVneq x)` and
`case: (eqVneq x y)` respectively. This replacement slightly simplifies and
reduces proof scripts.
- use `have [] :=` rather than `case` if it is better.
- `by apply:` -> `exact:`.
- `apply/lem1; apply/lem2` or `apply: lem1; apply: lem2` -> `apply/lem1/lem2`.
- `move/lem1; move/lem2` -> `move/lem1/lem2`.
- Remove `GRing.` prefix if applicable.
- `negbTE` -> `negPf`, `eq_refl` -> `eqxx` and `sym_equal` -> `esym`.
|
|
Added lemmas `big_enum_cond`, `big_enum` and `big_enumP` to handle more
explicitly big ops iterating over explicit enumerations in a `finType`.
The previous practice was to rely on the convertibility between
`enum A` and `filter A (index_enum T)`, sometimes explicitly via the
`filter_index_enum` equality, more often than not implicitly.
Both are likely to fail after the integration of `finmap`, as the
`choiceType` theory can’t guarantee that the order in selected
enumerations is consistent.
For this reason `big_enum` and the related (but currently unused)
`big_image` lemmas are restricted to the abelian case. The `big_enumP`
lemma can be used to handle enumerations in the non-abelian case, as
explained in the `bigop.v` internal documentation.
The Changelog entry enjoins clients to stop relying on either
`filter_index_enum` and convertibility (though this PR still provides
both), and warns about the restriction of the `big_image` lemma set to
the abelian case, as it it a possible source of incompatibility.
|
|
Replaced the legacy generalised induction idiom with a more robust one
that does not rely on the `{-2}` numerical occurrence selector, using
either new helper lemmas `ubnP` and `ltnSE` or a specific `nat`
induction principle `ltn_ind`.
Added (non-strict in)equality induction helper lemmas
Added `ubnP[lg]?eq` helper lemmas that abstract an integer expression
along with some (in)equality, in preparation for some generalised
induction. Note that while `ubnPleq` is very similar to `ubnP` (indeed
`ubnP M` is basically `ubnPleq M.+1`), `ubnPgeq` is used to remember
that the inductive value remains below the initial one.
Used the change log to give notice to users to update the generalised
induction idioms in their proofs to one of the new forms before
Mathcomp 1.11.
|
|
add existsPn/forallPn lemmas
|
|
|
|
|
|
- Change the naming of permutation lemmas so they conform to a
consistent policy: `perm_eq` lemmas have a `perm_` (_not_ `perm_eq`)
prefix, or sometimes a `_perm` suffix for lemmas that _prove_ `perm_eq`
using a property when there is also a lemma _using_ `perm_eq` for the
same property. Lemmas that do not concern `perm_eq` do _not_ have
`perm` in their name.
- Change the definition of `permutations` for a time- and space-
back-to-front generation algorithm.
- Add frequency tally operations `tally`, `incr_tally`, `wf_tally` and
`tally_seq`, used by the improved `permutation` algorithm.
- add deprecated aliases for renamed lemmas
|
|
- add notation support for lemma renaming PRs, helping clients adjust
to the name change by emitting warning or raising errors when the old
name is used. The default is to emit warnings, but clients can change
this to silence (if electing to delay migration) or errors (to help
with actual migration). Usage:
Notation old_id := (deprecate old_id new_id) (only parsing).
—> Caveat 1: only prenex maximal implicit of `new_id` are preserved, so,
as `Notation` does not support on-demand implicits, the latter should
be added explicitly as in `(deprecate old new _ _)`.
—> Caveat 2: the warnings are emitted by a tactic-in-term, which
is interpreted during type elaboration. As the SSReflect elaborator may
re-infer type in arguments multiple times (notably, views and arguments
to `apply` and `rewrite`), clients may see duplicate warnings.
- use the `deprecate` facility to introduce the first part of the
refactoring of `seq` permutation lemmas : only lemmas concerning
`perm_eq` should have `perm` as a component of their name.
- document local additions in `ssreflect.v` and `ssrbool.v`
- add 8.8 `odd-order` regression
- the explicit beta-redex idiom use in the `perm_uniq` and
`leq_min_perm` aliases avoids a strange name-sensitive bug of view
interpretation in Coq 8.7.
|
|
|
|
Like injectivity lemmas, instances of cancellation lemmas (whose
conclusion is `cancel ? ?`, `{in ?, cancel ? ?}`, `pcancel`, or
`ocancel`) are passed to
generic lemmas such as `canRL` or `canLR_in`. Thus such lemmas should
not have trailing on-demand implicits _just before_ the `cancel`
conclusion, as these would be inconvenient to insert (requiring
essentially an explicit eta-expansion).
We therefore use `Arguments` or `Prenex Implicits` directives to make
all such arguments maximally inserted implicits. We don’t, however make
other arguments implicit, so as not to spoil direct instantiation of
the lemmas (in, e.g., `rewrite -[y](invmK injf)`).
We have also tried to do this with lemmas whose statement matches a
`cancel`, i.e., ending in `forall x, g (E[x]) = x` (where pattern
unification will pick up `f = fun x => E[x]`).
We also adjusted implicits of a few stray injectivity
lemmas, and defined constants.
We provide a shorthand for reindexing a bigop with a permutation.
Finally we used the new implicit signatures to simplify proofs that
use injectivity or cancellation lemmas.
|
|
This increases performance 10% - 15% for Coq v8.6.1 - v8.9.dev.
Tested on a Debain-based 16-core build server and
a Macbook Pro laptop with 2,3 GHz Intel Core i5.
| | Compilation time, old | Compilation | Speedup |
| | (mathcomp commit 967088a6f87) | time, new | |
| Coq 8.6.1 | 10min 33s | 9min 10s | 15% |
| Coq 8.7.2 | 10min 12s | 8min 50s | 15% |
| Coq 8.8.2 | 9min 39s | 8min 32s | 13% |
| Coq 8.9.dev(05d827c800544) | 9min 12s | 8min 16s | 11% |
| | | | |
It seems Coq at some point fixed the problem `_ : Type` was
supposed to solve.
|
|
See the discussion here:
https://github.com/math-comp/math-comp/pull/242#discussion_r233778114
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
