Commit graph

2 commits

Author SHA1 Message Date
Joachim Breitner
d1174e10e6
feat: always run clean_wf, even before decreasing_by (#5016)
Previously, the tactic state shown at `decreasing_by` would leak lots of
details about the translation, and mention `invImage`, `PSigma` etc.
This is not nice.
  
So this introduces `clean_wf`, which is like `simp_wf` but using
`simp`'s `only` mode, and runs this unconditionally. This should clean
up the goal to a reasonable extent.
  
Previously `simp_wf` was an unrestricted `simp […]` call, but we
probably don’t want arbitrary simplification to happen at this point, so
this now became `simp only` call. For backwards compatibility,
`decreasing_with` begins with `try simp`. The `simp_wf` tactic
is still available to not break too much existing code; it’s docstring
suggests to no longer use it.

With `set_option cleanDecreasingByGoal false` one can disable the use of
`clean_wf`. I hope this is only needed for debugging and understanding.
  
Migration advise: If your `decreasing_by` proof begins with `simp_wf`,
either remove that (if the proof still goes through), or replace with
`simp`.
  
I am a bit anxious about running even `simp only` unconditionally here,
as it may do more than some user might want, e.g. because of options
like `zetaDelta := true`. We'll see if we need to reign in this tactic
some more.

I wonder if in corner cases the `simp_wf` tactic might be able to close
the goal, and if that is a problem. If so, we may have to promote simp’s
internal `mayCloseGoal` parameter to a simp configuration option and use
that here.
  
fixes #4928
2024-08-15 14:42:15 +00:00
Joachim Breitner
4c57da4b0f
feat: infer termination arguments like xs.size - i (#3666)
a common pattern for recursive functions is
```
def countUp (n i acc : Nat) : Nat :=
  if i < n then
    countUp n (i+1) (acc + i)
  else
    acc
```
where we increase a value `i` until it hits an upper bound. This is
particularly common with array processing functions:
```
$ git grep 'termination_by.*size.*-' src/|wc -l
26
```

GuessLex now recognizes this pattern. The general approach is:

For every recursive call, check if the context contains hypotheses of
the form `e₁ < e₂` (or similar comparisions), and then consider `e₂ -
e₁` as a termination argument.

Currently, this only fires when `e₁` and `e₂` only depend on the
functions parameters, but not local let-bindings or variables bound in
local pattern matches.

Duplicates are removed.

In the table showing the termination argument failures, long termination
arguments are now given a number and abbreviated as e.g. `#4` in the
table headers.

More examples in the test file, here as some highlights:
```
def distinct (xs : Array Nat) : Bool :=
  let rec loop (i j : Nat) : Bool :=
    if _ : i < xs.size then
      if _ : j < i then
        if xs[j] = xs[i] then
          false
        else
          loop i (j+1)
      else
        loop (i+1) 0
    else
      true
  loop 0 0
```
infers
```
termination_by (Array.size xs - i, i - j)
```
and the weird functions where `i` goes up or down
```
def weird (xs : Array Nat) (i : Nat) : Bool :=
  if _ : i < xs.size then
    if _ : 0 < i then
      if xs[i] = 42 then
        weird xs.pop (i - 1)
      else
        weird xs (i+1)
    else
      weird xs (i+1)
  else
    true
decreasing_by all_goals simp_wf; omega
```
infers
```
termination_by (Array.size xs - i, i)
```
but unfortunately needs `decreasing_by` pending the “big
decreasing_tactic refactor” that
I expect we’ll want to do at some point.
2024-03-16 12:27:35 +00:00