adam@372
|
1 (* Copyright (c) 2011-2012, Adam Chlipala
|
adam@324
|
2 *
|
adam@324
|
3 * This work is licensed under a
|
adam@324
|
4 * Creative Commons Attribution-Noncommercial-No Derivative Works 3.0
|
adam@324
|
5 * Unported License.
|
adam@324
|
6 * The license text is available at:
|
adam@324
|
7 * http://creativecommons.org/licenses/by-nc-nd/3.0/
|
adam@324
|
8 *)
|
adam@324
|
9
|
adam@324
|
10 (* begin hide *)
|
adam@324
|
11
|
adam@324
|
12 Require Import List.
|
adam@324
|
13
|
adam@324
|
14 Require Import CpdtTactics.
|
adam@324
|
15
|
adam@324
|
16 Set Implicit Arguments.
|
adam@324
|
17
|
adam@324
|
18 (* end hide *)
|
adam@324
|
19
|
adam@324
|
20 (** %\part{Proof Engineering}
|
adam@324
|
21
|
adam@324
|
22 \chapter{Proof Search by Logic Programming}% *)
|
adam@324
|
23
|
adam@430
|
24 (** The Curry-Howard correspondence tells us that proving is "just" programming, but the pragmatics of the two activities are very different. Generally we care about properties of a program besides its type, but the same is not true about proofs. Any proof of a theorem will do just as well. As a result, automated proof search is conceptually simpler than automated programming.
|
adam@372
|
25
|
adam@395
|
26 The paradigm of %\index{logic programming}%logic programming%~\cite{LogicProgramming}%, as embodied in languages like %\index{Prolog}%Prolog%~\cite{Prolog}%, is a good match for proof search in higher-order logic. This chapter introduces the details, attempting to avoid any dependence on past logic programming experience. *)
|
adam@324
|
27
|
adam@324
|
28
|
adam@324
|
29 (** * Introducing Logic Programming *)
|
adam@324
|
30
|
adam@372
|
31 (** Recall the definition of addition from the standard library. *)
|
adam@372
|
32
|
adam@324
|
33 Print plus.
|
adam@372
|
34 (** %\vspace{-.15in}%[[
|
adam@372
|
35 plus =
|
adam@372
|
36 fix plus (n m : nat) : nat := match n with
|
adam@372
|
37 | 0 => m
|
adam@372
|
38 | S p => S (plus p m)
|
adam@372
|
39 end
|
adam@372
|
40
|
adam@372
|
41 ]]
|
adam@372
|
42
|
adam@372
|
43 This is a recursive definition, in the style of functional programming. We might also follow the style of logic programming, which corresponds to the inductive relations we have defined in previous chapters. *)
|
adam@324
|
44
|
adam@324
|
45 Inductive plusR : nat -> nat -> nat -> Prop :=
|
adam@324
|
46 | PlusO : forall m, plusR O m m
|
adam@324
|
47 | PlusS : forall n m r, plusR n m r
|
adam@324
|
48 -> plusR (S n) m (S r).
|
adam@324
|
49
|
adam@372
|
50 (** Intuitively, a fact [plusR n m r] only holds when [plus n m = r]. It is not hard to prove this correspondence formally. *)
|
adam@372
|
51
|
adam@324
|
52 (* begin thide *)
|
adam@324
|
53 Hint Constructors plusR.
|
adam@324
|
54 (* end thide *)
|
adam@324
|
55
|
adam@324
|
56 Theorem plus_plusR : forall n m,
|
adam@324
|
57 plusR n m (n + m).
|
adam@324
|
58 (* begin thide *)
|
adam@324
|
59 induction n; crush.
|
adam@324
|
60 Qed.
|
adam@324
|
61 (* end thide *)
|
adam@324
|
62
|
adam@324
|
63 Theorem plusR_plus : forall n m r,
|
adam@324
|
64 plusR n m r
|
adam@324
|
65 -> r = n + m.
|
adam@324
|
66 (* begin thide *)
|
adam@324
|
67 induction 1; crush.
|
adam@324
|
68 Qed.
|
adam@324
|
69 (* end thide *)
|
adam@324
|
70
|
adam@372
|
71 (** With the functional definition of [plus], simple equalities about arithmetic follow by computation. *)
|
adam@372
|
72
|
adam@324
|
73 Example four_plus_three : 4 + 3 = 7.
|
adam@324
|
74 (* begin thide *)
|
adam@324
|
75 reflexivity.
|
adam@324
|
76 Qed.
|
adam@324
|
77 (* end thide *)
|
adam@324
|
78
|
adam@430
|
79 (* begin hide *)
|
adam@430
|
80 Definition er := @eq_refl.
|
adam@430
|
81 (* end hide *)
|
adam@430
|
82
|
adam@372
|
83 Print four_plus_three.
|
adam@372
|
84 (** %\vspace{-.15in}%[[
|
adam@372
|
85 four_plus_three = eq_refl
|
adam@372
|
86 ]]
|
adam@372
|
87
|
adam@372
|
88 With the relational definition, the same equalities take more steps to prove, but the process is completely mechanical. For example, consider this simple-minded manual proof search strategy. The steps with error messages shown afterward will be omitted from the final script.
|
adam@372
|
89 *)
|
adam@372
|
90
|
adam@324
|
91 Example four_plus_three' : plusR 4 3 7.
|
adam@324
|
92 (* begin thide *)
|
adam@372
|
93 (** %\vspace{-.2in}%[[
|
adam@372
|
94 apply PlusO.
|
adam@372
|
95 ]]
|
adam@372
|
96 %\vspace{-.2in}%
|
adam@372
|
97 <<
|
adam@372
|
98 Error: Impossible to unify "plusR 0 ?24 ?24" with "plusR 4 3 7".
|
adam@372
|
99 >> *)
|
adam@372
|
100 apply PlusS.
|
adam@372
|
101 (** %\vspace{-.2in}%[[
|
adam@372
|
102 apply PlusO.
|
adam@372
|
103 ]]
|
adam@372
|
104 %\vspace{-.2in}%
|
adam@372
|
105 <<
|
adam@372
|
106 Error: Impossible to unify "plusR 0 ?25 ?25" with "plusR 3 3 6".
|
adam@372
|
107 >> *)
|
adam@372
|
108 apply PlusS.
|
adam@372
|
109 (** %\vspace{-.2in}%[[
|
adam@372
|
110 apply PlusO.
|
adam@372
|
111 ]]
|
adam@372
|
112 %\vspace{-.2in}%
|
adam@372
|
113 <<
|
adam@372
|
114 Error: Impossible to unify "plusR 0 ?26 ?26" with "plusR 2 3 5".
|
adam@372
|
115 >> *)
|
adam@372
|
116 apply PlusS.
|
adam@372
|
117 (** %\vspace{-.2in}%[[
|
adam@372
|
118 apply PlusO.
|
adam@372
|
119 ]]
|
adam@372
|
120 %\vspace{-.2in}%
|
adam@372
|
121 <<
|
adam@372
|
122 Error: Impossible to unify "plusR 0 ?27 ?27" with "plusR 1 3 4".
|
adam@372
|
123 >> *)
|
adam@372
|
124 apply PlusS.
|
adam@372
|
125 apply PlusO.
|
adam@372
|
126
|
adam@372
|
127 (** At this point the proof is completed. It is no doubt clear that a simple procedure could find all proofs of this kind for us. We are just exploring all possible proof trees, built from the two candidate steps [apply PlusO] and [apply PlusS]. The built-in tactic %\index{tactics!auto}%[auto] does exactly that, since above we used [Hint Constructors] to register the two candidate proof steps as hints. *)
|
adam@372
|
128
|
adam@372
|
129 Restart.
|
adam@324
|
130 auto.
|
adam@324
|
131 Qed.
|
adam@324
|
132 (* end thide *)
|
adam@324
|
133
|
adam@372
|
134 Print four_plus_three'.
|
adam@372
|
135 (** %\vspace{-.15in}%[[
|
adam@372
|
136 four_plus_three' = PlusS (PlusS (PlusS (PlusS (PlusO 3))))
|
adam@372
|
137 ]]
|
adam@372
|
138 *)
|
adam@372
|
139
|
adam@372
|
140 (** Let us try the same approach on a slightly more complex goal. *)
|
adam@372
|
141
|
adam@372
|
142 Example five_plus_three : plusR 5 3 8.
|
adam@324
|
143 (* begin thide *)
|
adam@372
|
144 auto.
|
adam@372
|
145
|
adam@372
|
146 (** This time, [auto] is not enough to make any progress. Since even a single candidate step may lead to an infinite space of possible proof trees, [auto] is parameterized on the maximum depth of trees to consider. The default depth is 5, and it turns out that we need depth 6 to prove the goal. *)
|
adam@372
|
147
|
adam@324
|
148 auto 6.
|
adam@372
|
149
|
adam@410
|
150 (** Sometimes it is useful to see a description of the proof tree that [auto] finds, with the %\index{tactics!info}%[info] tactical. (This tactical is not available in Coq 8.4 as of this writing, but I hope it reappears soon.) *)
|
adam@372
|
151
|
adam@324
|
152 Restart.
|
adam@324
|
153 info auto 6.
|
adam@372
|
154 (** %\vspace{-.15in}%[[
|
adam@372
|
155 == apply PlusS; apply PlusS; apply PlusS; apply PlusS;
|
adam@372
|
156 apply PlusS; apply PlusO.
|
adam@372
|
157 ]]
|
adam@372
|
158 *)
|
adam@324
|
159 Qed.
|
adam@324
|
160 (* end thide *)
|
adam@324
|
161
|
adam@410
|
162 (** The two key components of logic programming are%\index{backtracking}% _backtracking_ and%\index{unification}% _unification_. To see these techniques in action, consider this further silly example. Here our candidate proof steps will be reflexivity and quantifier instantiation. *)
|
adam@324
|
163
|
adam@324
|
164 Example seven_minus_three : exists x, x + 3 = 7.
|
adam@324
|
165 (* begin thide *)
|
adam@372
|
166 (** For explanatory purposes, let us simulate a user with minimal understanding of arithmetic. We start by choosing an instantiation for the quantifier. Recall that [ex_intro] is the constructor for existentially quantified formulas. *)
|
adam@372
|
167
|
adam@372
|
168 apply ex_intro with 0.
|
adam@372
|
169 (** %\vspace{-.2in}%[[
|
adam@372
|
170 reflexivity.
|
adam@372
|
171 ]]
|
adam@372
|
172 %\vspace{-.2in}%
|
adam@372
|
173 <<
|
adam@372
|
174 Error: Impossible to unify "7" with "0 + 3".
|
adam@372
|
175 >>
|
adam@372
|
176
|
adam@398
|
177 This seems to be a dead end. Let us _backtrack_ to the point where we ran [apply] and make a better alternate choice.
|
adam@372
|
178 *)
|
adam@372
|
179
|
adam@372
|
180 Restart.
|
adam@372
|
181 apply ex_intro with 4.
|
adam@372
|
182 reflexivity.
|
adam@372
|
183 Qed.
|
adam@324
|
184 (* end thide *)
|
adam@324
|
185
|
adam@372
|
186 (** The above was a fairly tame example of backtracking. In general, any node in an under-construction proof tree may be the destination of backtracking an arbitrarily large number of times, as different candidate proof steps are found not to lead to full proof trees, within the depth bound passed to [auto].
|
adam@372
|
187
|
adam@372
|
188 Next we demonstrate unification, which will be easier when we switch to the relational formulation of addition. *)
|
adam@372
|
189
|
adam@324
|
190 Example seven_minus_three' : exists x, plusR x 3 7.
|
adam@324
|
191 (* begin thide *)
|
adam@410
|
192 (** We could attempt to guess the quantifier instantiation manually as before, but here there is no need. Instead of [apply], we use %\index{tactics!eapply}%[eapply] instead, which proceeds with placeholder%\index{unification variable}% _unification variables_ standing in for those parameters we wish to postpone guessing. *)
|
adam@372
|
193
|
adam@372
|
194 eapply ex_intro.
|
adam@372
|
195 (** [[
|
adam@372
|
196 1 subgoal
|
adam@372
|
197
|
adam@372
|
198 ============================
|
adam@372
|
199 plusR ?70 3 7
|
adam@372
|
200 ]]
|
adam@372
|
201
|
adam@372
|
202 Now we can finish the proof with the right applications of [plusR]'s constructors. Note that new unification variables are being generated to stand for new unknowns. *)
|
adam@372
|
203
|
adam@372
|
204 apply PlusS.
|
adam@372
|
205 (** [[
|
adam@372
|
206 ============================
|
adam@372
|
207 plusR ?71 3 6
|
adam@372
|
208 ]]
|
adam@372
|
209 *)
|
adam@372
|
210 apply PlusS. apply PlusS. apply PlusS.
|
adam@372
|
211 (** [[
|
adam@372
|
212 ============================
|
adam@372
|
213 plusR ?74 3 3
|
adam@372
|
214 ]]
|
adam@372
|
215 *)
|
adam@372
|
216 apply PlusO.
|
adam@372
|
217
|
adam@372
|
218 (** The [auto] tactic will not perform these sorts of steps that introduce unification variables, but the %\index{tactics!eauto}%[eauto] tactic will. It is helpful to work with two separate tactics, because proof search in the [eauto] style can uncover many more potential proof trees and hence take much longer to run. *)
|
adam@372
|
219
|
adam@372
|
220 Restart.
|
adam@324
|
221 info eauto 6.
|
adam@372
|
222 (** %\vspace{-.15in}%[[
|
adam@372
|
223 == eapply ex_intro; apply PlusS; apply PlusS;
|
adam@372
|
224 apply PlusS; apply PlusS; apply PlusO.
|
adam@372
|
225 ]]
|
adam@372
|
226 *)
|
adam@324
|
227 Qed.
|
adam@324
|
228 (* end thide *)
|
adam@324
|
229
|
adam@372
|
230 (** This proof gives us our first example where logic programming simplifies proof search compared to functional programming. In general, functional programs are only meant to be run in a single direction; a function has disjoint sets of inputs and outputs. In the last example, we effectively ran a logic program backwards, deducing an input that gives rise to a certain output. The same works for deducing an unknown value of the other input. *)
|
adam@372
|
231
|
adam@324
|
232 Example seven_minus_four' : exists x, plusR 4 x 7.
|
adam@324
|
233 (* begin thide *)
|
adam@372
|
234 eauto 6.
|
adam@324
|
235 Qed.
|
adam@324
|
236 (* end thide *)
|
adam@324
|
237
|
adam@372
|
238 (** By proving the right auxiliary facts, we can reason about specific functional programs in the same way as we did above for a logic program. Let us prove that the constructors of [plusR] have natural interpretations as lemmas about [plus]. We can find the first such lemma already proved in the standard library, using the %\index{Vernacular commands!SearchRewrite}%[SearchRewrite] command to find a library function proving an equality whose lefthand or righthand side matches a pattern with wildcards. *)
|
adam@372
|
239
|
adam@324
|
240 (* begin thide *)
|
adam@324
|
241 SearchRewrite (O + _).
|
adam@372
|
242 (** %\vspace{-.15in}%[[
|
adam@372
|
243 plus_O_n: forall n : nat, 0 + n = n
|
adam@372
|
244 ]]
|
adam@372
|
245
|
adam@372
|
246 The command %\index{Vernacular commands!Hint Immediate}%[Hint Immediate] asks [auto] and [eauto] to consider this lemma as a candidate step for any leaf of a proof tree. *)
|
adam@324
|
247
|
adam@324
|
248 Hint Immediate plus_O_n.
|
adam@324
|
249
|
adam@372
|
250 (** The counterpart to [PlusS] we will prove ourselves. *)
|
adam@372
|
251
|
adam@324
|
252 Lemma plusS : forall n m r,
|
adam@324
|
253 n + m = r
|
adam@324
|
254 -> S n + m = S r.
|
adam@324
|
255 crush.
|
adam@324
|
256 Qed.
|
adam@324
|
257
|
adam@372
|
258 (** The command %\index{Vernacular commands!Hint Resolve}%[Hint Resolve] adds a new candidate proof step, to be attempted at any level of a proof tree, not just at leaves. *)
|
adam@372
|
259
|
adam@324
|
260 Hint Resolve plusS.
|
adam@324
|
261 (* end thide *)
|
adam@324
|
262
|
adam@372
|
263 (** Now that we have registered the proper hints, we can replicate our previous examples with the normal, functional addition [plus]. *)
|
adam@372
|
264
|
adam@372
|
265 Example seven_minus_three'' : exists x, x + 3 = 7.
|
adam@324
|
266 (* begin thide *)
|
adam@372
|
267 eauto 6.
|
adam@324
|
268 Qed.
|
adam@324
|
269 (* end thide *)
|
adam@324
|
270
|
adam@324
|
271 Example seven_minus_four : exists x, 4 + x = 7.
|
adam@324
|
272 (* begin thide *)
|
adam@372
|
273 eauto 6.
|
adam@324
|
274 Qed.
|
adam@324
|
275 (* end thide *)
|
adam@324
|
276
|
adam@372
|
277 (** This new hint database is far from a complete decision procedure, as we see in a further example that [eauto] does not finish. *)
|
adam@372
|
278
|
adam@372
|
279 Example seven_minus_four_zero : exists x, 4 + x + 0 = 7.
|
adam@324
|
280 (* begin thide *)
|
adam@324
|
281 eauto 6.
|
adam@324
|
282 Abort.
|
adam@324
|
283 (* end thide *)
|
adam@324
|
284
|
adam@372
|
285 (** A further lemma will be helpful. *)
|
adam@372
|
286
|
adam@324
|
287 (* begin thide *)
|
adam@324
|
288 Lemma plusO : forall n m,
|
adam@324
|
289 n = m
|
adam@324
|
290 -> n + 0 = m.
|
adam@324
|
291 crush.
|
adam@324
|
292 Qed.
|
adam@324
|
293
|
adam@324
|
294 Hint Resolve plusO.
|
adam@324
|
295 (* end thide *)
|
adam@324
|
296
|
adam@372
|
297 (** Note that, if we consider the inputs to [plus] as the inputs of a corresponding logic program, the new rule [plusO] introduces an ambiguity. For instance, a sum [0 + 0] would match both of [plus_O_n] and [plusO], depending on which operand we focus on. This ambiguity may increase the number of potential search trees, slowing proof search, but semantically it presents no problems, and in fact it leads to an automated proof of the present example. *)
|
adam@372
|
298
|
adam@324
|
299 Example seven_minus_four_zero : exists x, 4 + x + 0 = 7.
|
adam@324
|
300 (* begin thide *)
|
adam@372
|
301 eauto 7.
|
adam@324
|
302 Qed.
|
adam@324
|
303 (* end thide *)
|
adam@324
|
304
|
adam@372
|
305 (** Just how much damage can be done by adding hints that grow the space of possible proof trees? A classic gotcha comes from unrestricted use of transitivity, as embodied in this library theorem about equality: *)
|
adam@372
|
306
|
adam@324
|
307 Check eq_trans.
|
adam@372
|
308 (** %\vspace{-.15in}%[[
|
adam@372
|
309 eq_trans
|
adam@372
|
310 : forall (A : Type) (x y z : A), x = y -> y = z -> x = z
|
adam@372
|
311 ]]
|
adam@372
|
312 *)
|
adam@372
|
313
|
adam@372
|
314 (** Hints are scoped over sections, so let us enter a section to contain the effects of an unfortunate hint choice. *)
|
adam@324
|
315
|
adam@324
|
316 Section slow.
|
adam@324
|
317 Hint Resolve eq_trans.
|
adam@324
|
318
|
adam@372
|
319 (** The following fact is false, but that does not stop [eauto] from taking a very long time to search for proofs of it. We use the handy %\index{Vernacular commands!Time}%[Time] command to measure how long a proof step takes to run. None of the following steps make any progress. *)
|
adam@372
|
320
|
adam@324
|
321 Example three_minus_four_zero : exists x, 1 + x = 0.
|
adam@324
|
322 Time eauto 1.
|
adam@372
|
323 (** %\vspace{-.15in}%
|
adam@372
|
324 <<
|
adam@372
|
325 Finished transaction in 0. secs (0.u,0.s)
|
adam@372
|
326 >>
|
adam@372
|
327 *)
|
adam@372
|
328
|
adam@324
|
329 Time eauto 2.
|
adam@372
|
330 (** %\vspace{-.15in}%
|
adam@372
|
331 <<
|
adam@372
|
332 Finished transaction in 0. secs (0.u,0.s)
|
adam@372
|
333 >>
|
adam@372
|
334 *)
|
adam@372
|
335
|
adam@324
|
336 Time eauto 3.
|
adam@372
|
337 (** %\vspace{-.15in}%
|
adam@372
|
338 <<
|
adam@372
|
339 Finished transaction in 0. secs (0.008u,0.s)
|
adam@372
|
340 >>
|
adam@372
|
341 *)
|
adam@372
|
342
|
adam@324
|
343 Time eauto 4.
|
adam@372
|
344 (** %\vspace{-.15in}%
|
adam@372
|
345 <<
|
adam@372
|
346 Finished transaction in 0. secs (0.068005u,0.004s)
|
adam@372
|
347 >>
|
adam@372
|
348 *)
|
adam@372
|
349
|
adam@324
|
350 Time eauto 5.
|
adam@372
|
351 (** %\vspace{-.15in}%
|
adam@372
|
352 <<
|
adam@372
|
353 Finished transaction in 2. secs (1.92012u,0.044003s)
|
adam@372
|
354 >>
|
adam@372
|
355 *)
|
adam@372
|
356
|
adam@372
|
357 (** We see worrying exponential growth in running time, and the %\index{tactics!debug}%[debug] tactical helps us see where [eauto] is wasting its time, outputting a trace of every proof step that is attempted. The rule [eq_trans] applies at every node of a proof tree, and [eauto] tries all such positions. *)
|
adam@372
|
358
|
adam@324
|
359 debug eauto 3.
|
adam@372
|
360 (** [[
|
adam@372
|
361 1 depth=3
|
adam@372
|
362 1.1 depth=2 eapply ex_intro
|
adam@372
|
363 1.1.1 depth=1 apply plusO
|
adam@372
|
364 1.1.1.1 depth=0 eapply eq_trans
|
adam@372
|
365 1.1.2 depth=1 eapply eq_trans
|
adam@372
|
366 1.1.2.1 depth=1 apply plus_n_O
|
adam@372
|
367 1.1.2.1.1 depth=0 apply plusO
|
adam@372
|
368 1.1.2.1.2 depth=0 eapply eq_trans
|
adam@372
|
369 1.1.2.2 depth=1 apply @eq_refl
|
adam@372
|
370 1.1.2.2.1 depth=0 apply plusO
|
adam@372
|
371 1.1.2.2.2 depth=0 eapply eq_trans
|
adam@372
|
372 1.1.2.3 depth=1 apply eq_add_S ; trivial
|
adam@372
|
373 1.1.2.3.1 depth=0 apply plusO
|
adam@372
|
374 1.1.2.3.2 depth=0 eapply eq_trans
|
adam@372
|
375 1.1.2.4 depth=1 apply eq_sym ; trivial
|
adam@372
|
376 1.1.2.4.1 depth=0 eapply eq_trans
|
adam@372
|
377 1.1.2.5 depth=0 apply plusO
|
adam@372
|
378 1.1.2.6 depth=0 apply plusS
|
adam@372
|
379 1.1.2.7 depth=0 apply f_equal (A:=nat)
|
adam@372
|
380 1.1.2.8 depth=0 apply f_equal2 (A1:=nat) (A2:=nat)
|
adam@372
|
381 1.1.2.9 depth=0 eapply eq_trans
|
adam@372
|
382 ]]
|
adam@372
|
383 *)
|
adam@324
|
384 Abort.
|
adam@324
|
385 End slow.
|
adam@324
|
386
|
adam@410
|
387 (** Sometimes, though, transitivity is just what is needed to get a proof to go through automatically with [eauto]. For those cases, we can use named%\index{hint databases}% _hint databases_ to segragate hints into different groups that may be called on as needed. Here we put [eq_trans] into the database [slow]. *)
|
adam@372
|
388
|
adam@324
|
389 (* begin thide *)
|
adam@324
|
390 Hint Resolve eq_trans : slow.
|
adam@324
|
391 (* end thide *)
|
adam@324
|
392
|
adam@324
|
393 Example three_minus_four_zero : exists x, 1 + x = 0.
|
adam@324
|
394 (* begin thide *)
|
adam@372
|
395 Time eauto.
|
adam@372
|
396 (** %\vspace{-.15in}%
|
adam@372
|
397 <<
|
adam@372
|
398 Finished transaction in 0. secs (0.004u,0.s)
|
adam@372
|
399 >>
|
adam@372
|
400
|
adam@372
|
401 This [eauto] fails to prove the goal, but it least it takes substantially less than the 2 seconds required above! *)
|
adam@372
|
402
|
adam@324
|
403 Abort.
|
adam@324
|
404 (* end thide *)
|
adam@324
|
405
|
adam@372
|
406 (** One simple example from before runs in the same amount of time, avoiding pollution by transivity. *)
|
adam@372
|
407
|
adam@324
|
408 Example seven_minus_three_again : exists x, x + 3 = 7.
|
adam@324
|
409 (* begin thide *)
|
adam@372
|
410 Time eauto 6.
|
adam@372
|
411 (** %\vspace{-.15in}%
|
adam@372
|
412 <<
|
adam@372
|
413 Finished transaction in 0. secs (0.004001u,0.s)
|
adam@372
|
414 >>
|
adam@372
|
415 %\vspace{-.2in}% *)
|
adam@372
|
416
|
adam@324
|
417 Qed.
|
adam@324
|
418 (* end thide *)
|
adam@324
|
419
|
adam@398
|
420 (** When we _do_ need transitivity, we ask for it explicitly. *)
|
adam@372
|
421
|
adam@324
|
422 Example needs_trans : forall x y, 1 + x = y
|
adam@324
|
423 -> y = 2
|
adam@324
|
424 -> exists z, z + x = 3.
|
adam@324
|
425 (* begin thide *)
|
adam@324
|
426 info eauto with slow.
|
adam@372
|
427 (** %\vspace{-.2in}%[[
|
adam@372
|
428 == intro x; intro y; intro H; intro H0; simple eapply ex_intro;
|
adam@372
|
429 apply plusS; simple eapply eq_trans.
|
adam@372
|
430 exact H.
|
adam@372
|
431
|
adam@372
|
432 exact H0.
|
adam@372
|
433 ]]
|
adam@372
|
434 *)
|
adam@324
|
435 Qed.
|
adam@324
|
436 (* end thide *)
|
adam@324
|
437
|
adam@372
|
438 (** The [info] trace shows that [eq_trans] was used in just the position where it is needed to complete the proof. We also see that [auto] and [eauto] always perform [intro] steps without counting them toward the bound on proof tree depth. *)
|
adam@372
|
439
|
adam@324
|
440
|
adam@324
|
441 (** * Searching for Underconstrained Values *)
|
adam@324
|
442
|
adam@373
|
443 (** Recall the definition of the list length function. *)
|
adam@373
|
444
|
adam@324
|
445 Print length.
|
adam@373
|
446 (** %\vspace{-.15in}%[[
|
adam@373
|
447 length =
|
adam@373
|
448 fun A : Type =>
|
adam@373
|
449 fix length (l : list A) : nat :=
|
adam@373
|
450 match l with
|
adam@373
|
451 | nil => 0
|
adam@373
|
452 | _ :: l' => S (length l')
|
adam@373
|
453 end
|
adam@373
|
454 ]]
|
adam@373
|
455
|
adam@373
|
456 This function is easy to reason about in the forward direction, computing output from input. *)
|
adam@324
|
457
|
adam@324
|
458 Example length_1_2 : length (1 :: 2 :: nil) = 2.
|
adam@324
|
459 auto.
|
adam@324
|
460 Qed.
|
adam@324
|
461
|
adam@324
|
462 Print length_1_2.
|
adam@373
|
463 (** %\vspace{-.15in}%[[
|
adam@373
|
464 length_1_2 = eq_refl
|
adam@373
|
465 ]]
|
adam@373
|
466
|
adam@373
|
467 As in the last section, we will prove some lemmas to recast [length] in logic programming style, to help us compute inputs from outputs. *)
|
adam@324
|
468
|
adam@324
|
469 (* begin thide *)
|
adam@324
|
470 Theorem length_O : forall A, length (nil (A := A)) = O.
|
adam@324
|
471 crush.
|
adam@324
|
472 Qed.
|
adam@324
|
473
|
adam@324
|
474 Theorem length_S : forall A (h : A) t n,
|
adam@324
|
475 length t = n
|
adam@324
|
476 -> length (h :: t) = S n.
|
adam@324
|
477 crush.
|
adam@324
|
478 Qed.
|
adam@324
|
479
|
adam@324
|
480 Hint Resolve length_O length_S.
|
adam@324
|
481 (* end thide *)
|
adam@324
|
482
|
adam@373
|
483 (** Let us apply these hints to prove that a [list nat] of length 2 exists. *)
|
adam@373
|
484
|
adam@324
|
485 Example length_is_2 : exists ls : list nat, length ls = 2.
|
adam@324
|
486 (* begin thide *)
|
adam@324
|
487 eauto.
|
adam@373
|
488 (** <<
|
adam@373
|
489 No more subgoals but non-instantiated existential variables:
|
adam@373
|
490 Existential 1 = ?20249 : [ |- nat]
|
adam@373
|
491 Existential 2 = ?20252 : [ |- nat]
|
adam@373
|
492 >>
|
adam@373
|
493
|
adam@398
|
494 Coq complains that we finished the proof without determining the values of some unification variables created during proof search. The error message may seem a bit silly, since _any_ value of type [nat] (for instance, 0) can be plugged in for either variable! However, for more complex types, finding their inhabitants may be as complex as theorem-proving in general.
|
adam@373
|
495
|
adam@373
|
496 The %\index{Vernacular commands!Show Proof}%[Show Proof] command shows exactly which proof term [eauto] has found, with the undetermined unification variables appearing explicitly where they are used. *)
|
adam@324
|
497
|
adam@324
|
498 Show Proof.
|
adam@373
|
499 (** <<
|
adam@373
|
500 Proof: ex_intro (fun ls : list nat => length ls = 2)
|
adam@373
|
501 (?20249 :: ?20252 :: nil)
|
adam@373
|
502 (length_S ?20249 (?20252 :: nil)
|
adam@373
|
503 (length_S ?20252 nil (length_O nat)))
|
adam@373
|
504 >>
|
adam@373
|
505 %\vspace{-.2in}% *)
|
adam@373
|
506
|
adam@324
|
507 Abort.
|
adam@324
|
508 (* end thide *)
|
adam@324
|
509
|
adam@373
|
510 (** We see that the two unification variables stand for the two elements of the list. Indeed, list length is independent of data values. Paradoxically, we can make the proof search process easier by constraining the list further, so that proof search naturally locates appropriate data elements by unification. The library predicate [Forall] will be helpful. *)
|
adam@373
|
511
|
adam@430
|
512 (* begin hide *)
|
adam@430
|
513 Definition Forall_c := (@Forall_nil, @Forall_cons).
|
adam@430
|
514 (* end hide *)
|
adam@430
|
515
|
adam@324
|
516 Print Forall.
|
adam@373
|
517 (** %\vspace{-.15in}%[[
|
adam@373
|
518 Inductive Forall (A : Type) (P : A -> Prop) : list A -> Prop :=
|
adam@373
|
519 Forall_nil : Forall P nil
|
adam@373
|
520 | Forall_cons : forall (x : A) (l : list A),
|
adam@373
|
521 P x -> Forall P l -> Forall P (x :: l)
|
adam@373
|
522 ]]
|
adam@373
|
523 *)
|
adam@324
|
524
|
adam@324
|
525 Example length_is_2 : exists ls : list nat, length ls = 2
|
adam@324
|
526 /\ Forall (fun n => n >= 1) ls.
|
adam@324
|
527 (* begin thide *)
|
adam@324
|
528 eauto 9.
|
adam@324
|
529 Qed.
|
adam@324
|
530 (* end thide *)
|
adam@324
|
531
|
adam@373
|
532 (** We can see which list [eauto] found by printing the proof term. *)
|
adam@373
|
533
|
adam@430
|
534 (* begin hide *)
|
adam@430
|
535 Definition conj' := (conj, le_n).
|
adam@430
|
536 (* end hide *)
|
adam@430
|
537
|
adam@373
|
538 Print length_is_2.
|
adam@373
|
539 (** %\vspace{-.15in}%[[
|
adam@373
|
540 length_is_2 =
|
adam@373
|
541 ex_intro
|
adam@373
|
542 (fun ls : list nat => length ls = 2 /\ Forall (fun n : nat => n >= 1) ls)
|
adam@373
|
543 (1 :: 1 :: nil)
|
adam@373
|
544 (conj (length_S 1 (1 :: nil) (length_S 1 nil (length_O nat)))
|
adam@373
|
545 (Forall_cons 1 (le_n 1)
|
adam@373
|
546 (Forall_cons 1 (le_n 1) (Forall_nil (fun n : nat => n >= 1)))))
|
adam@373
|
547 ]]
|
adam@373
|
548 *)
|
adam@373
|
549
|
adam@373
|
550 (** Let us try one more, fancier example. First, we use a standard high-order function to define a function for summing all data elements of a list. *)
|
adam@373
|
551
|
adam@324
|
552 Definition sum := fold_right plus O.
|
adam@324
|
553
|
adam@373
|
554 (** Another basic lemma will be helpful to guide proof search. *)
|
adam@373
|
555
|
adam@324
|
556 (* begin thide *)
|
adam@324
|
557 Lemma plusO' : forall n m,
|
adam@324
|
558 n = m
|
adam@324
|
559 -> 0 + n = m.
|
adam@324
|
560 crush.
|
adam@324
|
561 Qed.
|
adam@324
|
562
|
adam@324
|
563 Hint Resolve plusO'.
|
adam@324
|
564
|
adam@373
|
565 (** Finally, we meet %\index{Vernacular commands!Hint Extern}%[Hint Extern], the command to register a custom hint. That is, we provide a pattern to match against goals during proof search. Whenever the pattern matches, a tactic (given to the right of an arrow [=>]) is attempted. Below, the number [1] gives a priority for this step. Lower priorities are tried before higher priorities, which can have a significant effect on proof search time. *)
|
adam@373
|
566
|
adam@324
|
567 Hint Extern 1 (sum _ = _) => simpl.
|
adam@324
|
568 (* end thide *)
|
adam@324
|
569
|
adam@373
|
570 (** Now we can find a length-2 list whose sum is 0. *)
|
adam@373
|
571
|
adam@324
|
572 Example length_and_sum : exists ls : list nat, length ls = 2
|
adam@324
|
573 /\ sum ls = O.
|
adam@324
|
574 (* begin thide *)
|
adam@324
|
575 eauto 7.
|
adam@324
|
576 Qed.
|
adam@324
|
577 (* end thide *)
|
adam@324
|
578
|
adam@373
|
579 (* begin hide *)
|
adam@324
|
580 Print length_and_sum.
|
adam@373
|
581 (* end hide *)
|
adam@373
|
582
|
adam@373
|
583 (** Printing the proof term shows the unsurprising list that is found. Here is an example where it is less obvious which list will be used. Can you guess which list [eauto] will choose? *)
|
adam@324
|
584
|
adam@324
|
585 Example length_and_sum' : exists ls : list nat, length ls = 5
|
adam@324
|
586 /\ sum ls = 42.
|
adam@324
|
587 (* begin thide *)
|
adam@324
|
588 eauto 15.
|
adam@324
|
589 Qed.
|
adam@324
|
590 (* end thide *)
|
adam@324
|
591
|
adam@373
|
592 (* begin hide *)
|
adam@324
|
593 Print length_and_sum'.
|
adam@373
|
594 (* end hide *)
|
adam@373
|
595
|
adam@373
|
596 (** We will give away part of the answer and say that the above list is less interesting than we would like, because it contains too many zeroes. A further constraint forces a different solution for a smaller instance of the problem. *)
|
adam@324
|
597
|
adam@324
|
598 Example length_and_sum'' : exists ls : list nat, length ls = 2
|
adam@324
|
599 /\ sum ls = 3
|
adam@324
|
600 /\ Forall (fun n => n <> 0) ls.
|
adam@324
|
601 (* begin thide *)
|
adam@324
|
602 eauto 11.
|
adam@324
|
603 Qed.
|
adam@324
|
604 (* end thide *)
|
adam@324
|
605
|
adam@373
|
606 (* begin hide *)
|
adam@324
|
607 Print length_and_sum''.
|
adam@373
|
608 (* end hide *)
|
adam@373
|
609
|
adam@398
|
610 (** We could continue through exercises of this kind, but even more interesting than finding lists automatically is finding _programs_ automatically. *)
|
adam@324
|
611
|
adam@324
|
612
|
adam@324
|
613 (** * Synthesizing Programs *)
|
adam@324
|
614
|
adam@374
|
615 (** Here is a simple syntax type for arithmetic expressions, similar to those we have used several times before in the book. In this case, we allow expressions to mention exactly one distinguished variable. *)
|
adam@374
|
616
|
adam@324
|
617 Inductive exp : Set :=
|
adam@324
|
618 | Const : nat -> exp
|
adam@324
|
619 | Var : exp
|
adam@324
|
620 | Plus : exp -> exp -> exp.
|
adam@324
|
621
|
adam@374
|
622 (** An inductive relation specifies the semantics of an expression, relating a variable value and an expression to the expression value. *)
|
adam@374
|
623
|
adam@324
|
624 Inductive eval (var : nat) : exp -> nat -> Prop :=
|
adam@324
|
625 | EvalConst : forall n, eval var (Const n) n
|
adam@324
|
626 | EvalVar : eval var Var var
|
adam@324
|
627 | EvalPlus : forall e1 e2 n1 n2, eval var e1 n1
|
adam@324
|
628 -> eval var e2 n2
|
adam@324
|
629 -> eval var (Plus e1 e2) (n1 + n2).
|
adam@324
|
630
|
adam@324
|
631 (* begin thide *)
|
adam@324
|
632 Hint Constructors eval.
|
adam@324
|
633 (* end thide *)
|
adam@324
|
634
|
adam@374
|
635 (** We can use [auto] to execute the semantics for specific expressions. *)
|
adam@374
|
636
|
adam@324
|
637 Example eval1 : forall var, eval var (Plus Var (Plus (Const 8) Var)) (var + (8 + var)).
|
adam@324
|
638 (* begin thide *)
|
adam@324
|
639 auto.
|
adam@324
|
640 Qed.
|
adam@324
|
641 (* end thide *)
|
adam@324
|
642
|
adam@374
|
643 (** Unfortunately, just the constructors of [eval] are not enough to prove theorems like the following, which depends on an arithmetic identity. *)
|
adam@374
|
644
|
adam@324
|
645 Example eval1' : forall var, eval var (Plus Var (Plus (Const 8) Var)) (2 * var + 8).
|
adam@324
|
646 (* begin thide *)
|
adam@324
|
647 eauto.
|
adam@324
|
648 Abort.
|
adam@324
|
649 (* end thide *)
|
adam@324
|
650
|
adam@374
|
651 (** To help prove [eval1'], we prove an alternate version of [EvalPlus] that inserts an extra equality premise. *)
|
adam@374
|
652
|
adam@324
|
653 (* begin thide *)
|
adam@324
|
654 Theorem EvalPlus' : forall var e1 e2 n1 n2 n, eval var e1 n1
|
adam@324
|
655 -> eval var e2 n2
|
adam@324
|
656 -> n1 + n2 = n
|
adam@324
|
657 -> eval var (Plus e1 e2) n.
|
adam@324
|
658 crush.
|
adam@324
|
659 Qed.
|
adam@324
|
660
|
adam@324
|
661 Hint Resolve EvalPlus'.
|
adam@324
|
662
|
adam@374
|
663 (** Further, we instruct [eauto] to apply %\index{tactics!omega}%[omega], a standard tactic that provides a complete decision procedure for quantifier-free linear arithmetic. Via [Hint Extern], we ask for use of [omega] on any equality goal. The [abstract] tactical generates a new lemma for every such successful proof, so that, in the final proof term, the lemma may be referenced in place of dropping in the full proof of the arithmetic equality. *)
|
adam@374
|
664
|
adam@324
|
665 Hint Extern 1 (_ = _) => abstract omega.
|
adam@324
|
666 (* end thide *)
|
adam@324
|
667
|
adam@374
|
668 (** Now we can return to [eval1'] and prove it automatically. *)
|
adam@374
|
669
|
adam@324
|
670 Example eval1' : forall var, eval var (Plus Var (Plus (Const 8) Var)) (2 * var + 8).
|
adam@324
|
671 (* begin thide *)
|
adam@324
|
672 eauto.
|
adam@324
|
673 Qed.
|
adam@324
|
674 (* end thide *)
|
adam@324
|
675
|
adam@430
|
676 (* begin hide *)
|
adam@430
|
677 Definition e1s := eval1'_subproof.
|
adam@430
|
678 (* end hide *)
|
adam@430
|
679
|
adam@324
|
680 Print eval1'.
|
adam@374
|
681 (** %\vspace{-.15in}%[[
|
adam@374
|
682 eval1' =
|
adam@374
|
683 fun var : nat =>
|
adam@374
|
684 EvalPlus' (EvalVar var) (EvalPlus (EvalConst var 8) (EvalVar var))
|
adam@374
|
685 (eval1'_subproof var)
|
adam@374
|
686 : forall var : nat,
|
adam@374
|
687 eval var (Plus Var (Plus (Const 8) Var)) (2 * var + 8)
|
adam@374
|
688 ]]
|
adam@374
|
689 *)
|
adam@374
|
690
|
adam@374
|
691 (** The lemma [eval1'_subproof] was generated by [abstract omega].
|
adam@374
|
692
|
adam@374
|
693 Now we are ready to take advantage of logic programming's flexibility by searching for a program (arithmetic expression) that always evaluates to a particular symbolic value. *)
|
adam@324
|
694
|
adam@324
|
695 Example synthesize1 : exists e, forall var, eval var e (var + 7).
|
adam@324
|
696 (* begin thide *)
|
adam@324
|
697 eauto.
|
adam@324
|
698 Qed.
|
adam@324
|
699 (* end thide *)
|
adam@324
|
700
|
adam@324
|
701 Print synthesize1.
|
adam@374
|
702 (** %\vspace{-.15in}%[[
|
adam@374
|
703 synthesize1 =
|
adam@374
|
704 ex_intro (fun e : exp => forall var : nat, eval var e (var + 7))
|
adam@374
|
705 (Plus Var (Const 7))
|
adam@374
|
706 (fun var : nat => EvalPlus (EvalVar var) (EvalConst var 7))
|
adam@374
|
707 ]]
|
adam@374
|
708 *)
|
adam@374
|
709
|
adam@374
|
710 (** Here are two more examples showing off our program synthesis abilities. *)
|
adam@324
|
711
|
adam@324
|
712 Example synthesize2 : exists e, forall var, eval var e (2 * var + 8).
|
adam@324
|
713 (* begin thide *)
|
adam@324
|
714 eauto.
|
adam@324
|
715 Qed.
|
adam@324
|
716 (* end thide *)
|
adam@324
|
717
|
adam@374
|
718 (* begin hide *)
|
adam@324
|
719 Print synthesize2.
|
adam@374
|
720 (* end hide *)
|
adam@324
|
721
|
adam@324
|
722 Example synthesize3 : exists e, forall var, eval var e (3 * var + 42).
|
adam@324
|
723 (* begin thide *)
|
adam@324
|
724 eauto.
|
adam@324
|
725 Qed.
|
adam@324
|
726 (* end thide *)
|
adam@324
|
727
|
adam@374
|
728 (* begin hide *)
|
adam@324
|
729 Print synthesize3.
|
adam@374
|
730 (* end hide *)
|
adam@374
|
731
|
adam@374
|
732 (** These examples show linear expressions over the variable [var]. Any such expression is equivalent to [k * var + n] for some [k] and [n]. It is probably not so surprising that we can prove that any expression's semantics is equivalent to some such linear expression, but it is tedious to prove such a fact manually. To finish this section, we will use [eauto] to complete the proof, finding [k] and [n] values automatically.
|
adam@374
|
733
|
adam@374
|
734 We prove a series of lemmas and add them as hints. We have alternate [eval] constructor lemmas and some facts about arithmetic. *)
|
adam@324
|
735
|
adam@324
|
736 (* begin thide *)
|
adam@324
|
737 Theorem EvalConst' : forall var n m, n = m
|
adam@324
|
738 -> eval var (Const n) m.
|
adam@324
|
739 crush.
|
adam@324
|
740 Qed.
|
adam@324
|
741
|
adam@324
|
742 Hint Resolve EvalConst'.
|
adam@324
|
743
|
adam@324
|
744 Theorem zero_times : forall n m r,
|
adam@324
|
745 r = m
|
adam@324
|
746 -> r = 0 * n + m.
|
adam@324
|
747 crush.
|
adam@324
|
748 Qed.
|
adam@324
|
749
|
adam@324
|
750 Hint Resolve zero_times.
|
adam@324
|
751
|
adam@324
|
752 Theorem EvalVar' : forall var n,
|
adam@324
|
753 var = n
|
adam@324
|
754 -> eval var Var n.
|
adam@324
|
755 crush.
|
adam@324
|
756 Qed.
|
adam@324
|
757
|
adam@324
|
758 Hint Resolve EvalVar'.
|
adam@324
|
759
|
adam@324
|
760 Theorem plus_0 : forall n r,
|
adam@324
|
761 r = n
|
adam@324
|
762 -> r = n + 0.
|
adam@324
|
763 crush.
|
adam@324
|
764 Qed.
|
adam@324
|
765
|
adam@324
|
766 Theorem times_1 : forall n, n = 1 * n.
|
adam@324
|
767 crush.
|
adam@324
|
768 Qed.
|
adam@324
|
769
|
adam@324
|
770 Hint Resolve plus_0 times_1.
|
adam@324
|
771
|
adam@398
|
772 (** We finish with one more arithmetic lemma that is particularly specialized to this theorem. This fact happens to follow by the axioms of the _ring_ algebraic structure, so, since the naturals form a ring, we can use the built-in tactic %\index{tactics!ring}%[ring]. *)
|
adam@374
|
773
|
adam@324
|
774 Require Import Arith Ring.
|
adam@324
|
775
|
adam@324
|
776 Theorem combine : forall x k1 k2 n1 n2,
|
adam@324
|
777 (k1 * x + n1) + (k2 * x + n2) = (k1 + k2) * x + (n1 + n2).
|
adam@324
|
778 intros; ring.
|
adam@324
|
779 Qed.
|
adam@324
|
780
|
adam@324
|
781 Hint Resolve combine.
|
adam@324
|
782
|
adam@374
|
783 (** Our choice of hints is cheating, to an extent, by telegraphing the procedure for choosing values of [k] and [n]. Nonetheless, with these lemmas in place, we achieve an automated proof without explicitly orchestrating the lemmas' composition. *)
|
adam@374
|
784
|
adam@324
|
785 Theorem linear : forall e, exists k, exists n,
|
adam@324
|
786 forall var, eval var e (k * var + n).
|
adam@324
|
787 induction e; crush; eauto.
|
adam@324
|
788 Qed.
|
adam@324
|
789
|
adam@374
|
790 (* begin hide *)
|
adam@324
|
791 Print linear.
|
adam@374
|
792 (* end hide *)
|
adam@324
|
793 (* end thide *)
|
adam@324
|
794
|
adam@374
|
795 (** By printing the proof term, it is possible to see the procedure that is used to choose the constants for each input term. *)
|
adam@374
|
796
|
adam@324
|
797
|
adam@324
|
798 (** * More on [auto] Hints *)
|
adam@324
|
799
|
adam@430
|
800 (** Let us stop at this point and take stock of the possibilities for [auto] and [eauto] hints. Hints are contained within _hint databases_, which we have seen extended in many examples so far. When no hint database is specified, a default database is used. Hints in the default database are always used by [auto] or [eauto]. The chance to extend hint databases imperatively is important, because, in Ltac programming, we cannot create "global variables" whose values can be extended seamlessly by different modules in different source files. We have seen the advantages of hints so far, where [crush] can be defined once and for all, while still automatically applying the hints we add throughout developments. In fact, [crush] is defined in terms of [auto], which explains how we achieve this extensibility. Other user-defined tactics can take similar advantage of [auto] and [eauto].
|
adam@324
|
801
|
adam@375
|
802 The basic hints for [auto] and [eauto] are %\index{Vernacular commands!Hint Immediate}%[Hint Immediate lemma], asking to try solving a goal immediately by applying a lemma and discharging any hypotheses with a single proof step each; %\index{Vernacular commands!Hint Resolve}%[Resolve lemma], which does the same but may add new premises that are themselves to be subjects of nested proof search; %\index{Vernacular commands!Hint Constructors}%[Constructors type], which acts like [Resolve] applied to every constructor of an inductive type; and %\index{Vernacular commands!Hint Unfold}%[Unfold ident], which tries unfolding [ident] when it appears at the head of a proof goal. Each of these [Hint] commands may be used with a suffix, as in [Hint Resolve lemma : my_db]. This adds the hint only to the specified database, so that it would only be used by, for instance, [auto with my_db]. An additional argument to [auto] specifies the maximum depth of proof trees to search in depth-first order, as in [auto 8] or [auto 8 with my_db]. The default depth is 5.
|
adam@324
|
803
|
adam@375
|
804 All of these [Hint] commands can be issued alternatively with a more primitive hint kind, [Extern]. A few more examples of [Hint Extern] should illustrate more of the possibilities. *)
|
adam@324
|
805
|
adam@324
|
806 Theorem bool_neq : true <> false.
|
adam@324
|
807 (* begin thide *)
|
adam@324
|
808 auto.
|
adam@324
|
809
|
adam@410
|
810 (** A call to [crush] would have discharged this goal, but the default hint database for [auto] contains no hint that applies. *)
|
adam@324
|
811
|
adam@324
|
812 Abort.
|
adam@324
|
813
|
adam@430
|
814 (* begin hide *)
|
adam@430
|
815 Definition boool := bool.
|
adam@430
|
816 (* end hide *)
|
adam@430
|
817
|
adam@375
|
818 (** It is hard to come up with a [bool]-specific hint that is not just a restatement of the theorem we mean to prove. Luckily, a simpler form suffices, by appealing to the built-in tactic %\index{tactics!congruence}%[congruence], a complete procedure for the theory of equality, uninterpreted functions, and datatype constructors. *)
|
adam@324
|
819
|
adam@324
|
820 Hint Extern 1 (_ <> _) => congruence.
|
adam@324
|
821
|
adam@324
|
822 Theorem bool_neq : true <> false.
|
adam@324
|
823 auto.
|
adam@324
|
824 Qed.
|
adam@324
|
825 (* end thide *)
|
adam@324
|
826
|
adam@410
|
827 (** A [Hint Extern] may be implemented with the full Ltac language. This example shows a case where a hint uses a [match]. *)
|
adam@324
|
828
|
adam@324
|
829 Section forall_and.
|
adam@324
|
830 Variable A : Set.
|
adam@324
|
831 Variables P Q : A -> Prop.
|
adam@324
|
832
|
adam@324
|
833 Hypothesis both : forall x, P x /\ Q x.
|
adam@324
|
834
|
adam@324
|
835 Theorem forall_and : forall z, P z.
|
adam@324
|
836 (* begin thide *)
|
adam@324
|
837 crush.
|
adam@324
|
838
|
adam@375
|
839 (** The [crush] invocation makes no progress beyond what [intros] would have accomplished. An [auto] invocation will not apply the hypothesis [both] to prove the goal, because the conclusion of [both] does not unify with the conclusion of the goal. However, we can teach [auto] to handle this kind of goal. *)
|
adam@324
|
840
|
adam@324
|
841 Hint Extern 1 (P ?X) =>
|
adam@324
|
842 match goal with
|
adam@324
|
843 | [ H : forall x, P x /\ _ |- _ ] => apply (proj1 (H X))
|
adam@324
|
844 end.
|
adam@324
|
845
|
adam@324
|
846 auto.
|
adam@324
|
847 Qed.
|
adam@324
|
848 (* end thide *)
|
adam@324
|
849
|
adam@375
|
850 (** We see that an [Extern] pattern may bind unification variables that we use in the associated tactic. The function [proj1] is from the standard library, for extracting a proof of [R] from a proof of [R /\ S]. *)
|
adam@324
|
851
|
adam@324
|
852 End forall_and.
|
adam@324
|
853
|
adam@430
|
854 (* begin hide *)
|
adam@430
|
855 Definition noot := (not, @eq).
|
adam@430
|
856 (* end hide *)
|
adam@430
|
857
|
adam@324
|
858 (** After our success on this example, we might get more ambitious and seek to generalize the hint to all possible predicates [P].
|
adam@324
|
859 [[
|
adam@430
|
860 Hint Extern 1 (?P ?X) =>
|
adam@430
|
861 match goal with
|
adam@430
|
862 | [ H : forall x, P x /\ _ |- _ ] => apply (proj1 (H X))
|
adam@430
|
863 end.
|
adam@375
|
864 ]]
|
adam@375
|
865 <<
|
adam@375
|
866 User error: Bound head variable
|
adam@375
|
867 >>
|
adam@324
|
868
|
adam@410
|
869 Coq's [auto] hint databases work as tables mapping%\index{head symbol}% _head symbols_ to lists of tactics to try. Because of this, the constant head of an [Extern] pattern must be determinable statically. In our first [Extern] hint, the head symbol was [not], since [x <> y] desugars to [not (eq x y)]; and, in the second example, the head symbol was [P].
|
adam@324
|
870
|
adam@375
|
871 Fortunately, a more basic form of [Hint Extern] also applies. We may simply leave out the pattern to the left of the [=>], incorporating the corresponding logic into the Ltac script. *)
|
adam@324
|
872
|
adam@375
|
873 Hint Extern 1 =>
|
adam@375
|
874 match goal with
|
adam@375
|
875 | [ H : forall x, ?P x /\ _ |- ?P ?X ] => apply (proj1 (H X))
|
adam@375
|
876 end.
|
adam@375
|
877
|
adam@398
|
878 (** Be forewarned that a [Hint Extern] of this kind will be applied at _every_ node of a proof tree, so an expensive Ltac script may slow proof search significantly. *)
|
adam@324
|
879
|
adam@324
|
880
|
adam@324
|
881 (** * Rewrite Hints *)
|
adam@324
|
882
|
adam@375
|
883 (** Another dimension of extensibility with hints is rewriting with quantified equalities. We have used the associated command %\index{Vernacular commands!Hint Rewrite}%[Hint Rewrite] in many examples so far. The [crush] tactic uses these hints by calling the built-in tactic %\index{tactics!autorewrite}%[autorewrite]. Our rewrite hints have taken the form [Hint Rewrite lemma], which by default adds them to the default hint database [core]; but alternate hint databases may also be specified just as with, e.g., [Hint Resolve].
|
adam@324
|
884
|
adam@375
|
885 The next example shows a direct use of [autorewrite]. Note that, while [Hint Rewrite] uses a default database, [autorewrite] requires that a database be named. *)
|
adam@324
|
886
|
adam@324
|
887 Section autorewrite.
|
adam@324
|
888 Variable A : Set.
|
adam@324
|
889 Variable f : A -> A.
|
adam@324
|
890
|
adam@324
|
891 Hypothesis f_f : forall x, f (f x) = f x.
|
adam@324
|
892
|
adam@375
|
893 Hint Rewrite f_f.
|
adam@324
|
894
|
adam@324
|
895 Lemma f_f_f : forall x, f (f (f x)) = f x.
|
adam@375
|
896 intros; autorewrite with core; reflexivity.
|
adam@324
|
897 Qed.
|
adam@324
|
898
|
adam@430
|
899 (** There are a few ways in which [autorewrite] can lead to trouble when insufficient care is taken in choosing hints. First, the set of hints may define a nonterminating rewrite system, in which case invocations to [autorewrite] may not terminate. Second, we may add hints that "lead [autorewrite] down the wrong path." For instance: *)
|
adam@324
|
900
|
adam@324
|
901 Section garden_path.
|
adam@324
|
902 Variable g : A -> A.
|
adam@324
|
903 Hypothesis f_g : forall x, f x = g x.
|
adam@375
|
904 Hint Rewrite f_g.
|
adam@324
|
905
|
adam@324
|
906 Lemma f_f_f' : forall x, f (f (f x)) = f x.
|
adam@375
|
907 intros; autorewrite with core.
|
adam@324
|
908 (** [[
|
adam@324
|
909 ============================
|
adam@324
|
910 g (g (g x)) = g x
|
adam@324
|
911 ]]
|
adam@324
|
912 *)
|
adam@324
|
913
|
adam@324
|
914 Abort.
|
adam@324
|
915
|
adam@430
|
916 (** Our new hint was used to rewrite the goal into a form where the old hint could no longer be applied. This "non-monotonicity" of rewrite hints contrasts with the situation for [auto], where new hints may slow down proof search but can never "break" old proofs. The key difference is that [auto] either solves a goal or makes no changes to it, while [autorewrite] may change goals without solving them. The situation for [eauto] is slightly more complicated, as changes to hint databases may change the proof found for a particular goal, and that proof may influence the settings of unification variables that appear elsewhere in the proof state. *)
|
adam@324
|
917
|
adam@324
|
918 Reset garden_path.
|
adam@324
|
919
|
adam@375
|
920 (** The [autorewrite] tactic also works with quantified equalities that include additional premises, but we must be careful to avoid similar incorrect rewritings. *)
|
adam@324
|
921
|
adam@324
|
922 Section garden_path.
|
adam@324
|
923 Variable P : A -> Prop.
|
adam@324
|
924 Variable g : A -> A.
|
adam@324
|
925 Hypothesis f_g : forall x, P x -> f x = g x.
|
adam@375
|
926 Hint Rewrite f_g.
|
adam@324
|
927
|
adam@324
|
928 Lemma f_f_f' : forall x, f (f (f x)) = f x.
|
adam@375
|
929 intros; autorewrite with core.
|
adam@324
|
930 (** [[
|
adam@324
|
931
|
adam@324
|
932 ============================
|
adam@324
|
933 g (g (g x)) = g x
|
adam@324
|
934
|
adam@324
|
935 subgoal 2 is:
|
adam@324
|
936 P x
|
adam@324
|
937 subgoal 3 is:
|
adam@324
|
938 P (f x)
|
adam@324
|
939 subgoal 4 is:
|
adam@324
|
940 P (f x)
|
adam@324
|
941 ]]
|
adam@324
|
942 *)
|
adam@324
|
943
|
adam@324
|
944 Abort.
|
adam@324
|
945
|
adam@324
|
946 (** The inappropriate rule fired the same three times as before, even though we know we will not be able to prove the premises. *)
|
adam@324
|
947
|
adam@324
|
948 Reset garden_path.
|
adam@324
|
949
|
adam@324
|
950 (** Our final, successful, attempt uses an extra argument to [Hint Rewrite] that specifies a tactic to apply to generated premises. Such a hint is only used when the tactic succeeds for all premises, possibly leaving further subgoals for some premises. *)
|
adam@324
|
951
|
adam@324
|
952 Section garden_path.
|
adam@324
|
953 Variable P : A -> Prop.
|
adam@324
|
954 Variable g : A -> A.
|
adam@324
|
955 Hypothesis f_g : forall x, P x -> f x = g x.
|
adam@324
|
956 (* begin thide *)
|
adam@375
|
957 Hint Rewrite f_g using assumption.
|
adam@324
|
958 (* end thide *)
|
adam@324
|
959
|
adam@324
|
960 Lemma f_f_f' : forall x, f (f (f x)) = f x.
|
adam@324
|
961 (* begin thide *)
|
adam@375
|
962 intros; autorewrite with core; reflexivity.
|
adam@324
|
963 Qed.
|
adam@324
|
964 (* end thide *)
|
adam@324
|
965
|
adam@375
|
966 (** We may still use [autorewrite] to apply [f_g] when the generated premise is among our assumptions. *)
|
adam@324
|
967
|
adam@324
|
968 Lemma f_f_f_g : forall x, P x -> f (f x) = g x.
|
adam@324
|
969 (* begin thide *)
|
adam@375
|
970 intros; autorewrite with core; reflexivity.
|
adam@324
|
971 (* end thide *)
|
adam@324
|
972 Qed.
|
adam@324
|
973 End garden_path.
|
adam@324
|
974
|
adam@375
|
975 (** It can also be useful to apply the [autorewrite with db in *] form, which does rewriting in hypotheses, as well as in the conclusion. *)
|
adam@324
|
976
|
adam@324
|
977 Lemma in_star : forall x y, f (f (f (f x))) = f (f y)
|
adam@324
|
978 -> f x = f (f (f y)).
|
adam@324
|
979 (* begin thide *)
|
adam@375
|
980 intros; autorewrite with core in *; assumption.
|
adam@324
|
981 (* end thide *)
|
adam@324
|
982 Qed.
|
adam@324
|
983
|
adam@324
|
984 End autorewrite.
|
adam@375
|
985
|
adam@375
|
986 (** Many proofs can be automated in pleasantly modular ways with deft combination of [auto] and [autorewrite]. *)
|