|
adamc@45
|
1 (* Copyright (c) 2008, Adam Chlipala
|
|
adamc@45
|
2 *
|
|
adamc@45
|
3 * This work is licensed under a
|
|
adamc@45
|
4 * Creative Commons Attribution-Noncommercial-No Derivative Works 3.0
|
|
adamc@45
|
5 * Unported License.
|
|
adamc@45
|
6 * The license text is available at:
|
|
adamc@45
|
7 * http://creativecommons.org/licenses/by-nc-nd/3.0/
|
|
adamc@45
|
8 *)
|
|
adamc@45
|
9
|
|
adamc@45
|
10 (* begin hide *)
|
|
adamc@45
|
11 Require Import List.
|
|
adamc@45
|
12
|
|
adamc@45
|
13 Require Import Tactics.
|
|
adamc@45
|
14
|
|
adamc@45
|
15 Set Implicit Arguments.
|
|
adamc@45
|
16 (* end hide *)
|
|
adamc@45
|
17
|
|
adamc@45
|
18
|
|
adamc@45
|
19 (** %\chapter{Inductive Predicates}% *)
|
|
adamc@45
|
20
|
|
adamc@45
|
21 (** The so-called "Curry-Howard Correspondence" states a formal connection between functional programs and mathematical proofs. In the last chapter, we snuck in a first introduction to this subject in Coq. Witness the close similarity between the types [unit] and [True] from the standard library: *)
|
|
adamc@45
|
22
|
|
adamc@45
|
23 Print unit.
|
|
adamc@45
|
24 (** [[
|
|
adamc@45
|
25
|
|
adamc@45
|
26 Inductive unit : Set := tt : unit
|
|
adamc@45
|
27 ]] *)
|
|
adamc@45
|
28
|
|
adamc@45
|
29 Print True.
|
|
adamc@45
|
30 (** [[
|
|
adamc@45
|
31
|
|
adamc@45
|
32 Inductive True : Prop := I : True
|
|
adamc@45
|
33 ]] *)
|
|
adamc@45
|
34
|
|
adamc@45
|
35 (** Recall that [unit] is the type with only one value, and [True] is the proposition that always holds. Despite this superficial difference between the two concepts, in both cases we can use the same inductive definition mechanism. The connection goes further than this. We see that we arrive at the definition of [True] by replacing [unit] by [True], [tt] by [I], and [Set] by [Prop]. The first two of these differences are superficial changes of names, while the third difference is the crucial one for separating programs from proofs. A term [T] of type [Set] is a type of programs, and a term of type [T] is a program. A term [T] of type [Prop] is a logical proposition, and its proofs are of type [T].
|
|
adamc@45
|
36
|
|
adamc@45
|
37 [unit] has one value, [tt]. [True] has one proof, [I]. Why distinguish between these two types? Many people who have read about Curry-Howard in an abstract context and not put it to use in proof engineering answer that the two types in fact %\textit{%#<i>#should not#</i>#%}% be distinguished. There is a certain aesthetic appeal to this point of view, but I want to argue that it is best to treat Curry-Howard very loosely in practical proving. There are Coq-specific reasons for preferring the distinction, involving efficient compilation and avoidance of paradoxes in the presence of classical math, but I will argue that there is a more general principle that should lead us to avoid conflating programming and proving.
|
|
adamc@45
|
38
|
|
adamc@45
|
39 The essence of the argument is roughly this: to an engineer, not all functions of type [A -> B] are created equal, but all proofs of a proposition [P -> Q] are. This idea is known as %\textit{%#<i>#proof irrelevance#</i>#%}%, and its formalizations in logics prevent us from distinguishing between alternate proofs of the same proposition. Proof irrelevance is compatible with, but not derivable in, Gallina. Apart from this theoretical concern, I will argue that it is most effective to do engineering with Coq by employing different techniques for programs versus proofs. Most of this book is organized around that distinction, describing how to program, by applying standard functional programming techniques in the presence of dependent types; and how to prove, by writing custom Ltac decision procedures.
|
|
adamc@45
|
40
|
|
adamc@45
|
41 With that perspective in mind, this chapter is sort of a mirror image of the last chapter, introducing how to define predicates with inductive definitions. We will point out similarities in places, but much of the effective Coq user's bag of tricks is disjoint for predicates versus "datatypes." This chapter is also a covert introduction to dependent types, which are the foundation on which interesting inductive predicates are built, though we will rely on tactics to build dependently-typed proof terms for us for now. A future chapter introduces more manual application of dependent types. *)
|
|
adamc@45
|
42
|
|
adamc@45
|
43
|
|
adamc@48
|
44 (** * Propositional Logic *)
|
|
adamc@45
|
45
|
|
adamc@45
|
46 (** Let us begin with a brief tour through the definitions of the connectives for propositional logic. We will work within a Coq section that provides us with a set of propositional variables. In Coq parlance, these are just terms of type [Prop.] *)
|
|
adamc@45
|
47
|
|
adamc@45
|
48 Section Propositional.
|
|
adamc@46
|
49 Variables P Q R : Prop.
|
|
adamc@45
|
50
|
|
adamc@45
|
51 (** In Coq, the most basic propositional connective is implication, written [->], which we have already used in almost every proof. Rather than being defined inductively, implication is built into Coq as the function type constructor.
|
|
adamc@45
|
52
|
|
adamc@45
|
53 We have also already seen the definition of [True]. For a demonstration of a lower-level way of establishing proofs of inductive predicates, we turn to this trivial theorem. *)
|
|
adamc@45
|
54
|
|
adamc@45
|
55 Theorem obvious : True.
|
|
adamc@55
|
56 (* begin thide *)
|
|
adamc@45
|
57 apply I.
|
|
adamc@55
|
58 (* end thide *)
|
|
adamc@45
|
59 Qed.
|
|
adamc@45
|
60
|
|
adamc@45
|
61 (** We may always use the [apply] tactic to take a proof step based on applying a particular constructor of the inductive predicate that we are trying to establish. Sometimes there is only one constructor that could possibly apply, in which case a shortcut is available: *)
|
|
adamc@45
|
62
|
|
adamc@55
|
63 (* begin thide *)
|
|
adamc@45
|
64 Theorem obvious' : True.
|
|
adamc@45
|
65 constructor.
|
|
adamc@45
|
66 Qed.
|
|
adamc@45
|
67
|
|
adamc@55
|
68 (* end thide *)
|
|
adamc@55
|
69
|
|
adamc@45
|
70 (** There is also a predicate [False], which is the Curry-Howard mirror image of [Empty_set] from the last chapter. *)
|
|
adamc@45
|
71
|
|
adamc@45
|
72 Print False.
|
|
adamc@45
|
73 (** [[
|
|
adamc@45
|
74
|
|
adamc@45
|
75 Inductive False : Prop :=
|
|
adamc@45
|
76 ]] *)
|
|
adamc@45
|
77
|
|
adamc@45
|
78 (** We can conclude anything from [False], doing case analysis on a proof of [False] in the same way we might do case analysis on, say, a natural number. Since there are no cases to consider, any such case analysis succeeds immediately in proving the goal. *)
|
|
adamc@45
|
79
|
|
adamc@45
|
80 Theorem False_imp : False -> 2 + 2 = 5.
|
|
adamc@55
|
81 (* begin thide *)
|
|
adamc@45
|
82 destruct 1.
|
|
adamc@55
|
83 (* end thide *)
|
|
adamc@45
|
84 Qed.
|
|
adamc@45
|
85
|
|
adamc@45
|
86 (** In a consistent context, we can never build a proof of [False]. In inconsistent contexts that appear in the courses of proofs, it is usually easiest to proceed by demonstrating that inconsistency with an explicit proof of [False]. *)
|
|
adamc@45
|
87
|
|
adamc@45
|
88 Theorem arith_neq : 2 + 2 = 5 -> 9 + 9 = 835.
|
|
adamc@55
|
89 (* begin thide *)
|
|
adamc@45
|
90 intro.
|
|
adamc@45
|
91
|
|
adamc@45
|
92 (** At this point, we have an inconsistent hypothesis [2 + 2 = 5], so the specific conclusion is not important. We use the [elimtype] tactic to state a proposition, telling Coq that we wish to construct a proof of the new proposition and then prove the original goal by case analysis on the structure of the new auxiliary proof. Since [False] has no constructors, [elimtype False] simply leaves us with the obligation to prove [False]. *)
|
|
adamc@45
|
93
|
|
adamc@45
|
94 elimtype False.
|
|
adamc@45
|
95 (** [[
|
|
adamc@45
|
96
|
|
adamc@45
|
97 H : 2 + 2 = 5
|
|
adamc@45
|
98 ============================
|
|
adamc@45
|
99 False
|
|
adamc@45
|
100 ]] *)
|
|
adamc@45
|
101
|
|
adamc@45
|
102 (** For now, we will leave the details of this proof about arithmetic to [crush]. *)
|
|
adamc@45
|
103
|
|
adamc@45
|
104 crush.
|
|
adamc@55
|
105 (* end thide *)
|
|
adamc@45
|
106 Qed.
|
|
adamc@45
|
107
|
|
adamc@45
|
108 (** A related notion to [False] is logical negation. *)
|
|
adamc@45
|
109
|
|
adamc@45
|
110 Print not.
|
|
adamc@45
|
111 (** [[
|
|
adamc@45
|
112
|
|
adamc@45
|
113 not = fun A : Prop => A -> False
|
|
adamc@45
|
114 : Prop -> Prop
|
|
adamc@45
|
115 ]] *)
|
|
adamc@45
|
116
|
|
adamc@45
|
117 (** We see that [not] is just shorthand for implication of [False]. We can use that fact explicitly in proofs. The syntax [~P] expands to [not P]. *)
|
|
adamc@45
|
118
|
|
adamc@45
|
119 Theorem arith_neq' : ~ (2 + 2 = 5).
|
|
adamc@55
|
120 (* begin thide *)
|
|
adamc@45
|
121 unfold not.
|
|
adamc@45
|
122
|
|
adamc@45
|
123 (** [[
|
|
adamc@45
|
124
|
|
adamc@45
|
125 ============================
|
|
adamc@45
|
126 2 + 2 = 5 -> False
|
|
adamc@45
|
127 ]] *)
|
|
adamc@45
|
128
|
|
adamc@45
|
129 crush.
|
|
adamc@55
|
130 (* end thide *)
|
|
adamc@45
|
131 Qed.
|
|
adamc@45
|
132
|
|
adamc@45
|
133 (** We also have conjunction, which we introduced in the last chapter. *)
|
|
adamc@45
|
134
|
|
adamc@45
|
135 Print and.
|
|
adamc@45
|
136 (** [[
|
|
adamc@45
|
137
|
|
adamc@45
|
138 Inductive and (A : Prop) (B : Prop) : Prop := conj : A -> B -> A /\ B
|
|
adamc@45
|
139 ]] *)
|
|
adamc@45
|
140
|
|
adamc@45
|
141 (** The interested reader can check that [and] has a Curry-Howard doppleganger called [prod], the type of pairs. However, it is generally most convenient to reason about conjunction using tactics. An explicit proof of commutativity of [and] illustrates the usual suspects for such tasks. [/\] is an infix shorthand for [and]. *)
|
|
adamc@45
|
142
|
|
adamc@45
|
143 Theorem and_comm : P /\ Q -> Q /\ P.
|
|
adamc@55
|
144 (* begin thide *)
|
|
adamc@45
|
145 (** We start by case analysis on the proof of [P /\ Q]. *)
|
|
adamc@45
|
146
|
|
adamc@45
|
147 destruct 1.
|
|
adamc@45
|
148 (** [[
|
|
adamc@45
|
149
|
|
adamc@45
|
150 H : P
|
|
adamc@45
|
151 H0 : Q
|
|
adamc@45
|
152 ============================
|
|
adamc@45
|
153 Q /\ P
|
|
adamc@45
|
154 ]] *)
|
|
adamc@45
|
155
|
|
adamc@45
|
156 (** Every proof of a conjunction provides proofs for both conjuncts, so we get a single subgoal reflecting that. We can proceed by splitting this subgoal into a case for each conjunct of [Q /\ P]. *)
|
|
adamc@45
|
157
|
|
adamc@45
|
158 split.
|
|
adamc@45
|
159 (** [[
|
|
adamc@45
|
160 2 subgoals
|
|
adamc@45
|
161
|
|
adamc@45
|
162 H : P
|
|
adamc@45
|
163 H0 : Q
|
|
adamc@45
|
164 ============================
|
|
adamc@45
|
165 Q
|
|
adamc@45
|
166
|
|
adamc@45
|
167 subgoal 2 is:
|
|
adamc@45
|
168 P
|
|
adamc@45
|
169 ]] *)
|
|
adamc@45
|
170
|
|
adamc@45
|
171 (** In each case, the conclusion is among our hypotheses, so the [assumption] tactic finishes the process. *)
|
|
adamc@45
|
172
|
|
adamc@45
|
173 assumption.
|
|
adamc@45
|
174 assumption.
|
|
adamc@55
|
175 (* end thide *)
|
|
adamc@45
|
176 Qed.
|
|
adamc@45
|
177
|
|
adamc@45
|
178 (** Coq disjunction is called [or] and abbreviated with the infix operator [\/]. *)
|
|
adamc@45
|
179
|
|
adamc@45
|
180 Print or.
|
|
adamc@45
|
181 (** [[
|
|
adamc@45
|
182
|
|
adamc@45
|
183 Inductive or (A : Prop) (B : Prop) : Prop :=
|
|
adamc@45
|
184 or_introl : A -> A \/ B | or_intror : B -> A \/ B
|
|
adamc@45
|
185 ]] *)
|
|
adamc@45
|
186
|
|
adamc@45
|
187 (** We see that there are two ways to prove a disjunction: prove the first disjunct or prove the second. The Curry-Howard analogue of this is the Coq [sum] type. We can demonstrate the main tactics here with another proof of commutativity. *)
|
|
adamc@45
|
188
|
|
adamc@45
|
189 Theorem or_comm : P \/ Q -> Q \/ P.
|
|
adamc@55
|
190
|
|
adamc@55
|
191 (* begin thide *)
|
|
adamc@45
|
192 (** As in the proof for [and], we begin with case analysis, though this time we are met by two cases instead of one. *)
|
|
adamc@45
|
193 destruct 1.
|
|
adamc@45
|
194 (** [[
|
|
adamc@45
|
195
|
|
adamc@45
|
196 2 subgoals
|
|
adamc@45
|
197
|
|
adamc@45
|
198 H : P
|
|
adamc@45
|
199 ============================
|
|
adamc@45
|
200 Q \/ P
|
|
adamc@45
|
201
|
|
adamc@45
|
202 subgoal 2 is:
|
|
adamc@45
|
203 Q \/ P
|
|
adamc@45
|
204 ]] *)
|
|
adamc@45
|
205
|
|
adamc@45
|
206 (** We can see that, in the first subgoal, we want to prove the disjunction by proving its second disjunct. The [right] tactic telegraphs this intent. *)
|
|
adamc@45
|
207
|
|
adamc@45
|
208 right; assumption.
|
|
adamc@45
|
209
|
|
adamc@45
|
210 (** The second subgoal has a symmetric proof.
|
|
adamc@45
|
211
|
|
adamc@45
|
212 [[
|
|
adamc@45
|
213
|
|
adamc@45
|
214 1 subgoal
|
|
adamc@45
|
215
|
|
adamc@45
|
216 H : Q
|
|
adamc@45
|
217 ============================
|
|
adamc@45
|
218 Q \/ P
|
|
adamc@45
|
219 ]] *)
|
|
adamc@45
|
220
|
|
adamc@45
|
221 left; assumption.
|
|
adamc@55
|
222 (* end thide *)
|
|
adamc@45
|
223 Qed.
|
|
adamc@45
|
224
|
|
adamc@46
|
225
|
|
adamc@46
|
226 (* begin hide *)
|
|
adamc@46
|
227 (* In-class exercises *)
|
|
adamc@46
|
228
|
|
adamc@46
|
229 Theorem contra : P -> ~P -> R.
|
|
adamc@52
|
230 (* begin thide *)
|
|
adamc@52
|
231 unfold not.
|
|
adamc@52
|
232 intros.
|
|
adamc@52
|
233 elimtype False.
|
|
adamc@52
|
234 apply H0.
|
|
adamc@52
|
235 assumption.
|
|
adamc@52
|
236 (* end thide *)
|
|
adamc@46
|
237 Admitted.
|
|
adamc@46
|
238
|
|
adamc@46
|
239 Theorem and_assoc : (P /\ Q) /\ R -> P /\ (Q /\ R).
|
|
adamc@52
|
240 (* begin thide *)
|
|
adamc@52
|
241 intros.
|
|
adamc@52
|
242 destruct H.
|
|
adamc@52
|
243 destruct H.
|
|
adamc@52
|
244 split.
|
|
adamc@52
|
245 assumption.
|
|
adamc@52
|
246 split.
|
|
adamc@52
|
247 assumption.
|
|
adamc@52
|
248 assumption.
|
|
adamc@52
|
249 (* end thide *)
|
|
adamc@46
|
250 Admitted.
|
|
adamc@46
|
251
|
|
adamc@46
|
252 Theorem or_assoc : (P \/ Q) \/ R -> P \/ (Q \/ R).
|
|
adamc@52
|
253 (* begin thide *)
|
|
adamc@52
|
254 intros.
|
|
adamc@52
|
255 destruct H.
|
|
adamc@52
|
256 destruct H.
|
|
adamc@52
|
257 left.
|
|
adamc@52
|
258 assumption.
|
|
adamc@52
|
259 right.
|
|
adamc@52
|
260 left.
|
|
adamc@52
|
261 assumption.
|
|
adamc@52
|
262 right.
|
|
adamc@52
|
263 right.
|
|
adamc@52
|
264 assumption.
|
|
adamc@52
|
265 (* end thide *)
|
|
adamc@46
|
266 Admitted.
|
|
adamc@46
|
267
|
|
adamc@46
|
268 (* end hide *)
|
|
adamc@46
|
269
|
|
adamc@46
|
270
|
|
adamc@46
|
271 (** It would be a shame to have to plod manually through all proofs about propositional logic. Luckily, there is no need. One of the most basic Coq automation tactics is [tauto], which is a complete decision procedure for constructive propositional logic. (More on what "constructive" means in the next section.) We can use [tauto] to dispatch all of the purely propositional theorems we have proved so far. *)
|
|
adamc@46
|
272
|
|
adamc@46
|
273 Theorem or_comm' : P \/ Q -> Q \/ P.
|
|
adamc@55
|
274 (* begin thide *)
|
|
adamc@46
|
275 tauto.
|
|
adamc@55
|
276 (* end thide *)
|
|
adamc@46
|
277 Qed.
|
|
adamc@46
|
278
|
|
adamc@46
|
279 (** Sometimes propositional reasoning forms important plumbing for the proof of a theorem, but we still need to apply some other smarts about, say, arithmetic. [intuition] is a generalization of [tauto] that proves everything it can using propositional reasoning. When some goals remain, it uses propositional laws to simplify them as far as possible. Consider this example, which uses the list concatenation operator [++] from the standard library. *)
|
|
adamc@46
|
280
|
|
adamc@46
|
281 Theorem arith_comm : forall ls1 ls2 : list nat,
|
|
adamc@46
|
282 length ls1 = length ls2 \/ length ls1 + length ls2 = 6
|
|
adamc@46
|
283 -> length (ls1 ++ ls2) = 6 \/ length ls1 = length ls2.
|
|
adamc@55
|
284 (* begin thide *)
|
|
adamc@46
|
285 intuition.
|
|
adamc@46
|
286
|
|
adamc@46
|
287 (** A lot of the proof structure has been generated for us by [intuition], but the final proof depends on a fact about lists. The remaining subgoal hints at what cleverness we need to inject. *)
|
|
adamc@46
|
288
|
|
adamc@46
|
289 (** [[
|
|
adamc@46
|
290
|
|
adamc@46
|
291 ls1 : list nat
|
|
adamc@46
|
292 ls2 : list nat
|
|
adamc@46
|
293 H0 : length ls1 + length ls2 = 6
|
|
adamc@46
|
294 ============================
|
|
adamc@46
|
295 length (ls1 ++ ls2) = 6 \/ length ls1 = length ls2
|
|
adamc@46
|
296 ]] *)
|
|
adamc@46
|
297
|
|
adamc@46
|
298 (** We can see that we need a theorem about lengths of concatenated lists, which we proved last chapter and is also in the standard library. *)
|
|
adamc@46
|
299
|
|
adamc@46
|
300 rewrite app_length.
|
|
adamc@46
|
301 (** [[
|
|
adamc@46
|
302
|
|
adamc@46
|
303 ls1 : list nat
|
|
adamc@46
|
304 ls2 : list nat
|
|
adamc@46
|
305 H0 : length ls1 + length ls2 = 6
|
|
adamc@46
|
306 ============================
|
|
adamc@46
|
307 length ls1 + length ls2 = 6 \/ length ls1 = length ls2
|
|
adamc@46
|
308 ]] *)
|
|
adamc@46
|
309
|
|
adamc@46
|
310 (** Now the subgoal follows by purely propositional reasoning. That is, we could replace [length ls1 + length ls2 = 6] with [P] and [length ls1 = length ls2] with [Q] and arrive at a tautology of propositional logic. *)
|
|
adamc@46
|
311
|
|
adamc@46
|
312 tauto.
|
|
adamc@55
|
313 (* end thide *)
|
|
adamc@46
|
314 Qed.
|
|
adamc@46
|
315
|
|
adamc@46
|
316 (** [intuition] is one of the main bits of glue in the implementation of [crush], so, with a little help, we can get a short automated proof of the theorem. *)
|
|
adamc@46
|
317
|
|
adamc@55
|
318 (* begin thide *)
|
|
adamc@46
|
319 Theorem arith_comm' : forall ls1 ls2 : list nat,
|
|
adamc@46
|
320 length ls1 = length ls2 \/ length ls1 + length ls2 = 6
|
|
adamc@46
|
321 -> length (ls1 ++ ls2) = 6 \/ length ls1 = length ls2.
|
|
adamc@46
|
322 Hint Rewrite app_length : cpdt.
|
|
adamc@46
|
323
|
|
adamc@46
|
324 crush.
|
|
adamc@46
|
325 Qed.
|
|
adamc@55
|
326 (* end thide *)
|
|
adamc@46
|
327
|
|
adamc@45
|
328 End Propositional.
|
|
adamc@45
|
329
|
|
adamc@46
|
330
|
|
adamc@47
|
331 (** * What Does It Mean to Be Constructive? *)
|
|
adamc@46
|
332
|
|
adamc@47
|
333 (** One potential point of confusion in the presentation so far is the distinction between [bool] and [Prop]. [bool] is a datatype whose two values are [true] and [false], while [Prop] is a more primitive type that includes among its members [True] and [False]. Why not collapse these two concepts into one, and why must there be more than two states of mathematical truth?
|
|
adamc@46
|
334
|
|
adamc@47
|
335 The answer comes from the fact that Coq implements %\textit{%#<i>#constructive#</i>#%}% or %\textit{%#<i>#intuitionistic#</i>#%}% logic, in contrast to the %\textit{%#<i>#classical#</i>#%}% logic that you may be more familiar with. In constructive logic, classical tautologies like [~ ~P -> P] and [P \/ ~P] do not always hold. In general, we can only prove these tautologies when [P] is %\textit{%#<i>#decidable#</i>#%}%, in the sense of computability theory. The Curry-Howard encoding that Coq uses for [or] allows us to extract either a proof of [P] or a proof of [~P] from any proof of [P \/ ~P]. Since our proofs are just functional programs which we can run, this would give us a decision procedure for the halting problem, where the instantiations of [P] would be formulas like "this particular Turing machine halts."
|
|
adamc@47
|
336
|
|
adamc@47
|
337 Hence the distinction between [bool] and [Prop]. Programs of type [bool] are computational by construction; we can always run them to determine their results. Many [Prop]s are undecidable, and so we can write more expressive formulas with [Prop]s than with [bool]s, but the inevitable consequence is that we cannot simply "run a [Prop] to determine its truth."
|
|
adamc@47
|
338
|
|
adamc@47
|
339 Constructive logic lets us define all of the logical connectives in an aesthetically-appealing way, with orthogonal inductive definitions. That is, each connective is defined independently using a simple, shared mechanism. Constructivity also enables a trick called %\textit{%#<i>#program extraction#</i>#%}%, where we write programs by phrasing them as theorems to be proved. Since our proofs are just functional programs, we can extract executable programs from our final proofs, which we could not do as naturally with classical proofs.
|
|
adamc@47
|
340
|
|
adamc@47
|
341 We will see more about Coq's program extraction facility in a later chapter. However, I think it is worth interjecting another warning at this point, following up on the prior warning about taking the Curry-Howard correspondence too literally. It is possible to write programs by theorem-proving methods in Coq, but hardly anyone does it. It is almost always most useful to maintain the distinction between programs and proofs. If you write a program by proving a theorem, you are likely to run into algorithmic inefficiencies that you introduced in your proof to make it easier to prove. It is a shame to have to worry about such situations while proving tricky theorems, and it is a happy state of affairs that you almost certainly will not need to, with the ideal of extracting programs from proofs being confined mostly to theoretical studies. *)
|
|
adamc@48
|
342
|
|
adamc@48
|
343
|
|
adamc@48
|
344 (** * First-Order Logic *)
|
|
adamc@48
|
345
|
|
adamc@48
|
346 (** The [forall] connective of first-order logic, which we have seen in many examples so far, is built into Coq. Getting ahead of ourselves a bit, we can see it as the dependent function type constructor. In fact, implication and universal quantification are just different syntactic shorthands for the same Coq mechanism. A formula [P -> Q] is equivalent to [forall x : P, Q], where [x] does not appear in [Q]. That is, the "real" type of the implication says "for every proof of [P], there exists a proof of [Q]."
|
|
adamc@48
|
347
|
|
adamc@48
|
348 Existential quantification is defined in the standard library. *)
|
|
adamc@48
|
349
|
|
adamc@48
|
350 Print ex.
|
|
adamc@48
|
351 (** [[
|
|
adamc@48
|
352
|
|
adamc@48
|
353 Inductive ex (A : Type) (P : A -> Prop) : Prop :=
|
|
adamc@48
|
354 ex_intro : forall x : A, P x -> ex P
|
|
adamc@48
|
355 ]] *)
|
|
adamc@48
|
356
|
|
adamc@48
|
357 (** [ex] is parameterized by the type [A] that we quantify over, and by a predicate [P] over [A]s. We prove an existential by exhibiting some [x] of type [A], along with a proof of [P x]. As usual, there are tactics that save us from worrying about the low-level details most of the time. *)
|
|
adamc@48
|
358
|
|
adamc@48
|
359 Theorem exist1 : exists x : nat, x + 1 = 2.
|
|
adamc@55
|
360 (* begin thide *)
|
|
adamc@67
|
361 (** remove printing exists *)
|
|
adamc@55
|
362 (** We can start this proof with a tactic [exists], which should not be confused with the formula constructor shorthand of the same name. (In the PDF version of this document, the reverse 'E' appears instead of the text "exists" in formulas.) *)
|
|
adamc@48
|
363 exists 1.
|
|
adamc@48
|
364
|
|
adamc@48
|
365 (** The conclusion is replaced with a version using the existential witness that we announced. *)
|
|
adamc@48
|
366
|
|
adamc@48
|
367 (** [[
|
|
adamc@48
|
368
|
|
adamc@48
|
369 ============================
|
|
adamc@48
|
370 1 + 1 = 2
|
|
adamc@48
|
371 ]] *)
|
|
adamc@48
|
372
|
|
adamc@48
|
373 reflexivity.
|
|
adamc@55
|
374 (* end thide *)
|
|
adamc@48
|
375 Qed.
|
|
adamc@48
|
376
|
|
adamc@48
|
377 (** printing exists $\exists$ *)
|
|
adamc@48
|
378
|
|
adamc@48
|
379 (** We can also use tactics to reason about existential hypotheses. *)
|
|
adamc@48
|
380
|
|
adamc@48
|
381 Theorem exist2 : forall n m : nat, (exists x : nat, n + x = m) -> n <= m.
|
|
adamc@55
|
382 (* begin thide *)
|
|
adamc@48
|
383 (** We start by case analysis on the proof of the existential fact. *)
|
|
adamc@48
|
384 destruct 1.
|
|
adamc@48
|
385 (** [[
|
|
adamc@48
|
386
|
|
adamc@48
|
387 n : nat
|
|
adamc@48
|
388 m : nat
|
|
adamc@48
|
389 x : nat
|
|
adamc@48
|
390 H : n + x = m
|
|
adamc@48
|
391 ============================
|
|
adamc@48
|
392 n <= m
|
|
adamc@48
|
393 ]] *)
|
|
adamc@48
|
394
|
|
adamc@48
|
395 (** The goal has been replaced by a form where there is a new free variable [x], and where we have a new hypothesis that the body of the existential holds with [x] substituted for the old bound variable. From here, the proof is just about arithmetic and is easy to automate. *)
|
|
adamc@48
|
396
|
|
adamc@48
|
397 crush.
|
|
adamc@55
|
398 (* end thide *)
|
|
adamc@48
|
399 Qed.
|
|
adamc@48
|
400
|
|
adamc@48
|
401
|
|
adamc@48
|
402 (* begin hide *)
|
|
adamc@48
|
403 (* In-class exercises *)
|
|
adamc@48
|
404
|
|
adamc@48
|
405 Theorem forall_exists_commute : forall (A B : Type) (P : A -> B -> Prop),
|
|
adamc@48
|
406 (exists x : A, forall y : B, P x y) -> (forall y : B, exists x : A, P x y).
|
|
adamc@52
|
407 (* begin thide *)
|
|
adamc@52
|
408 intros.
|
|
adamc@52
|
409 destruct H.
|
|
adamc@52
|
410 exists x.
|
|
adamc@52
|
411 apply H.
|
|
adamc@52
|
412 (* end thide *)
|
|
adamc@48
|
413 Admitted.
|
|
adamc@48
|
414
|
|
adamc@48
|
415 (* end hide *)
|
|
adamc@48
|
416
|
|
adamc@48
|
417
|
|
adamc@48
|
418 (** The tactic [intuition] has a first-order cousin called [firstorder]. [firstorder] proves many formulas when only first-order reasoning is needed, and it tries to perform first-order simplifications in any case. First-order reasoning is much harder than propositional reasoning, so [firstorder] is much more likely than [intuition] to get stuck in a way that makes it run for long enough to be useless. *)
|
|
adamc@49
|
419
|
|
adamc@49
|
420
|
|
adamc@49
|
421 (** * Predicates with Implicit Equality *)
|
|
adamc@49
|
422
|
|
adamc@49
|
423 (** We start our exploration of a more complicated class of predicates with a simple example: an alternative way of characterizing when a natural number is zero. *)
|
|
adamc@49
|
424
|
|
adamc@49
|
425 Inductive isZero : nat -> Prop :=
|
|
adamc@49
|
426 | IsZero : isZero 0.
|
|
adamc@49
|
427
|
|
adamc@49
|
428 Theorem isZero_zero : isZero 0.
|
|
adamc@55
|
429 (* begin thide *)
|
|
adamc@49
|
430 constructor.
|
|
adamc@55
|
431 (* end thide *)
|
|
adamc@49
|
432 Qed.
|
|
adamc@49
|
433
|
|
adamc@49
|
434 (** We can call [isZero] a %\textit{%#<i>#judgment#</i>#%}%, in the sense often used in the semantics of programming languages. Judgments are typically defined in the style of %\textit{%#<i>#natural deduction#</i>#%}%, where we write a number of %\textit{%#<i>#inference rules#</i>#%}% with premises appearing above a solid line and a conclusion appearing below the line. In this example, the sole constructor [IsZero] of [isZero] can be thought of as the single inference rule for deducing [isZero], with nothing above the line and [isZero 0] below it. The proof of [isZero_zero] demonstrates how we can apply an inference rule.
|
|
adamc@49
|
435
|
|
adamc@49
|
436 The definition of [isZero] differs in an important way from all of the other inductive definitions that we have seen in this and the previous chapter. Instead of writing just [Set] or [Prop] after the colon, here we write [nat -> Prop]. We saw examples of parameterized types like [list], but there the parameters appeared with names %\textit{%#<i>#before#</i>#%}% the colon. Every constructor of a parameterized inductive type must have a range type that uses the same parameter, whereas the form we use here enables us to use different arguments to the type for different constructors.
|
|
adamc@49
|
437
|
|
adamc@49
|
438 For instance, [isZero] forces its argument to be [0]. We can see that the concept of equality is somehow implicit in the inductive definition mechanism. The way this is accomplished is similar to the way that logic variables are used in Prolog, and it is a very powerful mechanism that forms a foundation for formalizing all of mathematics. In fact, though it is natural to think of inductive types as folding in the functionality of equality, in Coq, the true situation is reversed, with equality defined as just another inductive type! *)
|
|
adamc@49
|
439
|
|
adamc@49
|
440 Print eq.
|
|
adamc@49
|
441 (** [[
|
|
adamc@49
|
442
|
|
adamc@49
|
443 Inductive eq (A : Type) (x : A) : A -> Prop := refl_equal : x = x
|
|
adamc@49
|
444 ]] *)
|
|
adamc@49
|
445
|
|
adamc@202
|
446 (** [eq] is the type we get behind the scenes when uses of infix [=] are expanded. We see that [eq] has both a parameter [x] that is fixed and an extra unnamed argument of the same type. The type of [eq] allows us to state any equalities, even those that are provably false. However, examining the type of equality's sole constructor [refl_equal], we see that we can only %\textit{%#<i>#prove#</i>#%}% equality when its two arguments are syntactically equal. This definition turns out to capture all of the basic properties of equality, and the equality-manipulating tactics that we have seen so far, like [reflexivity] and [rewrite], are implemented treating [eq] as just another inductive type with a well-chosen definition.
|
|
adamc@49
|
447
|
|
adamc@49
|
448 Returning to the example of [isZero], we can see how to make use of hypotheses that use this predicate. *)
|
|
adamc@49
|
449
|
|
adamc@49
|
450 Theorem isZero_plus : forall n m : nat, isZero m -> n + m = n.
|
|
adamc@55
|
451 (* begin thide *)
|
|
adamc@49
|
452 (** We want to proceed by cases on the proof of the assumption about [isZero]. *)
|
|
adamc@49
|
453 destruct 1.
|
|
adamc@49
|
454 (** [[
|
|
adamc@49
|
455
|
|
adamc@49
|
456 n : nat
|
|
adamc@49
|
457 ============================
|
|
adamc@49
|
458 n + 0 = n
|
|
adamc@49
|
459 ]] *)
|
|
adamc@49
|
460
|
|
adamc@49
|
461 (** Since [isZero] has only one constructor, we are presented with only one subgoal. The argument [m] to [isZero] is replaced with that type's argument from the single constructor [IsZero]. From this point, the proof is trivial. *)
|
|
adamc@49
|
462
|
|
adamc@49
|
463 crush.
|
|
adamc@55
|
464 (* end thide *)
|
|
adamc@49
|
465 Qed.
|
|
adamc@49
|
466
|
|
adamc@49
|
467 (** Another example seems at first like it should admit an analogous proof, but in fact provides a demonstration of one of the most basic gotchas of Coq proving. *)
|
|
adamc@49
|
468
|
|
adamc@49
|
469 Theorem isZero_contra : isZero 1 -> False.
|
|
adamc@55
|
470 (* begin thide *)
|
|
adamc@49
|
471 (** Let us try a proof by cases on the assumption, as in the last proof. *)
|
|
adamc@49
|
472 destruct 1.
|
|
adamc@49
|
473 (** [[
|
|
adamc@49
|
474
|
|
adamc@49
|
475 ============================
|
|
adamc@49
|
476 False
|
|
adamc@49
|
477 ]] *)
|
|
adamc@49
|
478
|
|
adamc@49
|
479 (** It seems that case analysis has not helped us much at all! Our sole hypothesis disappears, leaving us, if anything, worse off than we were before. What went wrong? We have met an important restriction in tactics like [destruct] and [induction] when applied to types with arguments. If the arguments are not already free variables, they will be replaced by new free variables internally before doing the case analysis or induction. Since the argument [1] to [isZero] is replaced by a fresh variable, we lose the crucial fact that it is not equal to [0].
|
|
adamc@49
|
480
|
|
adamc@49
|
481 Why does Coq use this restriction? We will discuss the issue in detail in a future chapter, when we see the dependently-typed programming techniques that would allow us to write this proof term manually. For now, we just say that the algorithmic problem of "logically complete case analysis" is undecidable when phrased in Coq's logic. A few tactics and design patterns that we will present in this chapter suffice in almost all cases. For the current example, what we want is a tactic called [inversion], which corresponds to the concept of inversion that is frequently used with natural deduction proof systems. *)
|
|
adamc@49
|
482
|
|
adamc@49
|
483 Undo.
|
|
adamc@49
|
484 inversion 1.
|
|
adamc@55
|
485 (* end thide *)
|
|
adamc@49
|
486 Qed.
|
|
adamc@49
|
487
|
|
adamc@49
|
488 (** What does [inversion] do? Think of it as a version of [destruct] that does its best to take advantage of the structure of arguments to inductive types. In this case, [inversion] completed the proof immediately, because it was able to detect that we were using [isZero] with an impossible argument.
|
|
adamc@49
|
489
|
|
adamc@49
|
490 Sometimes using [destruct] when you should have used [inversion] can lead to confusing results. To illustrate, consider an alternate proof attempt for the last theorem. *)
|
|
adamc@49
|
491
|
|
adamc@49
|
492 Theorem isZero_contra' : isZero 1 -> 2 + 2 = 5.
|
|
adamc@49
|
493 destruct 1.
|
|
adamc@49
|
494 (** [[
|
|
adamc@49
|
495
|
|
adamc@49
|
496 ============================
|
|
adamc@49
|
497 1 + 1 = 4
|
|
adamc@49
|
498 ]] *)
|
|
adamc@49
|
499
|
|
adamc@49
|
500 (** What on earth happened here? Internally, [destruct] replaced [1] with a fresh variable, and, trying to be helpful, it also replaced the occurrence of [1] within the unary representation of each number in the goal. This has the net effect of decrementing each of these numbers. If you are doing a proof and encounter a strange transmutation like this, there is a good chance that you should go back and replace a use of [destruct] with [inversion]. *)
|
|
adamc@49
|
501 Abort.
|
|
adamc@49
|
502
|
|
adamc@49
|
503
|
|
adamc@49
|
504 (* begin hide *)
|
|
adamc@49
|
505 (* In-class exercises *)
|
|
adamc@49
|
506
|
|
adamc@49
|
507 (* EX: Define an inductive type capturing when a list has exactly two elements. Prove that your predicate does not hold of the empty list, and prove that, whenever it holds of a list, the length of that list is two. *)
|
|
adamc@49
|
508
|
|
adamc@52
|
509 (* begin thide *)
|
|
adamc@52
|
510 Section twoEls.
|
|
adamc@52
|
511 Variable A : Type.
|
|
adamc@52
|
512
|
|
adamc@52
|
513 Inductive twoEls : list A -> Prop :=
|
|
adamc@52
|
514 | TwoEls : forall x y, twoEls (x :: y :: nil).
|
|
adamc@52
|
515
|
|
adamc@52
|
516 Theorem twoEls_nil : twoEls nil -> False.
|
|
adamc@52
|
517 inversion 1.
|
|
adamc@52
|
518 Qed.
|
|
adamc@52
|
519
|
|
adamc@52
|
520 Theorem twoEls_two : forall ls, twoEls ls -> length ls = 2.
|
|
adamc@52
|
521 inversion 1.
|
|
adamc@52
|
522 reflexivity.
|
|
adamc@52
|
523 Qed.
|
|
adamc@52
|
524 End twoEls.
|
|
adamc@52
|
525 (* end thide *)
|
|
adamc@52
|
526
|
|
adamc@49
|
527 (* end hide *)
|
|
adamc@49
|
528
|
|
adamc@50
|
529
|
|
adamc@50
|
530 (** * Recursive Predicates *)
|
|
adamc@50
|
531
|
|
adamc@50
|
532 (** We have already seen all of the ingredients we need to build interesting recursive predicates, like this predicate capturing even-ness. *)
|
|
adamc@50
|
533
|
|
adamc@50
|
534 Inductive even : nat -> Prop :=
|
|
adamc@50
|
535 | EvenO : even O
|
|
adamc@50
|
536 | EvenSS : forall n, even n -> even (S (S n)).
|
|
adamc@50
|
537
|
|
adamc@50
|
538 (** Think of [even] as another judgment defined by natural deduction rules. [EvenO] is a rule with nothing above the line and [even O] below the line, and [EvenSS] is a rule with [even n] above the line and [even (S (S n))] below.
|
|
adamc@50
|
539
|
|
adamc@50
|
540 The proof techniques of the last section are easily adapted. *)
|
|
adamc@50
|
541
|
|
adamc@50
|
542 Theorem even_0 : even 0.
|
|
adamc@55
|
543 (* begin thide *)
|
|
adamc@50
|
544 constructor.
|
|
adamc@55
|
545 (* end thide *)
|
|
adamc@50
|
546 Qed.
|
|
adamc@50
|
547
|
|
adamc@50
|
548 Theorem even_4 : even 4.
|
|
adamc@55
|
549 (* begin thide *)
|
|
adamc@50
|
550 constructor; constructor; constructor.
|
|
adamc@55
|
551 (* end thide *)
|
|
adamc@50
|
552 Qed.
|
|
adamc@50
|
553
|
|
adamc@50
|
554 (** It is not hard to see that sequences of constructor applications like the above can get tedious. We can avoid them using Coq's hint facility. *)
|
|
adamc@50
|
555
|
|
adamc@55
|
556 (* begin thide *)
|
|
adamc@50
|
557 Hint Constructors even.
|
|
adamc@50
|
558
|
|
adamc@50
|
559 Theorem even_4' : even 4.
|
|
adamc@50
|
560 auto.
|
|
adamc@50
|
561 Qed.
|
|
adamc@50
|
562
|
|
adamc@55
|
563 (* end thide *)
|
|
adamc@55
|
564
|
|
adamc@50
|
565 Theorem even_1_contra : even 1 -> False.
|
|
adamc@55
|
566 (* begin thide *)
|
|
adamc@50
|
567 inversion 1.
|
|
adamc@55
|
568 (* end thide *)
|
|
adamc@50
|
569 Qed.
|
|
adamc@50
|
570
|
|
adamc@50
|
571 Theorem even_3_contra : even 3 -> False.
|
|
adamc@55
|
572 (* begin thide *)
|
|
adamc@50
|
573 inversion 1.
|
|
adamc@50
|
574 (** [[
|
|
adamc@50
|
575
|
|
adamc@50
|
576 H : even 3
|
|
adamc@50
|
577 n : nat
|
|
adamc@50
|
578 H1 : even 1
|
|
adamc@50
|
579 H0 : n = 1
|
|
adamc@50
|
580 ============================
|
|
adamc@50
|
581 False
|
|
adamc@50
|
582 ]] *)
|
|
adamc@50
|
583
|
|
adamc@50
|
584 (** [inversion] can be a little overzealous at times, as we can see here with the introduction of the unused variable [n] and an equality hypothesis about it. For more complicated predicates, though, adding such assumptions is critical to dealing with the undecidability of general inversion. *)
|
|
adamc@50
|
585
|
|
adamc@50
|
586 inversion H1.
|
|
adamc@55
|
587 (* end thide *)
|
|
adamc@50
|
588 Qed.
|
|
adamc@50
|
589
|
|
adamc@50
|
590 (** We can also do inductive proofs about [even]. *)
|
|
adamc@50
|
591
|
|
adamc@50
|
592 Theorem even_plus : forall n m, even n -> even m -> even (n + m).
|
|
adamc@55
|
593 (* begin thide *)
|
|
adamc@50
|
594 (** It seems a reasonable first choice to proceed by induction on [n]. *)
|
|
adamc@50
|
595 induction n; crush.
|
|
adamc@50
|
596 (** [[
|
|
adamc@50
|
597
|
|
adamc@50
|
598 n : nat
|
|
adamc@50
|
599 IHn : forall m : nat, even n -> even m -> even (n + m)
|
|
adamc@50
|
600 m : nat
|
|
adamc@50
|
601 H : even (S n)
|
|
adamc@50
|
602 H0 : even m
|
|
adamc@50
|
603 ============================
|
|
adamc@50
|
604 even (S (n + m))
|
|
adamc@50
|
605 ]] *)
|
|
adamc@50
|
606
|
|
adamc@50
|
607 (** We will need to use the hypotheses [H] and [H0] somehow. The most natural choice is to invert [H]. *)
|
|
adamc@50
|
608
|
|
adamc@50
|
609 inversion H.
|
|
adamc@50
|
610 (** [[
|
|
adamc@50
|
611
|
|
adamc@50
|
612 n : nat
|
|
adamc@50
|
613 IHn : forall m : nat, even n -> even m -> even (n + m)
|
|
adamc@50
|
614 m : nat
|
|
adamc@50
|
615 H : even (S n)
|
|
adamc@50
|
616 H0 : even m
|
|
adamc@50
|
617 n0 : nat
|
|
adamc@50
|
618 H2 : even n0
|
|
adamc@50
|
619 H1 : S n0 = n
|
|
adamc@50
|
620 ============================
|
|
adamc@50
|
621 even (S (S n0 + m))
|
|
adamc@50
|
622 ]] *)
|
|
adamc@50
|
623
|
|
adamc@50
|
624 (** Simplifying the conclusion brings us to a point where we can apply a constructor. *)
|
|
adamc@50
|
625 simpl.
|
|
adamc@50
|
626 (** [[
|
|
adamc@50
|
627
|
|
adamc@50
|
628 ============================
|
|
adamc@50
|
629 even (S (S (n0 + m)))
|
|
adamc@50
|
630 ]] *)
|
|
adamc@50
|
631
|
|
adamc@50
|
632 constructor.
|
|
adamc@50
|
633 (** [[
|
|
adamc@50
|
634
|
|
adamc@50
|
635 ============================
|
|
adamc@50
|
636 even (n0 + m)
|
|
adamc@50
|
637 ]] *)
|
|
adamc@50
|
638
|
|
adamc@50
|
639 (** At this point, we would like to apply the inductive hypothesis, which is: *)
|
|
adamc@50
|
640 (** [[
|
|
adamc@50
|
641
|
|
adamc@50
|
642 IHn : forall m : nat, even n -> even m -> even (n + m)
|
|
adamc@50
|
643 ]] *)
|
|
adamc@50
|
644
|
|
adamc@50
|
645 (** Unfortunately, the goal mentions [n0] where it would need to mention [n] to match [IHn]. We could keep looking for a way to finish this proof from here, but it turns out that we can make our lives much easier by changing our basic strategy. Instead of inducting on the structure of [n], we should induct %\textit{%#<i>#on the structure of one of the [even] proofs#</i>#%}%. This technique is commonly called %\textit{%#<i>#rule induction#</i>#%}% in programming language semantics. In the setting of Coq, we have already seen how predicates are defined using the same inductive type mechanism as datatypes, so the fundamental unity of rule induction with "normal" induction is apparent. *)
|
|
adamc@50
|
646
|
|
adamc@50
|
647 Restart.
|
|
adamc@50
|
648
|
|
adamc@50
|
649 induction 1.
|
|
adamc@50
|
650 (** [[
|
|
adamc@50
|
651
|
|
adamc@50
|
652 m : nat
|
|
adamc@50
|
653 ============================
|
|
adamc@50
|
654 even m -> even (0 + m)
|
|
adamc@50
|
655
|
|
adamc@50
|
656 subgoal 2 is:
|
|
adamc@50
|
657 even m -> even (S (S n) + m)
|
|
adamc@50
|
658 ]] *)
|
|
adamc@50
|
659
|
|
adamc@50
|
660 (** The first case is easily discharged by [crush], based on the hint we added earlier to try the constructors of [even]. *)
|
|
adamc@50
|
661
|
|
adamc@50
|
662 crush.
|
|
adamc@50
|
663
|
|
adamc@50
|
664 (** Now we focus on the second case: *)
|
|
adamc@50
|
665 intro.
|
|
adamc@50
|
666
|
|
adamc@50
|
667 (** [[
|
|
adamc@50
|
668
|
|
adamc@50
|
669 m : nat
|
|
adamc@50
|
670 n : nat
|
|
adamc@50
|
671 H : even n
|
|
adamc@50
|
672 IHeven : even m -> even (n + m)
|
|
adamc@50
|
673 H0 : even m
|
|
adamc@50
|
674 ============================
|
|
adamc@50
|
675 even (S (S n) + m)
|
|
adamc@50
|
676 ]] *)
|
|
adamc@50
|
677
|
|
adamc@50
|
678 (** We simplify and apply a constructor, as in our last proof attempt. *)
|
|
adamc@50
|
679
|
|
adamc@50
|
680 simpl; constructor.
|
|
adamc@50
|
681 (** [[
|
|
adamc@50
|
682
|
|
adamc@50
|
683 ============================
|
|
adamc@50
|
684 even (n + m)
|
|
adamc@50
|
685 ]] *)
|
|
adamc@50
|
686
|
|
adamc@50
|
687 (** Now we have an exact match with our inductive hypothesis, and the remainder of the proof is trivial. *)
|
|
adamc@50
|
688
|
|
adamc@50
|
689 apply IHeven; assumption.
|
|
adamc@50
|
690
|
|
adamc@50
|
691 (** In fact, [crush] can handle all of the details of the proof once we declare the induction strategy. *)
|
|
adamc@50
|
692
|
|
adamc@50
|
693 Restart.
|
|
adamc@50
|
694 induction 1; crush.
|
|
adamc@55
|
695 (* end thide *)
|
|
adamc@50
|
696 Qed.
|
|
adamc@50
|
697
|
|
adamc@50
|
698 (** Induction on recursive predicates has similar pitfalls to those we encountered with inversion in the last section. *)
|
|
adamc@50
|
699
|
|
adamc@50
|
700 Theorem even_contra : forall n, even (S (n + n)) -> False.
|
|
adamc@55
|
701 (* begin thide *)
|
|
adamc@50
|
702 induction 1.
|
|
adamc@50
|
703 (** [[
|
|
adamc@50
|
704
|
|
adamc@50
|
705 n : nat
|
|
adamc@50
|
706 ============================
|
|
adamc@50
|
707 False
|
|
adamc@50
|
708
|
|
adamc@50
|
709 subgoal 2 is:
|
|
adamc@50
|
710 False
|
|
adamc@50
|
711 ]] *)
|
|
adamc@50
|
712
|
|
adamc@50
|
713 (** We are already sunk trying to prove the first subgoal, since the argument to [even] was replaced by a fresh variable internally. This time, we find it easiest to prove this theorem by way of a lemma. Instead of trusting [induction] to replace expressions with fresh variables, we do it ourselves, explicitly adding the appropriate equalities as new assumptions. *)
|
|
adamc@50
|
714 Abort.
|
|
adamc@50
|
715
|
|
adamc@50
|
716 Lemma even_contra' : forall n', even n' -> forall n, n' = S (n + n) -> False.
|
|
adamc@50
|
717 induction 1; crush.
|
|
adamc@50
|
718
|
|
adamc@54
|
719 (** At this point, it is useful to consider all cases of [n] and [n0] being zero or nonzero. Only one of these cases has any trickiness to it. *)
|
|
adamc@50
|
720 destruct n; destruct n0; crush.
|
|
adamc@50
|
721
|
|
adamc@50
|
722 (** [[
|
|
adamc@50
|
723
|
|
adamc@50
|
724 n : nat
|
|
adamc@50
|
725 H : even (S n)
|
|
adamc@50
|
726 IHeven : forall n0 : nat, S n = S (n0 + n0) -> False
|
|
adamc@50
|
727 n0 : nat
|
|
adamc@50
|
728 H0 : S n = n0 + S n0
|
|
adamc@50
|
729 ============================
|
|
adamc@50
|
730 False
|
|
adamc@50
|
731 ]] *)
|
|
adamc@50
|
732
|
|
adamc@50
|
733 (** At this point it is useful to use a theorem from the standard library, which we also proved with a different name in the last chapter. *)
|
|
adamc@50
|
734 Check plus_n_Sm.
|
|
adamc@50
|
735 (** [[
|
|
adamc@50
|
736
|
|
adamc@50
|
737 plus_n_Sm
|
|
adamc@50
|
738 : forall n m : nat, S (n + m) = n + S m
|
|
adamc@50
|
739 ]] *)
|
|
adamc@50
|
740
|
|
adamc@50
|
741 rewrite <- plus_n_Sm in H0.
|
|
adamc@50
|
742
|
|
adamc@50
|
743 (** The induction hypothesis lets us complete the proof. *)
|
|
adamc@50
|
744 apply IHeven with n0; assumption.
|
|
adamc@50
|
745
|
|
adamc@202
|
746 (** As usual, we can rewrite the proof to avoid referencing any locally-generated names, which makes our proof script more readable and more robust to changes in the theorem statement. We use the notation [<-] to request a hint that does right-to-left rewriting, just like we can with the [rewrite] tactic. *)
|
|
adamc@50
|
747 Restart.
|
|
adamc@50
|
748 Hint Rewrite <- plus_n_Sm : cpdt.
|
|
adamc@50
|
749
|
|
adamc@50
|
750 induction 1; crush;
|
|
adamc@50
|
751 match goal with
|
|
adamc@50
|
752 | [ H : S ?N = ?N0 + ?N0 |- _ ] => destruct N; destruct N0
|
|
adamc@50
|
753 end; crush; eauto.
|
|
adamc@50
|
754 Qed.
|
|
adamc@50
|
755
|
|
adamc@50
|
756 (** We write the proof in a way that avoids the use of local variable or hypothesis names, using the [match] tactic form to do pattern-matching on the goal. We use unification variables prefixed by question marks in the pattern, and we take advantage of the possibility to mention a unification variable twice in one pattern, to enforce equality between occurrences. The hint to rewrite with [plus_n_Sm] in a particular direction saves us from having to figure out the right place to apply that theorem, and we also take critical advantage of a new tactic, [eauto].
|
|
adamc@50
|
757
|
|
adamc@55
|
758 [crush] uses the tactic [intuition], which, when it runs out of tricks to try using only propositional logic, by default tries the tactic [auto], which we saw in an earlier example. [auto] attempts Prolog-style logic programming, searching through all proof trees up to a certain depth that are built only out of hints that have been registered with [Hint] commands. Compared to Prolog, [auto] places an important restriction: it never introduces new unification variables during search. That is, every time a rule is applied during proof search, all of its arguments must be deducible by studying the form of the goal. [eauto] relaxes this restriction, at the cost of possibly exponentially greater running time. In this particular case, we know that [eauto] has only a small space of proofs to search, so it makes sense to run it. It is common in effectively-automated Coq proofs to see a bag of standard tactics applied to pick off the "easy" subgoals, finishing with [eauto] to handle the tricky parts that can benefit from ad-hoc exhaustive search.
|
|
adamc@50
|
759
|
|
adamc@50
|
760 The original theorem now follows trivially from our lemma. *)
|
|
adamc@50
|
761
|
|
adamc@50
|
762 Theorem even_contra : forall n, even (S (n + n)) -> False.
|
|
adamc@52
|
763 intros; eapply even_contra'; eauto.
|
|
adamc@50
|
764 Qed.
|
|
adamc@52
|
765
|
|
adamc@52
|
766 (** We use a variant [eapply] of [apply] which has the same relationship to [apply] as [eauto] has to [auto]. [apply] only succeeds if all arguments to the rule being used can be determined from the form of the goal, whereas [eapply] will introduce unification variables for undetermined arguments. [eauto] is able to determine the right values for those unification variables.
|
|
adamc@52
|
767
|
|
adamc@52
|
768 By considering an alternate attempt at proving the lemma, we can see another common pitfall of inductive proofs in Coq. Imagine that we had tried to prove [even_contra'] with all of the [forall] quantifiers moved to the front of the lemma statement. *)
|
|
adamc@52
|
769
|
|
adamc@52
|
770 Lemma even_contra'' : forall n' n, even n' -> n' = S (n + n) -> False.
|
|
adamc@52
|
771 induction 1; crush;
|
|
adamc@52
|
772 match goal with
|
|
adamc@52
|
773 | [ H : S ?N = ?N0 + ?N0 |- _ ] => destruct N; destruct N0
|
|
adamc@52
|
774 end; crush; eauto.
|
|
adamc@52
|
775
|
|
adamc@52
|
776 (** One subgoal remains: *)
|
|
adamc@52
|
777
|
|
adamc@52
|
778 (** [[
|
|
adamc@52
|
779
|
|
adamc@52
|
780 n : nat
|
|
adamc@52
|
781 H : even (S (n + n))
|
|
adamc@52
|
782 IHeven : S (n + n) = S (S (S (n + n))) -> False
|
|
adamc@52
|
783 ============================
|
|
adamc@52
|
784 False
|
|
adamc@52
|
785 ]] *)
|
|
adamc@52
|
786
|
|
adamc@52
|
787 (** We are out of luck here. The inductive hypothesis is trivially true, since its assumption is false. In the version of this proof that succeeded, [IHeven] had an explicit quantification over [n]. This is because the quantification of [n] %\textit{%#<i>#appeared after the thing we are inducting on#</i>#%}% in the theorem statement. In general, quantified variables and hypotheses that appear before the induction object in the theorem statement stay fixed throughout the inductive proof. Variables and hypotheses that are quantified after the induction object may be varied explicitly in uses of inductive hypotheses.
|
|
adamc@52
|
788
|
|
adamc@52
|
789 Why should Coq implement [induction] this way? One answer is that it avoids burdening this basic tactic with additional heuristic smarts, but that is not the whole picture. Imagine that [induction] analyzed dependencies among variables and reordered quantifiers to preserve as much freedom as possible in later uses of inductive hypotheses. This could make the inductive hypotheses more complex, which could in turn cause particular automation machinery to fail when it would have succeeded before. In general, we want to avoid quantifiers in our proofs whenever we can, and that goal is furthered by the refactoring that the [induction] tactic forces us to do. *)
|
|
adamc@55
|
790 (* end thide *)
|
|
adamc@51
|
791 Abort.
|
|
adamc@51
|
792
|
|
adamc@52
|
793
|
|
adamc@52
|
794 (* begin hide *)
|
|
adamc@52
|
795 (* In-class exercises *)
|
|
adamc@52
|
796
|
|
adamc@52
|
797 (* EX: Define a type [prop] of simple boolean formulas made up only of truth, falsehood, binary conjunction, and binary disjunction. Define an inductive predicate [holds] that captures when [prop]s are valid, and define a predicate [falseFree] that captures when a [prop] does not contain the "false" formula. Prove that every false-free [prop] is valid. *)
|
|
adamc@52
|
798
|
|
adamc@52
|
799 (* begin thide *)
|
|
adamc@52
|
800 Inductive prop : Set :=
|
|
adamc@52
|
801 | Tru : prop
|
|
adamc@52
|
802 | Fals : prop
|
|
adamc@52
|
803 | And : prop -> prop -> prop
|
|
adamc@52
|
804 | Or : prop -> prop -> prop.
|
|
adamc@52
|
805
|
|
adamc@52
|
806 Inductive holds : prop -> Prop :=
|
|
adamc@52
|
807 | HTru : holds Tru
|
|
adamc@52
|
808 | HAnd : forall p1 p2, holds p1 -> holds p2 -> holds (And p1 p2)
|
|
adamc@52
|
809 | HOr1 : forall p1 p2, holds p1 -> holds (Or p1 p2)
|
|
adamc@52
|
810 | HOr2 : forall p1 p2, holds p2 -> holds (Or p1 p2).
|
|
adamc@52
|
811
|
|
adamc@52
|
812 Inductive falseFree : prop -> Prop :=
|
|
adamc@52
|
813 | FFTru : falseFree Tru
|
|
adamc@52
|
814 | FFAnd : forall p1 p2, falseFree p1 -> falseFree p2 -> falseFree (And p1 p2)
|
|
adamc@52
|
815 | FFNot : forall p1 p2, falseFree p1 -> falseFree p2 -> falseFree (Or p1 p2).
|
|
adamc@52
|
816
|
|
adamc@52
|
817 Hint Constructors holds.
|
|
adamc@52
|
818
|
|
adamc@52
|
819 Theorem falseFree_holds : forall p, falseFree p -> holds p.
|
|
adamc@52
|
820 induction 1; crush.
|
|
adamc@52
|
821 Qed.
|
|
adamc@52
|
822 (* end thide *)
|
|
adamc@52
|
823
|
|
adamc@52
|
824
|
|
adamc@52
|
825 (* EX: Define an inductive type [prop'] that is the same as [prop] but omits the possibility for falsehood. Define a proposition [holds'] for [prop'] that is analogous to [holds]. Define a function [propify] for translating [prop']s to [prop]s. Prove that, for any [prop'] [p], if [propify p] is valid, then so is [p]. *)
|
|
adamc@52
|
826
|
|
adamc@52
|
827 (* begin thide *)
|
|
adamc@52
|
828 Inductive prop' : Set :=
|
|
adamc@52
|
829 | Tru' : prop'
|
|
adamc@52
|
830 | And' : prop' -> prop' -> prop'
|
|
adamc@52
|
831 | Or' : prop' -> prop' -> prop'.
|
|
adamc@52
|
832
|
|
adamc@52
|
833 Inductive holds' : prop' -> Prop :=
|
|
adamc@52
|
834 | HTru' : holds' Tru'
|
|
adamc@52
|
835 | HAnd' : forall p1 p2, holds' p1 -> holds' p2 -> holds' (And' p1 p2)
|
|
adamc@52
|
836 | HOr1' : forall p1 p2, holds' p1 -> holds' (Or' p1 p2)
|
|
adamc@52
|
837 | HOr2' : forall p1 p2, holds' p2 -> holds' (Or' p1 p2).
|
|
adamc@52
|
838
|
|
adamc@52
|
839 Fixpoint propify (p : prop') : prop :=
|
|
adamc@52
|
840 match p with
|
|
adamc@52
|
841 | Tru' => Tru
|
|
adamc@52
|
842 | And' p1 p2 => And (propify p1) (propify p2)
|
|
adamc@52
|
843 | Or' p1 p2 => Or (propify p1) (propify p2)
|
|
adamc@52
|
844 end.
|
|
adamc@52
|
845
|
|
adamc@52
|
846 Hint Constructors holds'.
|
|
adamc@52
|
847
|
|
adamc@52
|
848 Lemma propify_holds' : forall p', holds p' -> forall p, p' = propify p -> holds' p.
|
|
adamc@52
|
849 induction 1; crush; destruct p; crush.
|
|
adamc@52
|
850 Qed.
|
|
adamc@52
|
851
|
|
adamc@52
|
852 Theorem propify_holds : forall p, holds (propify p) -> holds' p.
|
|
adamc@52
|
853 intros; eapply propify_holds'; eauto.
|
|
adamc@52
|
854 Qed.
|
|
adamc@52
|
855 (* end thide *)
|
|
adamc@52
|
856
|
|
adamc@52
|
857 (* end hide *)
|
|
adamc@58
|
858
|
|
adamc@58
|
859
|
|
adamc@58
|
860 (** * Exercises *)
|
|
adamc@58
|
861
|
|
adamc@58
|
862 (** %\begin{enumerate}%#<ol>#
|
|
adamc@58
|
863
|
|
adamc@58
|
864 %\item%#<li># Prove these tautologies of propositional logic, using only the tactics [apply], [assumption], [constructor], [destruct], [intro], [intros], [left], [right], [split], and [unfold].
|
|
adamc@58
|
865 %\begin{enumerate}%#<ol>#
|
|
adamc@58
|
866 %\item%#<li># [(True \/ False) /\ (False \/ True)]#</li>#
|
|
adamc@58
|
867 %\item%#<li># [P -> ~ ~P]#</li>#
|
|
adamc@58
|
868 %\item%#<li># [P /\ (Q \/ R) -> (P /\ Q) \/ (P /\ R)]#</li>#
|
|
adamc@61
|
869 #</ol> </li>#%\end{enumerate}%
|
|
adamc@58
|
870
|
|
adamc@61
|
871 %\item%#<li># Prove the following tautology of first-order logic, using only the tactics [apply], [assert], [assumption], [destruct], [eapply], [eassumption], and %\textit{%#<tt>#exists#</tt>#%}%. You will probably find [assert] useful for stating and proving an intermediate lemma, enabling a kind of "forward reasoning," in contrast to the "backward reasoning" that is the default for Coq tactics. [eassumption] is a version of [assumption] that will do matching of unification variables. Let some variable [T] of type [Set] be the set of individuals. [x] is a constant symbol, [p] is a unary predicate symbol, [q] is a binary predicate symbol, and [f] is a unary function symbol.
|
|
adamc@61
|
872 %\begin{enumerate}%#<ol>#
|
|
adamc@58
|
873 %\item%#<li># [p x -> (forall x, p x -> exists y, q x y) -> (forall x y, q x y -> q y (f y)) -> exists z, q z (f z)]#</li>#
|
|
adamc@58
|
874 #</ol> </li>#%\end{enumerate}%
|
|
adamc@58
|
875
|
|
adamc@59
|
876 %\item%#<li># Define an inductive predicate capturing when a natural number is an integer multiple of either 6 or 10. Prove that 13 does not satisfy your predicate, and prove that any number satisfying the predicate is not odd. It is probably easiest to prove the second theorem by indicating "odd-ness" as equality to [2 * n + 1] for some [n].#</li>#
|
|
adamc@59
|
877
|
|
adamc@60
|
878 %\item%#<li># Define a simple programming language, its semantics, and its typing rules, and then prove that well-typed programs cannot go wrong. Specifically:
|
|
adamc@60
|
879 %\begin{enumerate}%#<ol>#
|
|
adamc@60
|
880 %\item%#<li># Define [var] as a synonym for the natural numbers.#</li>#
|
|
adamc@60
|
881 %\item%#<li># Define an inductive type [exp] of expressions, containing natural number constants, natural number addition, pairing of two other expressions, extraction of the first component of a pair, extraction of the second component of a pair, and variables (based on the [var] type you defined).#</li>#
|
|
adamc@60
|
882 %\item%#<li># Define an inductive type [cmd] of commands, containing expressions and variable assignments. A variable assignment node should contain the variable being assigned, the expression being assigned to it, and the command to run afterward.#</li>#
|
|
adamc@60
|
883 %\item%#<li># Define an inductive type [val] of values, containing natural number constants and pairings of values.#</li>#
|
|
adamc@60
|
884 %\item%#<li># Define a type of variable assignments, which assign a value to each variable.#</li>#
|
|
adamc@60
|
885 %\item%#<li># Define a big-step evaluation relation [eval], capturing what it means for an expression to evaluate to a value under a particular variable assignment. "Big step" means that the evaluation of every expression should evaluatable with a single instance of the inductive predicate you will define. For instance, "[1 + 1] evaluates to [2] under assignment [va]" should be derivable for any assignment [va].#</li>#
|
|
adamc@60
|
886 %\item%#<li># Define a big-step evaluation relation [run], capturing what it means for a command to run to a value under a particular variable assignment. The value of a command is the result of evaluating its final expression.#</li>#
|
|
adamc@60
|
887 %\item%#<li># Define a type of variable typings, which are like variable assignments, but map variables to types instead of values. You might use polymorphism to share some code with your variable assignments.#</li>#
|
|
adamc@60
|
888 %\item%#<li># Define typing judgments for expressions, values, and commands. The expression and command cases will be in terms of a typing assignment.#</li>#
|
|
adamc@60
|
889 %\item%#<li># Define a predicate [varsType] to express when a variable assignment and a variable typing agree on the types of variables.#</li>#
|
|
adamc@60
|
890 %\item%#<li># Prove that any expression that has type [t] under variable typing [vt] evaluates under variable assignment [va] to some value that also has type [t] in [vt], as long as [va] and [vt] agree.#</li>#
|
|
adamc@60
|
891 %\item%#<li># Prove that any command that has type [t] under variable typing [vt] evaluates under variable assignment [va] to some value that also has type [t] in [vt], as long as [va] and [vt] agree.#</li>#
|
|
adamc@60
|
892 #</ol> </li>#%\end{enumerate}%
|
|
adamc@60
|
893 A few hints that may be helpful:
|
|
adamc@60
|
894 %\begin{enumerate}%#<ol>#
|
|
adamc@60
|
895 %\item%#<li># One easy way of defining variable assignments and typings is to define both as instances of a polymorphic map type. The map type at parameter [T] can be defined to be the type of arbitrary functions from variables to [T]. A helpful function for implementing insertion into such a functional map is [eq_nat_dec], which you can make available with [Require Import Arith.]. [eq_nat_dec] has a dependent type that tells you that it makes accurate decisions on whether two natural numbers are equal, but you can use it as if it returned a boolean, e.g., [if eq_nat_dec n m then E1 else E2].#</li>#
|
|
adamc@60
|
896 %\item%#<li># If you follow the last hint, you may find yourself writing a proof that involves an expression with [eq_nat_dec] that you would like to simplify. Running [destruct] on the particular call to [eq_nat_dec] should do the trick. You can automate this advice with a piece of Ltac: [[
|
|
adamc@60
|
897
|
|
adamc@60
|
898 match goal with
|
|
adamc@60
|
899 | [ |- context[eq_nat_dec ?X ?Y] ] => destruct (eq_nat_dec X Y)
|
|
adamc@60
|
900 end
|
|
adamc@60
|
901 ]] #</li>#
|
|
adamc@60
|
902 %\item%#<li># You probably do not want to use an inductive definition for compatibility of variable assignments and typings.#</li>#
|
|
adamc@60
|
903 %\item%#<li># The [Tactics] module from this book contains a variant [crush'] of [crush]. [crush'] takes two arguments. The first argument is a list of lemmas and other functions to be tried automatically in "forward reasoning" style, where we add new facts without being sure yet that they link into a proof of the conclusion. The second argument is a list of predicates on which inverison should be attempted automatically. For instance, running [crush' (lemma1, lemma2) pred] will search for chances to apply [lemma1] and [lemma2] to hypotheses that are already available, adding the new concluded fact if suitable hypotheses can be found. Inversion will be attempted on any hypothesis using [pred], but only those inversions that narrow the field of possibilities to one possible rule will be kept. The format of the list arguments to [crush'] is that you can pass an empty list as [tt], a singleton list as the unadorned single element, and a multiple-element list as a tuple of the elements.#</li>#
|
|
adamc@60
|
904 %\item%#<li># If you want [crush'] to apply polymorphic lemmas, you may have to do a little extra work, if the type parameter is not a free variable of your proof context (so that [crush'] does not know to try it). For instance, if you define a polymorphic map insert function [assign] of some type [forall T : Set, ...], and you want particular applications of [assign] added automatically with type parameter [U], you would need to include [assign] in the lemma list as [assign U] (if you have implicit arguments off) or [assign (T := U)] or [@assign U] (if you have implicit arguments on).#</li>#
|
|
adamc@60
|
905 #</ol> </li>#%\end{enumerate}%
|
|
adamc@60
|
906
|
|
adamc@60
|
907 #</li>#
|
|
adamc@60
|
908
|
|
adamc@58
|
909 #</ol>#%\end{enumerate}% *)
|
