blanchet@49395
|
1 |
(* Title: HOL/Tools/Sledgehammer/sledgehammer_mash.ML
|
blanchet@49263
|
2 |
Author: Jasmin Blanchette, TU Muenchen
|
blanchet@49263
|
3 |
|
blanchet@49263
|
4 |
Sledgehammer's machine-learning-based relevance filter (MaSh).
|
blanchet@49263
|
5 |
*)
|
blanchet@49263
|
6 |
|
blanchet@49396
|
7 |
signature SLEDGEHAMMER_MASH =
|
blanchet@49263
|
8 |
sig
|
blanchet@49266
|
9 |
type stature = ATP_Problem_Generate.stature
|
blanchet@49311
|
10 |
type fact = Sledgehammer_Fact.fact
|
blanchet@49311
|
11 |
type fact_override = Sledgehammer_Fact.fact_override
|
blanchet@49266
|
12 |
type params = Sledgehammer_Provers.params
|
blanchet@49303
|
13 |
type relevance_fudge = Sledgehammer_Provers.relevance_fudge
|
blanchet@49266
|
14 |
type prover_result = Sledgehammer_Provers.prover_result
|
blanchet@49266
|
15 |
|
blanchet@49323
|
16 |
val trace : bool Config.T
|
blanchet@49334
|
17 |
val MaShN : string
|
blanchet@49394
|
18 |
val mepoN : string
|
blanchet@49394
|
19 |
val mashN : string
|
blanchet@49329
|
20 |
val meshN : string
|
blanchet@49407
|
21 |
val unlearnN : string
|
blanchet@49407
|
22 |
val learn_isarN : string
|
blanchet@49407
|
23 |
val learn_atpN : string
|
blanchet@49407
|
24 |
val relearn_isarN : string
|
blanchet@49407
|
25 |
val relearn_atpN : string
|
blanchet@49329
|
26 |
val fact_filters : string list
|
blanchet@49318
|
27 |
val escape_meta : string -> string
|
blanchet@49318
|
28 |
val escape_metas : string list -> string
|
blanchet@49323
|
29 |
val unescape_meta : string -> string
|
blanchet@49323
|
30 |
val unescape_metas : string -> string list
|
blanchet@49421
|
31 |
val extract_query : string -> string * (string * real) list
|
blanchet@49393
|
32 |
val nickname_of : thm -> string
|
blanchet@49421
|
33 |
val suggested_facts :
|
blanchet@49421
|
34 |
(string * 'a) list -> ('b * thm) list -> (('b * thm) * 'a) list
|
blanchet@49336
|
35 |
val mesh_facts :
|
blanchet@49421
|
36 |
int -> ((('a * thm) * real) list * ('a * thm) list) list -> ('a * thm) list
|
blanchet@49266
|
37 |
val theory_ord : theory * theory -> order
|
blanchet@49266
|
38 |
val thm_ord : thm * thm -> order
|
blanchet@49266
|
39 |
val goal_of_thm : theory -> thm -> thm
|
blanchet@49336
|
40 |
val run_prover_for_mash :
|
blanchet@49333
|
41 |
Proof.context -> params -> string -> fact list -> thm -> prover_result
|
blanchet@49407
|
42 |
val features_of :
|
blanchet@49407
|
43 |
Proof.context -> string -> theory -> stature -> term list -> string list
|
blanchet@49419
|
44 |
val isar_dependencies_of : unit Symtab.table -> thm -> string list option
|
blanchet@49407
|
45 |
val atp_dependencies_of :
|
blanchet@49419
|
46 |
Proof.context -> params -> string -> int -> fact list -> unit Symtab.table
|
blanchet@49419
|
47 |
-> thm -> string list option
|
blanchet@49347
|
48 |
val mash_CLEAR : Proof.context -> unit
|
blanchet@49323
|
49 |
val mash_ADD :
|
blanchet@49331
|
50 |
Proof.context -> bool
|
blanchet@49331
|
51 |
-> (string * string list * string list * string list) list -> unit
|
blanchet@49419
|
52 |
val mash_REPROVE :
|
blanchet@49419
|
53 |
Proof.context -> bool -> (string * string list) list -> unit
|
blanchet@49331
|
54 |
val mash_QUERY :
|
blanchet@49421
|
55 |
Proof.context -> bool -> int -> string list * string list
|
blanchet@49421
|
56 |
-> (string * real) list
|
blanchet@49347
|
57 |
val mash_unlearn : Proof.context -> unit
|
blanchet@49333
|
58 |
val mash_could_suggest_facts : unit -> bool
|
blanchet@49336
|
59 |
val mash_can_suggest_facts : Proof.context -> bool
|
blanchet@49421
|
60 |
val mash_suggested_facts :
|
blanchet@49336
|
61 |
Proof.context -> params -> string -> int -> term list -> term
|
blanchet@49421
|
62 |
-> ('a * thm) list -> (('a * thm) * real) list * ('a * thm) list
|
blanchet@49398
|
63 |
val mash_learn_proof :
|
blanchet@49399
|
64 |
Proof.context -> params -> string -> term -> ('a * thm) list -> thm list
|
blanchet@49399
|
65 |
-> unit
|
blanchet@49410
|
66 |
val mash_learn :
|
blanchet@49410
|
67 |
Proof.context -> params -> fact_override -> thm list -> bool -> unit
|
blanchet@49303
|
68 |
val relevant_facts :
|
blanchet@49307
|
69 |
Proof.context -> params -> string -> int -> fact_override -> term list
|
blanchet@49311
|
70 |
-> term -> fact list -> fact list
|
blanchet@49334
|
71 |
val kill_learners : unit -> unit
|
blanchet@49334
|
72 |
val running_learners : unit -> unit
|
blanchet@49263
|
73 |
end;
|
blanchet@49263
|
74 |
|
blanchet@49396
|
75 |
structure Sledgehammer_MaSh : SLEDGEHAMMER_MASH =
|
blanchet@49263
|
76 |
struct
|
blanchet@49264
|
77 |
|
blanchet@49266
|
78 |
open ATP_Util
|
blanchet@49266
|
79 |
open ATP_Problem_Generate
|
blanchet@49266
|
80 |
open Sledgehammer_Util
|
blanchet@49266
|
81 |
open Sledgehammer_Fact
|
blanchet@49266
|
82 |
open Sledgehammer_Provers
|
blanchet@49333
|
83 |
open Sledgehammer_Minimize
|
blanchet@49396
|
84 |
open Sledgehammer_MePo
|
blanchet@49266
|
85 |
|
blanchet@49323
|
86 |
val trace =
|
blanchet@49395
|
87 |
Attrib.setup_config_bool @{binding sledgehammer_mash_trace} (K false)
|
blanchet@49323
|
88 |
fun trace_msg ctxt msg = if Config.get ctxt trace then tracing (msg ()) else ()
|
blanchet@49323
|
89 |
|
blanchet@49334
|
90 |
val MaShN = "MaSh"
|
blanchet@49334
|
91 |
|
blanchet@49394
|
92 |
val mepoN = "mepo"
|
blanchet@49394
|
93 |
val mashN = "mash"
|
blanchet@49329
|
94 |
val meshN = "mesh"
|
blanchet@49329
|
95 |
|
blanchet@49394
|
96 |
val fact_filters = [meshN, mepoN, mashN]
|
blanchet@49329
|
97 |
|
blanchet@49407
|
98 |
val unlearnN = "unlearn"
|
blanchet@49407
|
99 |
val learn_isarN = "learn_isar"
|
blanchet@49407
|
100 |
val learn_atpN = "learn_atp"
|
blanchet@49407
|
101 |
val relearn_isarN = "relearn_isar"
|
blanchet@49407
|
102 |
val relearn_atpN = "relearn_atp"
|
blanchet@49407
|
103 |
|
blanchet@49329
|
104 |
fun mash_home () = getenv "MASH_HOME"
|
blanchet@49409
|
105 |
fun mash_model_dir () =
|
blanchet@49324
|
106 |
getenv "ISABELLE_HOME_USER" ^ "/mash"
|
blanchet@49331
|
107 |
|> tap (Isabelle_System.mkdir o Path.explode)
|
blanchet@49409
|
108 |
val mash_state_dir = mash_model_dir
|
blanchet@49329
|
109 |
fun mash_state_path () = mash_state_dir () ^ "/state" |> Path.explode
|
blanchet@49266
|
110 |
|
blanchet@49345
|
111 |
|
blanchet@49266
|
112 |
(*** Isabelle helpers ***)
|
blanchet@49266
|
113 |
|
blanchet@49323
|
114 |
fun meta_char c =
|
blanchet@49266
|
115 |
if Char.isAlphaNum c orelse c = #"_" orelse c = #"." orelse c = #"(" orelse
|
blanchet@49266
|
116 |
c = #")" orelse c = #"," then
|
blanchet@49266
|
117 |
String.str c
|
blanchet@49266
|
118 |
else
|
blanchet@49266
|
119 |
(* fixed width, in case more digits follow *)
|
blanchet@49410
|
120 |
"%" ^ stringN_of_int 3 (Char.ord c)
|
blanchet@49266
|
121 |
|
blanchet@49323
|
122 |
fun unmeta_chars accum [] = String.implode (rev accum)
|
blanchet@49410
|
123 |
| unmeta_chars accum (#"%" :: d1 :: d2 :: d3 :: cs) =
|
blanchet@49323
|
124 |
(case Int.fromString (String.implode [d1, d2, d3]) of
|
blanchet@49323
|
125 |
SOME n => unmeta_chars (Char.chr n :: accum) cs
|
blanchet@49323
|
126 |
| NONE => "" (* error *))
|
blanchet@49410
|
127 |
| unmeta_chars _ (#"%" :: _) = "" (* error *)
|
blanchet@49323
|
128 |
| unmeta_chars accum (c :: cs) = unmeta_chars (c :: accum) cs
|
blanchet@49323
|
129 |
|
blanchet@49323
|
130 |
val escape_meta = String.translate meta_char
|
blanchet@49318
|
131 |
val escape_metas = map escape_meta #> space_implode " "
|
blanchet@49330
|
132 |
val unescape_meta = String.explode #> unmeta_chars []
|
blanchet@49330
|
133 |
val unescape_metas =
|
blanchet@49330
|
134 |
space_explode " " #> filter_out (curry (op =) "") #> map unescape_meta
|
blanchet@49266
|
135 |
|
blanchet@49421
|
136 |
fun extract_node line =
|
blanchet@49421
|
137 |
case space_explode ":" line of
|
blanchet@49421
|
138 |
[name, parents] => (unescape_meta name, unescape_metas parents)
|
blanchet@49421
|
139 |
| _ => ("", [])
|
blanchet@49421
|
140 |
|
blanchet@49421
|
141 |
fun extract_suggestion sugg =
|
blanchet@49421
|
142 |
case space_explode "=" sugg of
|
blanchet@49421
|
143 |
[name, weight] =>
|
blanchet@49421
|
144 |
SOME (unescape_meta name, Real.fromString weight |> the_default 0.0)
|
blanchet@49421
|
145 |
| _ => NONE
|
blanchet@49421
|
146 |
|
blanchet@49326
|
147 |
fun extract_query line =
|
blanchet@49326
|
148 |
case space_explode ":" line of
|
blanchet@49421
|
149 |
[goal, suggs] =>
|
blanchet@49421
|
150 |
(unescape_meta goal,
|
blanchet@49421
|
151 |
map_filter extract_suggestion (space_explode " " suggs))
|
blanchet@49327
|
152 |
| _ => ("", [])
|
blanchet@49326
|
153 |
|
blanchet@49393
|
154 |
fun parent_of_local_thm th =
|
blanchet@49393
|
155 |
let
|
blanchet@49393
|
156 |
val thy = th |> Thm.theory_of_thm
|
blanchet@49393
|
157 |
val facts = thy |> Global_Theory.facts_of
|
blanchet@49393
|
158 |
val space = facts |> Facts.space_of
|
blanchet@49393
|
159 |
fun id_of s = #id (Name_Space.the_entry space s)
|
blanchet@49393
|
160 |
fun max_id (s', _) (s, id) =
|
blanchet@49393
|
161 |
let val id' = id_of s' in if id > id' then (s, id) else (s', id') end
|
blanchet@49393
|
162 |
in ("", ~1) |> Facts.fold_static max_id facts |> fst end
|
blanchet@49393
|
163 |
|
blanchet@49393
|
164 |
val local_prefix = "local" ^ Long_Name.separator
|
blanchet@49393
|
165 |
|
blanchet@49393
|
166 |
fun nickname_of th =
|
blanchet@49409
|
167 |
if Thm.has_name_hint th then
|
blanchet@49409
|
168 |
let val hint = Thm.get_name_hint th in
|
blanchet@49409
|
169 |
(* FIXME: There must be a better way to detect local facts. *)
|
blanchet@49409
|
170 |
case try (unprefix local_prefix) hint of
|
blanchet@49409
|
171 |
SOME suf =>
|
blanchet@49409
|
172 |
parent_of_local_thm th ^ Long_Name.separator ^ Long_Name.separator ^ suf
|
blanchet@49409
|
173 |
| NONE => hint
|
blanchet@49409
|
174 |
end
|
blanchet@49409
|
175 |
else
|
blanchet@49409
|
176 |
backquote_thm th
|
blanchet@49393
|
177 |
|
blanchet@49345
|
178 |
fun suggested_facts suggs facts =
|
blanchet@49345
|
179 |
let
|
blanchet@49393
|
180 |
fun add_fact (fact as (_, th)) = Symtab.default (nickname_of th, fact)
|
blanchet@49345
|
181 |
val tab = Symtab.empty |> fold add_fact facts
|
blanchet@49421
|
182 |
fun find_sugg (name, weight) =
|
blanchet@49421
|
183 |
Symtab.lookup tab name |> Option.map (rpair weight)
|
blanchet@49421
|
184 |
in map_filter find_sugg suggs end
|
blanchet@49326
|
185 |
|
blanchet@49421
|
186 |
fun sum_avg [] = 0
|
blanchet@49422
|
187 |
| sum_avg xs =
|
blanchet@49422
|
188 |
Real.ceil (100000000.0 * fold (curry (op +)) xs 0.0) div length xs
|
blanchet@49343
|
189 |
|
blanchet@49421
|
190 |
fun normalize_scores [] = []
|
blanchet@49421
|
191 |
| normalize_scores ((fact, score) :: tail) =
|
blanchet@49421
|
192 |
(fact, 1.0) :: map (apsnd (curry Real.* (1.0 / score))) tail
|
blanchet@49328
|
193 |
|
blanchet@49421
|
194 |
fun mesh_facts max_facts [(sels, unks)] =
|
blanchet@49421
|
195 |
map fst (take max_facts sels) @ take (max_facts - length sels) unks
|
blanchet@49329
|
196 |
| mesh_facts max_facts mess =
|
blanchet@49329
|
197 |
let
|
blanchet@49421
|
198 |
val mess = mess |> map (apfst (normalize_scores #> `length))
|
blanchet@49329
|
199 |
val fact_eq = Thm.eq_thm o pairself snd
|
blanchet@49421
|
200 |
fun score_at sels = try (nth sels) #> Option.map snd
|
blanchet@49335
|
201 |
fun score_in fact ((sel_len, sels), unks) =
|
blanchet@49421
|
202 |
case find_index (curry fact_eq fact o fst) sels of
|
blanchet@49335
|
203 |
~1 => (case find_index (curry fact_eq fact) unks of
|
blanchet@49421
|
204 |
~1 => score_at sels sel_len
|
blanchet@49335
|
205 |
| _ => NONE)
|
blanchet@49421
|
206 |
| rank => score_at sels rank
|
blanchet@49421
|
207 |
fun weight_of fact = mess |> map_filter (score_in fact) |> sum_avg
|
blanchet@49421
|
208 |
val facts =
|
blanchet@49421
|
209 |
fold (union fact_eq o map fst o take max_facts o snd o fst) mess []
|
blanchet@49329
|
210 |
in
|
blanchet@49421
|
211 |
facts |> map (`weight_of) |> sort (int_ord o swap o pairself fst)
|
blanchet@49343
|
212 |
|> map snd |> take max_facts
|
blanchet@49329
|
213 |
end
|
blanchet@49327
|
214 |
|
blanchet@49410
|
215 |
val thy_feature_name_of = prefix "y"
|
blanchet@49410
|
216 |
val const_name_of = prefix "c"
|
blanchet@49410
|
217 |
val type_name_of = prefix "t"
|
blanchet@49410
|
218 |
val class_name_of = prefix "s"
|
blanchet@49266
|
219 |
|
blanchet@49339
|
220 |
fun theory_ord p =
|
blanchet@49339
|
221 |
if Theory.eq_thy p then
|
blanchet@49339
|
222 |
EQUAL
|
blanchet@49339
|
223 |
else if Theory.subthy p then
|
blanchet@49339
|
224 |
LESS
|
blanchet@49339
|
225 |
else if Theory.subthy (swap p) then
|
blanchet@49339
|
226 |
GREATER
|
blanchet@49339
|
227 |
else case int_ord (pairself (length o Theory.ancestors_of) p) of
|
blanchet@49339
|
228 |
EQUAL => string_ord (pairself Context.theory_name p)
|
blanchet@49339
|
229 |
| order => order
|
blanchet@49339
|
230 |
|
blanchet@49339
|
231 |
val thm_ord = theory_ord o pairself theory_of_thm
|
blanchet@49339
|
232 |
|
blanchet@49407
|
233 |
val freezeT = Type.legacy_freeze_type
|
blanchet@49407
|
234 |
|
blanchet@49407
|
235 |
fun freeze (t $ u) = freeze t $ freeze u
|
blanchet@49407
|
236 |
| freeze (Abs (s, T, t)) = Abs (s, freezeT T, freeze t)
|
blanchet@49407
|
237 |
| freeze (Var ((s, _), T)) = Free (s, freezeT T)
|
blanchet@49407
|
238 |
| freeze (Const (s, T)) = Const (s, freezeT T)
|
blanchet@49407
|
239 |
| freeze (Free (s, T)) = Free (s, freezeT T)
|
blanchet@49407
|
240 |
| freeze t = t
|
blanchet@49407
|
241 |
|
blanchet@49407
|
242 |
fun goal_of_thm thy = prop_of #> freeze #> cterm_of thy #> Goal.init
|
blanchet@49407
|
243 |
|
blanchet@49407
|
244 |
fun run_prover_for_mash ctxt params prover facts goal =
|
blanchet@49407
|
245 |
let
|
blanchet@49407
|
246 |
val problem =
|
blanchet@49407
|
247 |
{state = Proof.init ctxt, goal = goal, subgoal = 1, subgoal_count = 1,
|
blanchet@49407
|
248 |
facts = facts |> map (apfst (apfst (fn name => name ())))
|
blanchet@49407
|
249 |
|> map Untranslated_Fact}
|
blanchet@49407
|
250 |
in
|
blanchet@49414
|
251 |
get_minimizing_prover ctxt MaSh (K (K ())) prover params (K (K (K "")))
|
blanchet@49407
|
252 |
problem
|
blanchet@49407
|
253 |
end
|
blanchet@49407
|
254 |
|
blanchet@49341
|
255 |
val bad_types = [@{type_name prop}, @{type_name bool}, @{type_name fun}]
|
blanchet@49341
|
256 |
|
blanchet@49413
|
257 |
val logical_consts =
|
blanchet@49413
|
258 |
[@{const_name prop}, @{const_name Pure.conjunction}] @ atp_logical_consts
|
blanchet@49413
|
259 |
|
blanchet@49333
|
260 |
fun interesting_terms_types_and_classes ctxt prover term_max_depth
|
blanchet@49333
|
261 |
type_max_depth ts =
|
blanchet@49266
|
262 |
let
|
blanchet@49333
|
263 |
fun is_bad_const (x as (s, _)) args =
|
blanchet@49413
|
264 |
member (op =) logical_consts s orelse
|
blanchet@49333
|
265 |
fst (is_built_in_const_for_prover ctxt prover x args)
|
blanchet@49319
|
266 |
fun add_classes @{sort type} = I
|
blanchet@49319
|
267 |
| add_classes S = union (op =) (map class_name_of S)
|
blanchet@49266
|
268 |
fun do_add_type (Type (s, Ts)) =
|
blanchet@49266
|
269 |
(not (member (op =) bad_types s) ? insert (op =) (type_name_of s))
|
blanchet@49266
|
270 |
#> fold do_add_type Ts
|
blanchet@49319
|
271 |
| do_add_type (TFree (_, S)) = add_classes S
|
blanchet@49319
|
272 |
| do_add_type (TVar (_, S)) = add_classes S
|
blanchet@49266
|
273 |
fun add_type T = type_max_depth >= 0 ? do_add_type T
|
blanchet@49266
|
274 |
fun mk_app s args =
|
blanchet@49266
|
275 |
if member (op <>) args "" then s ^ "(" ^ space_implode "," args ^ ")"
|
blanchet@49266
|
276 |
else s
|
blanchet@49266
|
277 |
fun patternify ~1 _ = ""
|
blanchet@49266
|
278 |
| patternify depth t =
|
blanchet@49266
|
279 |
case strip_comb t of
|
blanchet@49413
|
280 |
(Const (x as (s, _)), args) =>
|
blanchet@49413
|
281 |
if is_bad_const x args then ""
|
blanchet@49413
|
282 |
else mk_app (const_name_of s) (map (patternify (depth - 1)) args)
|
blanchet@49266
|
283 |
| _ => ""
|
blanchet@49421
|
284 |
fun add_pattern depth t =
|
blanchet@49421
|
285 |
case patternify depth t of "" => I | s => insert (op =) s
|
blanchet@49266
|
286 |
fun add_term_patterns ~1 _ = I
|
blanchet@49266
|
287 |
| add_term_patterns depth t =
|
blanchet@49421
|
288 |
add_pattern depth t #> add_term_patterns (depth - 1) t
|
blanchet@49266
|
289 |
val add_term = add_term_patterns term_max_depth
|
blanchet@49266
|
290 |
fun add_patterns t =
|
blanchet@49266
|
291 |
let val (head, args) = strip_comb t in
|
blanchet@49266
|
292 |
(case head of
|
blanchet@49413
|
293 |
Const (_, T) => add_term t #> add_type T
|
blanchet@49266
|
294 |
| Free (_, T) => add_type T
|
blanchet@49266
|
295 |
| Var (_, T) => add_type T
|
blanchet@49266
|
296 |
| Abs (_, T, body) => add_type T #> add_patterns body
|
blanchet@49266
|
297 |
| _ => I)
|
blanchet@49266
|
298 |
#> fold add_patterns args
|
blanchet@49266
|
299 |
end
|
blanchet@49341
|
300 |
in [] |> fold add_patterns ts end
|
blanchet@49266
|
301 |
|
blanchet@49266
|
302 |
fun is_exists (s, _) = (s = @{const_name Ex} orelse s = @{const_name Ex1})
|
blanchet@49266
|
303 |
|
blanchet@49312
|
304 |
val term_max_depth = 1
|
blanchet@49312
|
305 |
val type_max_depth = 1
|
blanchet@49266
|
306 |
|
blanchet@49266
|
307 |
(* TODO: Generate type classes for types? *)
|
blanchet@49400
|
308 |
fun features_of ctxt prover thy (scope, status) ts =
|
blanchet@49318
|
309 |
thy_feature_name_of (Context.theory_name thy) ::
|
blanchet@49333
|
310 |
interesting_terms_types_and_classes ctxt prover term_max_depth type_max_depth
|
blanchet@49333
|
311 |
ts
|
blanchet@49347
|
312 |
|> forall is_lambda_free ts ? cons "no_lams"
|
blanchet@49347
|
313 |
|> forall (not o exists_Const is_exists) ts ? cons "no_skos"
|
blanchet@49400
|
314 |
|> scope <> Global ? cons "local"
|
blanchet@49317
|
315 |
|> (case status of
|
blanchet@49317
|
316 |
General => I
|
blanchet@49317
|
317 |
| Induction => cons "induction"
|
blanchet@49317
|
318 |
| Intro => cons "intro"
|
blanchet@49317
|
319 |
| Inductive => cons "inductive"
|
blanchet@49317
|
320 |
| Elim => cons "elim"
|
blanchet@49317
|
321 |
| Simp => cons "simp"
|
blanchet@49317
|
322 |
| Def => cons "def")
|
blanchet@49266
|
323 |
|
blanchet@49419
|
324 |
(* Too many dependencies is a sign that a decision procedure is at work. There
|
blanchet@49419
|
325 |
isn't much too learn from such proofs. *)
|
blanchet@49419
|
326 |
val max_dependencies = 10
|
blanchet@49419
|
327 |
val atp_dependency_default_max_fact = 50
|
blanchet@49266
|
328 |
|
blanchet@49419
|
329 |
fun trim_dependencies deps =
|
blanchet@49419
|
330 |
if length deps <= max_dependencies then SOME deps else NONE
|
blanchet@49266
|
331 |
|
blanchet@49421
|
332 |
fun isar_dependencies_of all_names =
|
blanchet@49421
|
333 |
thms_in_proof (SOME all_names) #> trim_dependencies
|
blanchet@49419
|
334 |
|
blanchet@49419
|
335 |
fun atp_dependencies_of ctxt (params as {verbose, max_facts, ...}) prover
|
blanchet@49419
|
336 |
auto_level facts all_names th =
|
blanchet@49407
|
337 |
case isar_dependencies_of all_names th of
|
blanchet@49419
|
338 |
SOME [] => NONE
|
blanchet@49407
|
339 |
| isar_deps =>
|
blanchet@49407
|
340 |
let
|
blanchet@49407
|
341 |
val thy = Proof_Context.theory_of ctxt
|
blanchet@49407
|
342 |
val goal = goal_of_thm thy th
|
blanchet@49407
|
343 |
val (_, hyp_ts, concl_t) = ATP_Util.strip_subgoal ctxt goal 1
|
blanchet@49407
|
344 |
val facts = facts |> filter (fn (_, th') => thm_ord (th', th) = LESS)
|
blanchet@49407
|
345 |
fun fix_name ((_, stature), th) = ((fn () => nickname_of th, stature), th)
|
blanchet@49407
|
346 |
fun is_dep dep (_, th) = nickname_of th = dep
|
blanchet@49407
|
347 |
fun add_isar_dep facts dep accum =
|
blanchet@49407
|
348 |
if exists (is_dep dep) accum then
|
blanchet@49407
|
349 |
accum
|
blanchet@49407
|
350 |
else case find_first (is_dep dep) facts of
|
blanchet@49407
|
351 |
SOME ((name, status), th) => accum @ [((name, status), th)]
|
blanchet@49407
|
352 |
| NONE => accum (* shouldn't happen *)
|
blanchet@49407
|
353 |
val facts =
|
blanchet@49421
|
354 |
facts |> mepo_suggested_facts ctxt params prover
|
blanchet@49419
|
355 |
(max_facts |> the_default atp_dependency_default_max_fact)
|
blanchet@49419
|
356 |
NONE hyp_ts concl_t
|
blanchet@49419
|
357 |
|> fold (add_isar_dep facts) (these isar_deps)
|
blanchet@49407
|
358 |
|> map fix_name
|
blanchet@49407
|
359 |
in
|
blanchet@49419
|
360 |
if verbose andalso auto_level = 0 then
|
blanchet@49407
|
361 |
let val num_facts = length facts in
|
blanchet@49407
|
362 |
"MaSh: " ^ quote prover ^ " on " ^ quote (nickname_of th) ^
|
blanchet@49407
|
363 |
" with " ^ string_of_int num_facts ^ " fact" ^ plural_s num_facts ^
|
blanchet@49407
|
364 |
"."
|
blanchet@49407
|
365 |
|> Output.urgent_message
|
blanchet@49407
|
366 |
end
|
blanchet@49407
|
367 |
else
|
blanchet@49407
|
368 |
();
|
blanchet@49407
|
369 |
case run_prover_for_mash ctxt params prover facts goal of
|
blanchet@49407
|
370 |
{outcome = NONE, used_facts, ...} =>
|
blanchet@49419
|
371 |
(if verbose andalso auto_level = 0 then
|
blanchet@49407
|
372 |
let val num_facts = length used_facts in
|
blanchet@49407
|
373 |
"Found proof with " ^ string_of_int num_facts ^ " fact" ^
|
blanchet@49407
|
374 |
plural_s num_facts ^ "."
|
blanchet@49407
|
375 |
|> Output.urgent_message
|
blanchet@49407
|
376 |
end
|
blanchet@49407
|
377 |
else
|
blanchet@49407
|
378 |
();
|
blanchet@49419
|
379 |
used_facts |> map fst |> trim_dependencies)
|
blanchet@49419
|
380 |
| _ => NONE
|
blanchet@49407
|
381 |
end
|
blanchet@49266
|
382 |
|
blanchet@49266
|
383 |
|
blanchet@49317
|
384 |
(*** Low-level communication with MaSh ***)
|
blanchet@49317
|
385 |
|
blanchet@49409
|
386 |
(* more friendly than "try o File.rm" for those who keep the files open in their
|
blanchet@49409
|
387 |
text editor *)
|
blanchet@49409
|
388 |
fun wipe_out file = File.write file ""
|
blanchet@49409
|
389 |
|
blanchet@49338
|
390 |
fun write_file (xs, f) file =
|
blanchet@49333
|
391 |
let val path = Path.explode file in
|
blanchet@49409
|
392 |
wipe_out path;
|
blanchet@49338
|
393 |
xs |> chunk_list 500
|
blanchet@49338
|
394 |
|> List.app (File.append path o space_implode "" o map f)
|
blanchet@49333
|
395 |
end
|
blanchet@49331
|
396 |
|
blanchet@49409
|
397 |
fun run_mash_tool ctxt overlord save max_suggs write_cmds read_suggs =
|
blanchet@49326
|
398 |
let
|
blanchet@49409
|
399 |
val (temp_dir, serial) =
|
blanchet@49409
|
400 |
if overlord then (getenv "ISABELLE_HOME_USER", "")
|
blanchet@49409
|
401 |
else (getenv "ISABELLE_TMP", serial_string ())
|
blanchet@49409
|
402 |
val log_file = if overlord then temp_dir ^ "/mash_log" else "/dev/null"
|
blanchet@49331
|
403 |
val err_file = temp_dir ^ "/mash_err" ^ serial
|
blanchet@49333
|
404 |
val sugg_file = temp_dir ^ "/mash_suggs" ^ serial
|
blanchet@49331
|
405 |
val cmd_file = temp_dir ^ "/mash_commands" ^ serial
|
blanchet@49409
|
406 |
val core =
|
blanchet@49409
|
407 |
"--inputFile " ^ cmd_file ^ " --predictions " ^ sugg_file ^
|
blanchet@49409
|
408 |
" --numberOfPredictions " ^ string_of_int max_suggs ^
|
blanchet@49409
|
409 |
(if save then " --saveModel" else "")
|
blanchet@49409
|
410 |
val command =
|
blanchet@49409
|
411 |
mash_home () ^ "/mash --quiet --outputDir " ^ mash_model_dir () ^
|
blanchet@49409
|
412 |
" --log " ^ log_file ^ " " ^ core ^ " >& " ^ err_file
|
blanchet@49331
|
413 |
in
|
blanchet@49338
|
414 |
write_file ([], K "") sugg_file;
|
blanchet@49331
|
415 |
write_file write_cmds cmd_file;
|
blanchet@49409
|
416 |
trace_msg ctxt (fn () => "Running " ^ command);
|
blanchet@49409
|
417 |
Isabelle_System.bash command;
|
blanchet@49409
|
418 |
read_suggs (fn () => try File.read_lines (Path.explode sugg_file) |> these)
|
blanchet@49409
|
419 |
|> tap (fn _ => trace_msg ctxt (fn () =>
|
blanchet@49409
|
420 |
case try File.read (Path.explode err_file) of
|
blanchet@49409
|
421 |
NONE => "Done"
|
blanchet@49409
|
422 |
| SOME "" => "Done"
|
blanchet@49409
|
423 |
| SOME s => "Error: " ^ elide_string 1000 s))
|
blanchet@49409
|
424 |
|> not overlord
|
blanchet@49409
|
425 |
? tap (fn _ => List.app (wipe_out o Path.explode)
|
blanchet@49409
|
426 |
[err_file, sugg_file, cmd_file])
|
blanchet@49331
|
427 |
end
|
blanchet@49331
|
428 |
|
blanchet@49419
|
429 |
fun str_of_add (name, parents, feats, deps) =
|
blanchet@49331
|
430 |
"! " ^ escape_meta name ^ ": " ^ escape_metas parents ^ "; " ^
|
blanchet@49326
|
431 |
escape_metas feats ^ "; " ^ escape_metas deps ^ "\n"
|
blanchet@49326
|
432 |
|
blanchet@49419
|
433 |
fun str_of_reprove (name, deps) =
|
blanchet@49419
|
434 |
"p " ^ escape_meta name ^ ": " ^ escape_metas deps ^ "\n"
|
blanchet@49419
|
435 |
|
blanchet@49331
|
436 |
fun str_of_query (parents, feats) =
|
blanchet@49421
|
437 |
"? " ^ escape_metas parents ^ "; " ^ escape_metas feats ^ "\n"
|
blanchet@49331
|
438 |
|
blanchet@49347
|
439 |
fun mash_CLEAR ctxt =
|
blanchet@49409
|
440 |
let val path = mash_model_dir () |> Path.explode in
|
blanchet@49347
|
441 |
trace_msg ctxt (K "MaSh CLEAR");
|
blanchet@49409
|
442 |
File.fold_dir (fn file => fn _ =>
|
blanchet@49409
|
443 |
try File.rm (Path.append path (Path.basic file)))
|
blanchet@49409
|
444 |
path NONE;
|
blanchet@49409
|
445 |
()
|
blanchet@49324
|
446 |
end
|
blanchet@49317
|
447 |
|
blanchet@49331
|
448 |
fun mash_ADD _ _ [] = ()
|
blanchet@49419
|
449 |
| mash_ADD ctxt overlord adds =
|
blanchet@49331
|
450 |
(trace_msg ctxt (fn () => "MaSh ADD " ^
|
blanchet@49419
|
451 |
elide_string 1000 (space_implode " " (map #1 adds)));
|
blanchet@49419
|
452 |
run_mash_tool ctxt overlord true 0 (adds, str_of_add) (K ()))
|
blanchet@49419
|
453 |
|
blanchet@49419
|
454 |
fun mash_REPROVE _ _ [] = ()
|
blanchet@49419
|
455 |
| mash_REPROVE ctxt overlord reps =
|
blanchet@49419
|
456 |
(trace_msg ctxt (fn () => "MaSh REPROVE " ^
|
blanchet@49419
|
457 |
elide_string 1000 (space_implode " " (map #1 reps)));
|
blanchet@49419
|
458 |
run_mash_tool ctxt overlord true 0 (reps, str_of_reprove) (K ()))
|
blanchet@49317
|
459 |
|
blanchet@49333
|
460 |
fun mash_QUERY ctxt overlord max_suggs (query as (_, feats)) =
|
blanchet@49329
|
461 |
(trace_msg ctxt (fn () => "MaSh QUERY " ^ space_implode " " feats);
|
blanchet@49409
|
462 |
run_mash_tool ctxt overlord false max_suggs
|
blanchet@49338
|
463 |
([query], str_of_query)
|
blanchet@49409
|
464 |
(fn suggs =>
|
blanchet@49409
|
465 |
case suggs () of
|
blanchet@49409
|
466 |
[] => []
|
blanchet@49409
|
467 |
| suggs => snd (extract_query (List.last suggs)))
|
blanchet@49326
|
468 |
handle List.Empty => [])
|
blanchet@49317
|
469 |
|
blanchet@49317
|
470 |
|
blanchet@49266
|
471 |
(*** High-level communication with MaSh ***)
|
blanchet@49266
|
472 |
|
blanchet@49336
|
473 |
fun try_graph ctxt when def f =
|
blanchet@49336
|
474 |
f ()
|
blanchet@49336
|
475 |
handle Graph.CYCLES (cycle :: _) =>
|
blanchet@49336
|
476 |
(trace_msg ctxt (fn () =>
|
blanchet@49336
|
477 |
"Cycle involving " ^ commas cycle ^ " when " ^ when); def)
|
blanchet@49407
|
478 |
| Graph.DUP name =>
|
blanchet@49407
|
479 |
(trace_msg ctxt (fn () =>
|
blanchet@49407
|
480 |
"Duplicate fact " ^ quote name ^ " when " ^ when); def)
|
blanchet@49336
|
481 |
| Graph.UNDEF name =>
|
blanchet@49336
|
482 |
(trace_msg ctxt (fn () =>
|
blanchet@49336
|
483 |
"Unknown fact " ^ quote name ^ " when " ^ when); def)
|
blanchet@49407
|
484 |
| exn =>
|
blanchet@49407
|
485 |
if Exn.is_interrupt exn then
|
blanchet@49407
|
486 |
reraise exn
|
blanchet@49407
|
487 |
else
|
blanchet@49407
|
488 |
(trace_msg ctxt (fn () =>
|
blanchet@49407
|
489 |
"Internal error when " ^ when ^ ":\n" ^
|
blanchet@49407
|
490 |
ML_Compiler.exn_message exn); def)
|
blanchet@49336
|
491 |
|
blanchet@49421
|
492 |
fun graph_info G =
|
blanchet@49421
|
493 |
string_of_int (length (Graph.keys G)) ^ " node(s), " ^
|
blanchet@49421
|
494 |
string_of_int (fold (Integer.add o length o snd) (Graph.dest G) 0) ^
|
blanchet@49421
|
495 |
" edge(s), " ^
|
blanchet@49421
|
496 |
string_of_int (length (Graph.minimals G)) ^ " minimal, " ^
|
blanchet@49421
|
497 |
string_of_int (length (Graph.maximals G)) ^ " maximal"
|
blanchet@49421
|
498 |
|
blanchet@49415
|
499 |
type mash_state = {fact_G : unit Graph.T}
|
blanchet@49264
|
500 |
|
blanchet@49415
|
501 |
val empty_state = {fact_G = Graph.empty}
|
blanchet@49316
|
502 |
|
blanchet@49316
|
503 |
local
|
blanchet@49316
|
504 |
|
blanchet@49405
|
505 |
val version = "*** MaSh 0.0 ***"
|
blanchet@49405
|
506 |
|
blanchet@49405
|
507 |
fun load _ (state as (true, _)) = state
|
blanchet@49405
|
508 |
| load ctxt _ =
|
blanchet@49324
|
509 |
let val path = mash_state_path () in
|
blanchet@49317
|
510 |
(true,
|
blanchet@49317
|
511 |
case try File.read_lines path of
|
blanchet@49421
|
512 |
SOME (version' :: node_lines) =>
|
blanchet@49317
|
513 |
let
|
blanchet@49337
|
514 |
fun add_edge_to name parent =
|
blanchet@49421
|
515 |
Graph.default_node (parent, ()) #> Graph.add_edge (parent, name)
|
blanchet@49421
|
516 |
fun add_node line =
|
blanchet@49421
|
517 |
case extract_node line of
|
blanchet@49331
|
518 |
("", _) => I (* shouldn't happen *)
|
blanchet@49331
|
519 |
| (name, parents) =>
|
blanchet@49421
|
520 |
Graph.default_node (name, ()) #> fold (add_edge_to name) parents
|
blanchet@49415
|
521 |
val fact_G =
|
blanchet@49337
|
522 |
try_graph ctxt "loading state" Graph.empty (fn () =>
|
blanchet@49421
|
523 |
Graph.empty |> version' = version ? fold add_node node_lines)
|
blanchet@49421
|
524 |
in
|
blanchet@49421
|
525 |
trace_msg ctxt (fn () =>
|
blanchet@49421
|
526 |
"Loaded fact graph (" ^ graph_info fact_G ^ ")");
|
blanchet@49421
|
527 |
{fact_G = fact_G}
|
blanchet@49421
|
528 |
end
|
blanchet@49319
|
529 |
| _ => empty_state)
|
blanchet@49317
|
530 |
end
|
blanchet@49316
|
531 |
|
blanchet@49421
|
532 |
fun save ctxt {fact_G} =
|
blanchet@49316
|
533 |
let
|
blanchet@49324
|
534 |
val path = mash_state_path ()
|
blanchet@49333
|
535 |
fun fact_line_for name parents =
|
blanchet@49333
|
536 |
escape_meta name ^ ": " ^ escape_metas parents
|
blanchet@49331
|
537 |
val append_fact = File.append path o suffix "\n" oo fact_line_for
|
blanchet@49415
|
538 |
fun append_entry (name, ((), (parents, _))) () =
|
blanchet@49415
|
539 |
append_fact name (Graph.Keys.dest parents)
|
blanchet@49316
|
540 |
in
|
blanchet@49405
|
541 |
File.write path (version ^ "\n");
|
blanchet@49421
|
542 |
Graph.fold append_entry fact_G ();
|
blanchet@49421
|
543 |
trace_msg ctxt (fn () => "Saved fact graph (" ^ graph_info fact_G ^ ")")
|
blanchet@49316
|
544 |
end
|
blanchet@49316
|
545 |
|
blanchet@49317
|
546 |
val global_state =
|
blanchet@49396
|
547 |
Synchronized.var "Sledgehammer_MaSh.global_state" (false, empty_state)
|
blanchet@49316
|
548 |
|
blanchet@49316
|
549 |
in
|
blanchet@49316
|
550 |
|
blanchet@49336
|
551 |
fun mash_map ctxt f =
|
blanchet@49421
|
552 |
Synchronized.change global_state (load ctxt ##> (f #> tap (save ctxt)))
|
blanchet@49316
|
553 |
|
blanchet@49449
|
554 |
fun mash_peek ctxt f =
|
blanchet@49449
|
555 |
Synchronized.change_result global_state (load ctxt #> `snd #>> f)
|
blanchet@49449
|
556 |
|
blanchet@49336
|
557 |
fun mash_get ctxt =
|
blanchet@49405
|
558 |
Synchronized.change_result global_state (load ctxt #> `snd)
|
blanchet@49317
|
559 |
|
blanchet@49347
|
560 |
fun mash_unlearn ctxt =
|
blanchet@49317
|
561 |
Synchronized.change global_state (fn _ =>
|
blanchet@49409
|
562 |
(mash_CLEAR ctxt; wipe_out (mash_state_path ()); (true, empty_state)))
|
blanchet@49316
|
563 |
|
blanchet@49316
|
564 |
end
|
blanchet@49316
|
565 |
|
blanchet@49333
|
566 |
fun mash_could_suggest_facts () = mash_home () <> ""
|
blanchet@49415
|
567 |
fun mash_can_suggest_facts ctxt = not (Graph.is_empty (#fact_G (mash_get ctxt)))
|
blanchet@49264
|
568 |
|
blanchet@49422
|
569 |
fun num_keys keys = Graph.Keys.fold (K (Integer.add 1)) keys 0
|
blanchet@49415
|
570 |
|
blanchet@49422
|
571 |
fun maximal_in_graph fact_G facts =
|
blanchet@49331
|
572 |
let
|
blanchet@49393
|
573 |
val facts = [] |> fold (cons o nickname_of o snd) facts
|
blanchet@49422
|
574 |
val tab = Symtab.empty |> fold (fn name => Symtab.default (name, ())) facts
|
blanchet@49422
|
575 |
fun insert_new seen name =
|
blanchet@49422
|
576 |
not (Symtab.defined seen name) ? insert (op =) name
|
blanchet@49422
|
577 |
fun find_maxes _ (maxs, []) = map snd maxs
|
blanchet@49422
|
578 |
| find_maxes seen (maxs, new :: news) =
|
blanchet@49422
|
579 |
find_maxes
|
blanchet@49422
|
580 |
(seen |> num_keys (Graph.imm_succs fact_G new) > 1
|
blanchet@49422
|
581 |
? Symtab.default (new, ()))
|
blanchet@49422
|
582 |
(if Symtab.defined tab new then
|
blanchet@49422
|
583 |
let
|
blanchet@49422
|
584 |
val newp = Graph.all_preds fact_G [new]
|
blanchet@49422
|
585 |
fun is_ancestor x yp = member (op =) yp x
|
blanchet@49422
|
586 |
val maxs =
|
blanchet@49422
|
587 |
maxs |> filter (fn (_, max) => not (is_ancestor max newp))
|
blanchet@49422
|
588 |
in
|
blanchet@49422
|
589 |
if exists (is_ancestor new o fst) maxs then
|
blanchet@49422
|
590 |
(maxs, news)
|
blanchet@49422
|
591 |
else
|
blanchet@49422
|
592 |
((newp, new)
|
blanchet@49422
|
593 |
:: filter_out (fn (_, max) => is_ancestor max newp) maxs,
|
blanchet@49422
|
594 |
news)
|
blanchet@49422
|
595 |
end
|
blanchet@49422
|
596 |
else
|
blanchet@49422
|
597 |
(maxs, Graph.Keys.fold (insert_new seen)
|
blanchet@49422
|
598 |
(Graph.imm_preds fact_G new) news))
|
blanchet@49422
|
599 |
in find_maxes Symtab.empty ([], Graph.maximals fact_G) end
|
blanchet@49331
|
600 |
|
blanchet@49333
|
601 |
(* Generate more suggestions than requested, because some might be thrown out
|
blanchet@49333
|
602 |
later for various reasons and "meshing" gives better results with some
|
blanchet@49333
|
603 |
slack. *)
|
blanchet@49449
|
604 |
fun max_suggs_of max_facts = max_facts + Int.min (50, max_facts)
|
blanchet@49333
|
605 |
|
blanchet@49415
|
606 |
fun is_fact_in_graph fact_G (_, th) =
|
blanchet@49415
|
607 |
can (Graph.get_node fact_G) (nickname_of th)
|
blanchet@49335
|
608 |
|
blanchet@49421
|
609 |
fun mash_suggested_facts ctxt ({overlord, ...} : params) prover max_facts hyp_ts
|
blanchet@49421
|
610 |
concl_t facts =
|
blanchet@49316
|
611 |
let
|
blanchet@49317
|
612 |
val thy = Proof_Context.theory_of ctxt
|
blanchet@49449
|
613 |
val (fact_G, suggs) =
|
blanchet@49449
|
614 |
mash_peek ctxt (fn {fact_G} =>
|
blanchet@49449
|
615 |
if Graph.is_empty fact_G then
|
blanchet@49449
|
616 |
(fact_G, [])
|
blanchet@49449
|
617 |
else
|
blanchet@49449
|
618 |
let
|
blanchet@49449
|
619 |
val parents = maximal_in_graph fact_G facts
|
blanchet@49449
|
620 |
val feats =
|
blanchet@49449
|
621 |
features_of ctxt prover thy (Local, General) (concl_t :: hyp_ts)
|
blanchet@49449
|
622 |
in
|
blanchet@49449
|
623 |
(fact_G, mash_QUERY ctxt overlord (max_suggs_of max_facts)
|
blanchet@49449
|
624 |
(parents, feats))
|
blanchet@49449
|
625 |
end)
|
blanchet@49423
|
626 |
val selected =
|
blanchet@49423
|
627 |
facts |> suggested_facts suggs
|
blanchet@49423
|
628 |
(* The weights currently returned by "mash.py" are too extreme to
|
blanchet@49423
|
629 |
make any sense. *)
|
blanchet@49423
|
630 |
|> map fst |> weight_mepo_facts
|
blanchet@49415
|
631 |
val unknown = facts |> filter_out (is_fact_in_graph fact_G)
|
blanchet@49335
|
632 |
in (selected, unknown) end
|
blanchet@49264
|
633 |
|
blanchet@49419
|
634 |
fun add_to_fact_graph ctxt (name, parents, feats, deps) (adds, graph) =
|
blanchet@49331
|
635 |
let
|
blanchet@49331
|
636 |
fun maybe_add_from from (accum as (parents, graph)) =
|
blanchet@49336
|
637 |
try_graph ctxt "updating graph" accum (fn () =>
|
blanchet@49336
|
638 |
(from :: parents, Graph.add_edge_acyclic (from, name) graph))
|
blanchet@49336
|
639 |
val graph = graph |> Graph.default_node (name, ())
|
blanchet@49331
|
640 |
val (parents, graph) = ([], graph) |> fold maybe_add_from parents
|
blanchet@49422
|
641 |
val (deps, _) = ([], graph) |> fold maybe_add_from deps
|
blanchet@49419
|
642 |
in ((name, parents, feats, deps) :: adds, graph) end
|
blanchet@49321
|
643 |
|
blanchet@49399
|
644 |
val learn_timeout_slack = 2.0
|
blanchet@49333
|
645 |
|
blanchet@49399
|
646 |
fun launch_thread timeout task =
|
blanchet@49398
|
647 |
let
|
blanchet@49399
|
648 |
val hard_timeout = time_mult learn_timeout_slack timeout
|
blanchet@49399
|
649 |
val birth_time = Time.now ()
|
blanchet@49399
|
650 |
val death_time = Time.+ (birth_time, hard_timeout)
|
blanchet@49399
|
651 |
val desc = ("machine learner for Sledgehammer", "")
|
blanchet@49399
|
652 |
in Async_Manager.launch MaShN birth_time death_time desc task end
|
blanchet@49399
|
653 |
|
blanchet@49415
|
654 |
fun freshish_name () =
|
blanchet@49415
|
655 |
Date.fmt ".%Y_%m_%d_%H_%M_%S__" (Date.fromTimeLocal (Time.now ())) ^
|
blanchet@49415
|
656 |
serial_string ()
|
blanchet@49415
|
657 |
|
blanchet@49399
|
658 |
fun mash_learn_proof ctxt ({overlord, timeout, ...} : params) prover t facts
|
blanchet@49399
|
659 |
used_ths =
|
blanchet@49399
|
660 |
if is_smt_prover ctxt prover then
|
blanchet@49399
|
661 |
()
|
blanchet@49399
|
662 |
else
|
blanchet@49418
|
663 |
launch_thread timeout (fn () =>
|
blanchet@49418
|
664 |
let
|
blanchet@49418
|
665 |
val thy = Proof_Context.theory_of ctxt
|
blanchet@49418
|
666 |
val name = freshish_name ()
|
blanchet@49418
|
667 |
val feats = features_of ctxt prover thy (Local, General) [t]
|
blanchet@49418
|
668 |
val deps = used_ths |> map nickname_of
|
blanchet@49418
|
669 |
in
|
blanchet@49449
|
670 |
mash_peek ctxt (fn {fact_G} =>
|
blanchet@49449
|
671 |
let val parents = maximal_in_graph fact_G facts in
|
blanchet@49449
|
672 |
mash_ADD ctxt overlord [(name, parents, feats, deps)]
|
blanchet@49449
|
673 |
end);
|
blanchet@49449
|
674 |
(true, "")
|
blanchet@49418
|
675 |
end)
|
blanchet@49398
|
676 |
|
blanchet@49407
|
677 |
fun sendback sub =
|
blanchet@49407
|
678 |
Markup.markup Isabelle_Markup.sendback (sledgehammerN ^ " " ^ sub)
|
blanchet@49407
|
679 |
|
blanchet@49407
|
680 |
val commit_timeout = seconds 30.0
|
blanchet@49347
|
681 |
|
blanchet@49333
|
682 |
(* The timeout is understood in a very slack fashion. *)
|
blanchet@49419
|
683 |
fun mash_learn_facts ctxt (params as {debug, verbose, overlord, ...}) prover
|
blanchet@49419
|
684 |
auto_level atp learn_timeout facts =
|
blanchet@49319
|
685 |
let
|
blanchet@49333
|
686 |
val timer = Timer.startRealTimer ()
|
blanchet@49407
|
687 |
fun next_commit_time () =
|
blanchet@49407
|
688 |
Time.+ (Timer.checkRealTimer timer, commit_timeout)
|
blanchet@49415
|
689 |
val {fact_G} = mash_get ctxt
|
blanchet@49415
|
690 |
val (old_facts, new_facts) =
|
blanchet@49415
|
691 |
facts |> List.partition (is_fact_in_graph fact_G)
|
blanchet@49415
|
692 |
||> sort (thm_ord o pairself snd)
|
blanchet@49323
|
693 |
in
|
blanchet@49419
|
694 |
if null new_facts andalso (not atp orelse null old_facts) then
|
blanchet@49419
|
695 |
if auto_level < 2 then
|
blanchet@49419
|
696 |
"No new " ^ (if atp then "ATP" else "Isar") ^ " proofs to learn." ^
|
blanchet@49419
|
697 |
(if auto_level = 0 andalso not atp then
|
blanchet@49419
|
698 |
"\n\nHint: Try " ^ sendback learn_atpN ^ " to learn from ATP proofs."
|
blanchet@49419
|
699 |
else
|
blanchet@49419
|
700 |
"")
|
blanchet@49407
|
701 |
else
|
blanchet@49407
|
702 |
""
|
blanchet@49323
|
703 |
else
|
blanchet@49319
|
704 |
let
|
blanchet@49330
|
705 |
val all_names =
|
blanchet@49415
|
706 |
facts |> map snd
|
blanchet@49415
|
707 |
|> filter_out is_likely_tautology_or_too_meta
|
blanchet@49415
|
708 |
|> map (rpair () o nickname_of)
|
blanchet@49415
|
709 |
|> Symtab.make
|
blanchet@49419
|
710 |
val deps_of =
|
blanchet@49419
|
711 |
if atp then
|
blanchet@49419
|
712 |
atp_dependencies_of ctxt params prover auto_level facts all_names
|
blanchet@49419
|
713 |
else
|
blanchet@49419
|
714 |
isar_dependencies_of all_names
|
blanchet@49419
|
715 |
fun do_commit [] [] state = state
|
blanchet@49419
|
716 |
| do_commit adds reps {fact_G} =
|
blanchet@49407
|
717 |
let
|
blanchet@49419
|
718 |
val (adds, fact_G) =
|
blanchet@49419
|
719 |
([], fact_G) |> fold (add_to_fact_graph ctxt) adds
|
blanchet@49419
|
720 |
in
|
blanchet@49419
|
721 |
mash_ADD ctxt overlord (rev adds);
|
blanchet@49419
|
722 |
mash_REPROVE ctxt overlord reps;
|
blanchet@49419
|
723 |
{fact_G = fact_G}
|
blanchet@49419
|
724 |
end
|
blanchet@49419
|
725 |
fun commit last adds reps =
|
blanchet@49419
|
726 |
(if debug andalso auto_level = 0 then
|
blanchet@49419
|
727 |
Output.urgent_message "Committing..."
|
blanchet@49419
|
728 |
else
|
blanchet@49419
|
729 |
();
|
blanchet@49419
|
730 |
mash_map ctxt (do_commit (rev adds) reps);
|
blanchet@49419
|
731 |
if not last andalso auto_level = 0 then
|
blanchet@49419
|
732 |
let val num_proofs = length adds + length reps in
|
blanchet@49419
|
733 |
"Learned " ^ string_of_int num_proofs ^ " " ^
|
blanchet@49419
|
734 |
(if atp then "ATP" else "Isar") ^ " proof" ^
|
blanchet@49419
|
735 |
plural_s num_proofs ^ " in the last " ^
|
blanchet@49407
|
736 |
string_from_time commit_timeout ^ "."
|
blanchet@49407
|
737 |
|> Output.urgent_message
|
blanchet@49407
|
738 |
end
|
blanchet@49407
|
739 |
else
|
blanchet@49407
|
740 |
())
|
blanchet@49419
|
741 |
fun learn_new_fact _ (accum as (_, (_, _, _, true))) = accum
|
blanchet@49419
|
742 |
| learn_new_fact ((_, stature), th)
|
blanchet@49419
|
743 |
(adds, (parents, n, next_commit, _)) =
|
blanchet@49333
|
744 |
let
|
blanchet@49393
|
745 |
val name = nickname_of th
|
blanchet@49347
|
746 |
val feats =
|
blanchet@49400
|
747 |
features_of ctxt prover (theory_of_thm th) stature [prop_of th]
|
blanchet@49419
|
748 |
val deps = deps_of th |> these
|
blanchet@49409
|
749 |
val n = n |> not (null deps) ? Integer.add 1
|
blanchet@49419
|
750 |
val adds = (name, parents, feats, deps) :: adds
|
blanchet@49419
|
751 |
val (adds, next_commit) =
|
blanchet@49407
|
752 |
if Time.> (Timer.checkRealTimer timer, next_commit) then
|
blanchet@49419
|
753 |
(commit false adds []; ([], next_commit_time ()))
|
blanchet@49407
|
754 |
else
|
blanchet@49419
|
755 |
(adds, next_commit)
|
blanchet@49419
|
756 |
val timed_out = Time.> (Timer.checkRealTimer timer, learn_timeout)
|
blanchet@49419
|
757 |
in (adds, ([name], n, next_commit, timed_out)) end
|
blanchet@49419
|
758 |
val n =
|
blanchet@49419
|
759 |
if null new_facts then
|
blanchet@49419
|
760 |
0
|
blanchet@49419
|
761 |
else
|
blanchet@49419
|
762 |
let
|
blanchet@49419
|
763 |
val last_th = new_facts |> List.last |> snd
|
blanchet@49419
|
764 |
(* crude approximation *)
|
blanchet@49419
|
765 |
val ancestors =
|
blanchet@49419
|
766 |
old_facts
|
blanchet@49419
|
767 |
|> filter (fn (_, th) => thm_ord (th, last_th) <> GREATER)
|
blanchet@49422
|
768 |
val parents = maximal_in_graph fact_G ancestors
|
blanchet@49419
|
769 |
val (adds, (_, n, _, _)) =
|
blanchet@49419
|
770 |
([], (parents, 0, next_commit_time (), false))
|
blanchet@49419
|
771 |
|> fold learn_new_fact new_facts
|
blanchet@49419
|
772 |
in commit true adds []; n end
|
blanchet@49419
|
773 |
fun relearn_old_fact _ (accum as (_, (_, _, true))) = accum
|
blanchet@49419
|
774 |
| relearn_old_fact (_, th) (reps, (n, next_commit, _)) =
|
blanchet@49419
|
775 |
let
|
blanchet@49419
|
776 |
val name = nickname_of th
|
blanchet@49419
|
777 |
val (n, reps) =
|
blanchet@49419
|
778 |
case deps_of th of
|
blanchet@49419
|
779 |
SOME deps => (n + 1, (name, deps) :: reps)
|
blanchet@49419
|
780 |
| NONE => (n, reps)
|
blanchet@49419
|
781 |
val (reps, next_commit) =
|
blanchet@49419
|
782 |
if Time.> (Timer.checkRealTimer timer, next_commit) then
|
blanchet@49419
|
783 |
(commit false [] reps; ([], next_commit_time ()))
|
blanchet@49419
|
784 |
else
|
blanchet@49419
|
785 |
(reps, next_commit)
|
blanchet@49419
|
786 |
val timed_out = Time.> (Timer.checkRealTimer timer, learn_timeout)
|
blanchet@49419
|
787 |
in (reps, (n, next_commit, timed_out)) end
|
blanchet@49419
|
788 |
val n =
|
blanchet@49448
|
789 |
if not atp orelse null old_facts then
|
blanchet@49419
|
790 |
n
|
blanchet@49419
|
791 |
else
|
blanchet@49419
|
792 |
let
|
blanchet@49421
|
793 |
fun priority_of (_, th) =
|
blanchet@49419
|
794 |
random_range 0 (1000 * max_dependencies)
|
blanchet@49419
|
795 |
- 500 * (th |> isar_dependencies_of all_names
|
blanchet@49419
|
796 |
|> Option.map length
|
blanchet@49419
|
797 |
|> the_default max_dependencies)
|
blanchet@49419
|
798 |
val old_facts =
|
blanchet@49421
|
799 |
old_facts |> map (`priority_of)
|
blanchet@49419
|
800 |
|> sort (int_ord o pairself fst)
|
blanchet@49419
|
801 |
|> map snd
|
blanchet@49419
|
802 |
val (reps, (n, _, _)) =
|
blanchet@49419
|
803 |
([], (n, next_commit_time (), false))
|
blanchet@49419
|
804 |
|> fold relearn_old_fact old_facts
|
blanchet@49419
|
805 |
in commit true [] reps; n end
|
blanchet@49333
|
806 |
in
|
blanchet@49419
|
807 |
if verbose orelse auto_level < 2 then
|
blanchet@49419
|
808 |
"Learned " ^ string_of_int n ^ " nontrivial " ^
|
blanchet@49419
|
809 |
(if atp then "ATP" else "Isar") ^ " proof" ^ plural_s n ^
|
blanchet@49334
|
810 |
(if verbose then
|
blanchet@49334
|
811 |
" in " ^ string_from_time (Timer.checkRealTimer timer)
|
blanchet@49334
|
812 |
else
|
blanchet@49334
|
813 |
"") ^ "."
|
blanchet@49334
|
814 |
else
|
blanchet@49334
|
815 |
""
|
blanchet@49333
|
816 |
end
|
blanchet@49323
|
817 |
end
|
blanchet@49319
|
818 |
|
blanchet@49419
|
819 |
fun mash_learn ctxt (params as {provers, timeout, ...}) fact_override chained
|
blanchet@49419
|
820 |
atp =
|
blanchet@49331
|
821 |
let
|
blanchet@49411
|
822 |
val css = Sledgehammer_Fact.clasimpset_rule_table_of ctxt
|
blanchet@49410
|
823 |
val ctxt = ctxt |> Config.put instantiate_inducts false
|
blanchet@49410
|
824 |
val facts =
|
blanchet@49411
|
825 |
nearly_all_facts ctxt false fact_override Symtab.empty css chained []
|
blanchet@49411
|
826 |
@{prop True}
|
blanchet@49419
|
827 |
val num_facts = length facts
|
blanchet@49419
|
828 |
val prover = hd provers
|
blanchet@49419
|
829 |
fun learn auto_level atp =
|
blanchet@49419
|
830 |
mash_learn_facts ctxt params prover auto_level atp infinite_timeout facts
|
blanchet@49419
|
831 |
|> Output.urgent_message
|
blanchet@49331
|
832 |
in
|
blanchet@49419
|
833 |
(if atp then
|
blanchet@49419
|
834 |
("MaShing through " ^ string_of_int num_facts ^ " fact" ^
|
blanchet@49419
|
835 |
plural_s num_facts ^ " for ATP proofs (" ^ quote prover ^ " timeout: " ^
|
blanchet@49419
|
836 |
string_from_time timeout ^ ").\n\nCollecting Isar proofs first..."
|
blanchet@49419
|
837 |
|> Output.urgent_message;
|
blanchet@49419
|
838 |
learn 1 false;
|
blanchet@49419
|
839 |
"Now collecting ATP proofs. This may take several hours. You can \
|
blanchet@49419
|
840 |
\safely stop the learning process at any point."
|
blanchet@49419
|
841 |
|> Output.urgent_message;
|
blanchet@49419
|
842 |
learn 0 true)
|
blanchet@49419
|
843 |
else
|
blanchet@49419
|
844 |
("MaShing through " ^ string_of_int num_facts ^ " fact" ^
|
blanchet@49419
|
845 |
plural_s num_facts ^ " for Isar proofs..."
|
blanchet@49419
|
846 |
|> Output.urgent_message;
|
blanchet@49419
|
847 |
learn 0 false))
|
blanchet@49331
|
848 |
end
|
blanchet@49264
|
849 |
|
blanchet@49333
|
850 |
(* The threshold should be large enough so that MaSh doesn't kick in for Auto
|
blanchet@49333
|
851 |
Sledgehammer and Try. *)
|
blanchet@49333
|
852 |
val min_secs_for_learning = 15
|
blanchet@49333
|
853 |
|
blanchet@49336
|
854 |
fun relevant_facts ctxt (params as {learn, fact_filter, timeout, ...}) prover
|
blanchet@49336
|
855 |
max_facts ({add, only, ...} : fact_override) hyp_ts concl_t facts =
|
blanchet@49329
|
856 |
if not (subset (op =) (the_list fact_filter, fact_filters)) then
|
blanchet@49329
|
857 |
error ("Unknown fact filter: " ^ quote (the fact_filter) ^ ".")
|
blanchet@49329
|
858 |
else if only then
|
blanchet@49304
|
859 |
facts
|
blanchet@49336
|
860 |
else if max_facts <= 0 orelse null facts then
|
blanchet@49303
|
861 |
[]
|
blanchet@49303
|
862 |
else
|
blanchet@49303
|
863 |
let
|
blanchet@49342
|
864 |
fun maybe_learn () =
|
blanchet@49399
|
865 |
if learn andalso not (Async_Manager.has_running_threads MaShN) andalso
|
blanchet@49399
|
866 |
Time.toSeconds timeout >= min_secs_for_learning then
|
blanchet@49399
|
867 |
let val timeout = time_mult learn_timeout_slack timeout in
|
blanchet@49399
|
868 |
launch_thread timeout
|
blanchet@49419
|
869 |
(fn () => (true, mash_learn_facts ctxt params prover 2 false
|
blanchet@49407
|
870 |
timeout facts))
|
blanchet@49334
|
871 |
end
|
blanchet@49333
|
872 |
else
|
blanchet@49333
|
873 |
()
|
blanchet@49329
|
874 |
val fact_filter =
|
blanchet@49329
|
875 |
case fact_filter of
|
blanchet@49394
|
876 |
SOME ff => (() |> ff <> mepoN ? maybe_learn; ff)
|
blanchet@49333
|
877 |
| NONE =>
|
blanchet@49422
|
878 |
if is_smt_prover ctxt prover then
|
blanchet@49422
|
879 |
mepoN
|
blanchet@49422
|
880 |
else if mash_could_suggest_facts () then
|
blanchet@49422
|
881 |
(maybe_learn ();
|
blanchet@49422
|
882 |
if mash_can_suggest_facts ctxt then meshN else mepoN)
|
blanchet@49422
|
883 |
else
|
blanchet@49422
|
884 |
mepoN
|
blanchet@49303
|
885 |
val add_ths = Attrib.eval_thms ctxt add
|
blanchet@49307
|
886 |
fun prepend_facts ths accepts =
|
blanchet@49303
|
887 |
((facts |> filter (member Thm.eq_thm_prop ths o snd)) @
|
blanchet@49307
|
888 |
(accepts |> filter_out (member Thm.eq_thm_prop ths o snd)))
|
blanchet@49308
|
889 |
|> take max_facts
|
blanchet@49421
|
890 |
fun mepo () =
|
blanchet@49421
|
891 |
facts |> mepo_suggested_facts ctxt params prover max_facts NONE hyp_ts
|
blanchet@49421
|
892 |
concl_t
|
blanchet@49421
|
893 |
|> weight_mepo_facts
|
blanchet@49329
|
894 |
fun mash () =
|
blanchet@49421
|
895 |
mash_suggested_facts ctxt params prover max_facts hyp_ts concl_t facts
|
blanchet@49329
|
896 |
val mess =
|
blanchet@49421
|
897 |
[] |> (if fact_filter <> mashN then cons (mepo (), []) else I)
|
blanchet@49394
|
898 |
|> (if fact_filter <> mepoN then cons (mash ()) else I)
|
blanchet@49303
|
899 |
in
|
blanchet@49328
|
900 |
mesh_facts max_facts mess
|
blanchet@49303
|
901 |
|> not (null add_ths) ? prepend_facts add_ths
|
blanchet@49303
|
902 |
end
|
blanchet@49303
|
903 |
|
blanchet@49334
|
904 |
fun kill_learners () = Async_Manager.kill_threads MaShN "learner"
|
blanchet@49334
|
905 |
fun running_learners () = Async_Manager.running_threads MaShN "learner"
|
blanchet@49334
|
906 |
|
blanchet@49263
|
907 |
end;
|