1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194
|
; Copyright (C) 2017, Regents of the University of Texas
; Marijn Heule, Warren A. Hunt, Jr., and Matt Kaufmann
; License: A 3-clause BSD license. See the LICENSE file distributed with ACL2.
; See ../README.
(in-package "LRAT")
; Included books:
(include-book "std/util/bstar" :dir :system) ; b* is used below
(include-book "clrat-parser")
(include-book "tools/er-soft-logic" :dir :system)
; Locally included books:
(local (include-book "incremental"))
(local (include-book "../sorted/lrat-checker"))
; Avoid doing any proofs below by ensuring that every event is introduced in
; one of the books above:
(set-enforce-redundancy t)
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Include code from ../list-based/lrat-checker.lisp
; (some of which was defined in yet earlier checkers):
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun literalp (x)
(declare (xargs :guard t))
(and (integerp x)
(not (equal x 0))))
(defun literal-listp (x)
(declare (xargs :guard t))
(if (atom x)
(null x)
(and (literalp (car x))
(literal-listp (cdr x)))))
(defmacro negate (x)
`(- ,x))
(defun unique-literalsp (x)
(declare (xargs :guard (literal-listp x)))
(if (atom x)
t
(and (not (member (car x) (cdr x)))
(unique-literalsp (cdr x)))))
(defun conflicting-literalsp (x)
(declare (xargs :guard (literal-listp x)))
(if (atom x)
nil
(or (member (negate (car x)) (cdr x))
(conflicting-literalsp (cdr x)))))
(defun clause-or-assignment-p (clause)
(declare (xargs :guard t))
(and (literal-listp clause)
(unique-literalsp clause)
(not (conflicting-literalsp clause))))
(defconst *deleted-clause* :deleted)
(defmacro deleted-clause-p (val)
`(eq ,val *deleted-clause*))
(defun formula-p (fal)
; We recognize nil-terminated fast-alists (applicative hash tables), such that
; that every index is bound to a clause or *deleted-clause*.
(declare (xargs :guard t))
(if (atom fal)
(null fal)
(let ((pair (car fal)))
(and (consp pair)
(posp (car pair))
(let ((val (cdr pair)))
(or (deleted-clause-p val)
(clause-or-assignment-p val)))
(formula-p (cdr fal))))))
(defmacro index-listp (x)
`(pos-listp ,x))
(defun drat-hint-p (x)
(declare (xargs :guard t))
(and (consp x)
(posp (car x)) ; index
(index-listp (cdr x))))
(defun drat-hint-listp (x)
(declare (xargs :guard t))
(cond ((atom x) (null x))
(t (and (drat-hint-p (car x))
(drat-hint-listp (cdr x))))))
(defrec add-step
((index . clause)
.
(rup-indices . drat-hints))
t)
(defun add-step-p (x)
(declare (xargs :guard t))
(and (weak-add-step-p x)
(posp (access add-step x :index))
(clause-or-assignment-p (access add-step x :clause))
(index-listp (access add-step x :rup-indices))
(drat-hint-listp (access add-step x :drat-hints))))
(defun proof-entry-p (entry)
; This function recognizes a "line" in the proof, which can have either of the
; following two formats.
; Deletion: (T i1 i2 ...), indicating deletion of the specified (by index)
; clauses.
; Addition: an ADD-STEP record indication addition of a clause with a given
; index and suitable unit propagation hints.
(declare (xargs :guard t))
(cond ((and (consp entry)
(eq (car entry) t)) ; deletion
(index-listp (cdr entry)))
(t (add-step-p entry))))
(defmacro proof-entry-deletion-p (entry)
; assumes (proof-entry-p entry)
`(eq (car ,entry) t))
(defmacro proof-entry-deletion-indices (entry)
; assumes (proof-entry-p entry) and (proof-entry-deletion-p entry)
`(cdr ,entry))
(defun proofp (proof) ; primitive
; A proof is a true-list of proof-entry-p structures.
(declare (xargs :guard t))
(if (atom proof)
(null proof)
(and (proof-entry-p (car proof))
(proofp (cdr proof)))))
(defun negate-clause-or-assignment-rec (clause acc)
(declare (xargs :guard (and (literal-listp clause)
(literal-listp acc))))
(if (endp clause)
acc
(negate-clause-or-assignment-rec (cdr clause)
(cons (negate (car clause))
acc))))
(defund negate-clause-or-assignment (clause)
; When we originally proved soundness for this checker, we wrote
; negate-clause-or-assignment using a straightforward recursion (not
; tail-recursion). However, when we tried to prove a correspondence theorem
; between this checker and one with stobj-based assignments, we ran into
; trouble because the order of literals in this assignment was reversed from
; what is obtained by the stack. (Of course, we could have reversed what we
; produced from the stack; but then rat-assignment, which is already
; tail-recursive for this checker, would have things backwards instead.)
(declare (xargs :guard (literal-listp clause)))
(negate-clause-or-assignment-rec clause nil))
(defun-inline undefp (x)
(declare (xargs :guard t))
(not (booleanp x)))
(defun evaluate-literal (literal assignment)
(declare (xargs :guard (and (literalp literal)
(clause-or-assignment-p assignment))))
(cond
((member literal assignment) t)
((member (negate literal) assignment) nil)
;; When undefined, return 0.
(t 0)))
(defun evaluate-clause (clause assignment)
(declare (xargs :guard (and (clause-or-assignment-p clause)
(clause-or-assignment-p assignment))))
(if (atom clause)
nil
(let* ((literal (car clause))
(literal-value (evaluate-literal literal assignment)))
(if (eq literal-value t)
t
(let* ((remaining-clause (cdr clause))
(remaining-clause-value (evaluate-clause remaining-clause
assignment)))
(cond
((eq remaining-clause-value t) t)
((undefp literal-value) 0)
(t remaining-clause-value)))))))
; Change the following to defun-inline if you want a bit more performance and
; don't mind the inability to profile this function. Note that it is illegal
; to profile hons-get.
(defun-notinline my-hons-get (key alist)
(declare (xargs :guard t))
(hons-get key alist))
(defun remove-deleted-clauses (fal acc)
(declare (xargs :guard (alistp fal)))
(cond ((endp fal) (make-fast-alist acc))
(t (remove-deleted-clauses (cdr fal)
(if (deleted-clause-p (cdar fal))
acc
(cons (car fal) acc))))))
(defund shrink-formula (fal)
(declare (xargs :guard (formula-p fal)))
(let ((fal2 (fast-alist-clean fal)))
(fast-alist-free-on-exit fal2 (remove-deleted-clauses fal2 nil))))
(defun maybe-shrink-formula (ncls ndel formula factor)
; This function returns ncls unchanged, simply so that verify-clause can return
; directly by calling this function.
(declare (xargs :guard (and (integerp ncls) ; really natp; see verify-clause
(natp ndel)
(formula-p formula)
(rationalp factor))))
(cond ((> ndel (* factor ncls))
(let ((new-formula (shrink-formula formula)))
#+skip ; This is a nice check but we don't want to pay the price.
(assert$
(or (eql ncls (fast-alist-len new-formula))
(cw "ERROR: ncls = ~x0, (fast-alist-len new-formula) = ~x1"
ncls (fast-alist-len new-formula)))
(mv ncls 0 new-formula))
(mv ncls 0 new-formula)))
(t (mv ncls ndel formula))))
(defun delete-clauses (index-list fal)
(declare (xargs :guard (index-listp index-list)))
(cond ((endp index-list) fal)
(t (delete-clauses
(cdr index-list)
(hons-acons (car index-list) *deleted-clause* fal)))))
(defun add-proof-clause (index clause formula)
(declare (xargs :guard (and (posp index)
(formula-p formula))))
(hons-acons index clause formula))
; The functions defined below are only relevant to the correctness statement.
(defun-sk formula-truep (formula assignment)
(forall index
(let ((pair (hons-get index formula)))
(implies (and pair
(not (deleted-clause-p (cdr pair))))
(equal (evaluate-clause (cdr pair) assignment)
t)))))
(defun solution-p (assignment formula)
(and (clause-or-assignment-p assignment)
(formula-truep formula assignment)))
(defun-sk satisfiable (formula)
(exists assignment (solution-p assignment formula)))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Include code from ../stobj-based/lrat-checker.lisp:
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defstobj a$
; Note that a$stk contains positive integers, i.e., variables; its entries are
; indices into a$arr.
(a$ptr :type (integer 0 *) :initially 0) ; stack pointer
(a$stk :type (array t (1)) :resizable t) ; stack
(a$arr :type (array t (1)) ; array of 0, t, nil
:initially 0
:resizable t)
:renaming ((a$arrp a$arrp-weak)
(a$p a$p-weak))
:non-memoizable t
:inline t)
(defmacro varp (x)
`(posp ,x))
(defun varp$ (var a$)
(declare (xargs :stobjs a$ :guard t))
(and (varp var)
(< (abs var) (a$arr-length a$))))
(defun literalp$ (lit a$)
(declare (xargs :stobjs a$ :guard t))
(and (literalp lit)
(< (abs lit) (a$arr-length a$))))
(defun literal-listp$ (x a$)
(declare (xargs :stobjs a$ :guard t))
(if (atom x)
(null x)
(and (literalp$ (car x) a$)
(literal-listp$ (cdr x) a$))))
(defun clausep$ (x a$)
(declare (xargs :stobjs a$ :guard t))
(and (literal-listp$ x a$)
(unique-literalsp x)
(not (conflicting-literalsp x))))
(defun find-var-on-stk (var i a$)
; Return t if variable var = (a$stki j a$) for some j < i, else nil.
(declare (xargs :stobjs a$
:guard (and (varp var)
(natp i)
(<= i (a$ptr a$))
(<= (a$ptr a$) (a$stk-length a$)))))
(cond ((zp i) nil)
(t (let* ((i (1- i))
(var2 (a$stki i a$)))
(cond
((eql var var2) t)
(t (find-var-on-stk var i a$)))))))
(defun good-stk-p (i a$)
; This predicate holds when (varp$ (a$stk j a$) a$) is true for all j < i and
; moreover, there are no duplicate variables (a$stk j1 a$) = (a$stk j2 a$) for
; distinct j1, j2 < i. Here i is initially the (a$ptr a$).
(declare (xargs :stobjs a$
:guard (and (natp i)
(<= i (a$ptr a$))
(<= (a$ptr a$) (a$stk-length a$)))))
(cond ((zp i)
t)
(t (let* ((i (1- i))
(var (a$stki i a$)))
(and (varp$ var a$)
(not (find-var-on-stk var i a$))
(good-stk-p i a$))))))
(defun arr-matches-stk (i a$)
; Check whether for all 0 < j < i, the (a$arri j a$) is boolean iff j is on the
; stk of a$.
(declare (xargs :stobjs a$
:guard (and (natp i)
(<= i (a$arr-length a$))
(<= (a$ptr a$) (a$stk-length a$)))))
(cond ((or (zp i) (= i 1))
t)
(t (let ((var (1- i)))
(and (eq (booleanp (a$arri var a$))
(find-var-on-stk var (a$ptr a$) a$))
(arr-matches-stk var a$))))))
(defun a$arrp-rec (i a$)
(declare (xargs :stobjs a$
:guard (and (natp i)
(<= i (a$arr-length a$)))))
; Check that every value below i in a$arr is a Boolean or 0. Actually we don't
; need to check the value at index 0, since that is unused; but for simplicity
; (canonicity, maybe) let's check that too.
(cond ((zp i) t)
(t (let* ((i (1- i))
(v (a$arri i a$)))
(and (or (booleanp v)
(eql v 0))
(a$arrp-rec i a$))))))
(defun a$arrp (a$)
(declare (xargs :stobjs a$))
(a$arrp-rec (a$arr-length a$) a$))
(defun a$p (a$)
; From the defstobj form for a$, specifically from the definition of a$ptrp, we
; know that (a$ptr a) is a natp.
; Note that a$ptr is one more than the maximum valid index in a$stk; it is the
; index where we would place a variable for the next literal to be put into the
; assignment. Consider the following example to understand the relations
; below. Let the variables be 1, 2, and 3. Thus (a$stk-length a$) = 3 so that
; there is room for one literal for each of these variables, and (a$arr-length
; a$) = 4 since the maximum array index is 3. In the case that the stack is
; full, the stack has one literal on it for each of these variables, in which
; case a$ptr is 3, which is an illegal place to place the next variable --
; which is fine, since the stack is full.
(declare (xargs :stobjs a$))
(and (a$p-weak a$)
(<= (a$ptr a$) (a$stk-length a$))
(equal (a$arr-length a$) (1+ (a$stk-length a$)))
(good-stk-p (a$ptr a$) a$)
(a$arrp a$)
(arr-matches-stk (a$arr-length a$) a$)))
(defun-inline negate-value (val)
(declare (xargs :guard t))
(if (booleanp val) (not val) val))
(defun-inline evaluate-literal$ (lit a$)
(declare (xargs :stobjs a$
:guard (literalp$ lit a$)))
(if (< 0 lit)
(a$arri lit a$)
(negate-value (a$arri (negate lit) a$))))
(defun push-literal (lit a$)
; !! Possible future optimization:
; We may also want an optimized version of this function that assumes that lit
; is not already assigned. There could be at least two applications: when we
; create an assignment from a clause by pushing (successively unassigned)
; literals onto the empty stack, and probably also in unit-propagation since we
; only call push-literal when the literal is unassigned (would need to think
; through if that's really the case). Then the two tests below, (eq old t) and
; (eq old nil), would become part of the guard. The two versions could
; trivially be proved equal under the assumption of that strengthened guard.
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(literalp$ lit a$))))
(let* ((var (abs lit))
(old (a$arri var a$))
(lit-posp (eql var lit)))
(cond
((eq old t)
(mv (not lit-posp) a$))
((eq old nil)
(mv lit-posp a$))
(t (let* ((ptr (a$ptr a$))
(a$ (update-a$stki ptr var a$))
(a$ (update-a$ptr (1+ ptr) a$))
(a$ (update-a$arri var lit-posp a$)))
(mv nil a$))))))
(defun negate-clause (clause a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(clausep$ clause a$))))
(cond ((atom clause)
(mv nil a$))
(t (mv-let (flg a$)
(push-literal (negate (car clause)) a$)
(cond (flg (mv flg a$))
(t (negate-clause (cdr clause) a$)))))))
(defun evaluate-clause$ (clause a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(clausep$ clause a$))))
(if (atom clause)
nil
(let* ((literal (car clause))
(literal-value (evaluate-literal$ literal a$)))
(if (eq literal-value t)
t
(let* ((remaining-clause (cdr clause))
(remaining-clause-value (evaluate-clause$ remaining-clause
a$)))
(cond
((eq remaining-clause-value t) t)
((undefp literal-value) 0)
(t remaining-clause-value)))))))
(defun is-unit-clause$ (clause a$)
; If clause is a (pseudo) unit clause under assignment, return the unique
; unassigned literal (the others will be false). Otherwise return nil unless
; the clause is false under assignment, in which case return t.
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(clausep$ clause a$))))
(if (atom clause)
t ; top-level clause is false under assignment
(let ((val (evaluate-literal$ (car clause) a$)))
(cond
((eq val t) nil)
((undefp val)
(if (null (evaluate-clause$ (cdr clause) a$))
(car clause)
nil))
(t ; (null val)
(is-unit-clause$ (cdr clause) a$))))))
(defmacro unit-propagation$-error$ (msg formula indices a$)
`(prog2$ (er hard? 'unit-propagation "~@0" ,msg)
(unit-propagation$ ,formula (cdr ,indices) ,a$)))
(defun formula-p$ (formula a$)
; We recognize nil-terminated fast-alists (applicative hash tables), such that
; that every index is bound to a clause or *deleted-clause*.
(declare (xargs :stobjs a$ :guard (a$p a$)))
(if (atom formula)
(null formula)
(let ((pair (car formula)))
(and (consp pair)
(posp (car pair))
(let ((val (cdr pair)))
(or (deleted-clause-p val)
(clausep$ val a$)))
(formula-p$ (cdr formula) a$)))))
(defun unit-propagation$ (formula indices a$)
; Extend a$ by unit-propagation$ restricted to the given indices in formula.
; Return (mv flg a$) where flg indicates whether a contradiction is found.
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(formula-p$ formula a$)
(index-listp indices))))
(cond
((endp indices) (mv nil a$))
(t (let* ((pair (my-hons-get (car indices) formula))
(clause (and pair
(not (deleted-clause-p (cdr pair)))
(cdr pair)))
(unit-literal (and clause
(is-unit-clause$ clause a$))))
; Note that (member (- unit-literal) assignment) is false, because of how
; unit-literal is chosen. So we don't need to consider that case.
(cond ((not unit-literal)
; This is a surprising case. It is tempting simply to return
; assignment (hence failing to produce t). However, it seems that would cause
; monotonicity to fail (unit-propagation$-monotone), so reasoning would be more
; contorted: do all the reasoning about a version that recurs here (as we now
; do), and then fix the proof by connecting the two versions. Instead, we go
; ahead and recur, but cause an error if we encounter this situation.
(unit-propagation$-error$
(msg "unit-propagation$ has failed for index ~x0 because ~@1."
(car indices)
(cond ((null pair)
"no formula was found for that index")
((null clause)
"that clause had been deleted")
(t
"that clause is not a unit")))
formula indices a$))
((eq unit-literal t) ; found contradiction
(mv t a$))
(t (mv-let (flg a$)
(push-literal unit-literal a$)
(assert$
(null flg)
(unit-propagation$ formula (cdr indices) a$)))))))))
(defun rat-assignment$ (a$ nlit clause)
; This is approximately a tail-recursive, optimized version of:
; (union$ assignment
; (negate-clause-or-assignment
; (remove-literal nlit clause)))
; However, if a contradiction is discovered, then we return t.
(declare (xargs :stobjs a$
:guard
(and (a$p a$)
(literalp$ nlit a$)
(clausep$ clause a$))))
(cond ((endp clause) (mv nil a$))
((eql (car clause) nlit)
(rat-assignment$ a$ nlit (cdr clause)))
(t (mv-let (flg a$)
(push-literal (negate (car clause)) a$)
(cond (flg (mv flg a$))
(t (rat-assignment$ a$ nlit (cdr clause))))))))
(defun pop-literals (old-ptr a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(natp old-ptr)
(<= old-ptr (a$ptr a$)))
:measure (nfix (a$ptr a$))))
(cond
((and (mbt (and (a$p a$)
(natp old-ptr)
(<= old-ptr (a$ptr a$))))
(not (= old-ptr (a$ptr a$))))
(let* ((index (1- (a$ptr a$)))
(var (a$stki index a$))
(a$ (update-a$ptr index a$))
(a$ (update-a$arri var 0 a$)))
(pop-literals old-ptr a$)))
(t a$)))
(defun RATp1$ (alist formula nlit drat-hints a$)
; We think of assignment as being the result of having extended the global
; assignment with the negation of the current proof clause (to check that that
; clause is redundant with respect to formula).
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(formula-p$ alist a$)
(formula-p$ formula a$)
(literalp$ nlit a$)
(drat-hint-listp drat-hints))))
(if (endp alist)
(mv t a$)
(let* ((index (caar alist))
(clause (cdar alist)))
(cond
((deleted-clause-p clause)
(RATp1$ (cdr alist) formula nlit drat-hints a$))
((eql index (caar drat-hints)) ; perform RAT
(b* ((old-ptr (a$ptr a$))
((mv flg a$) (rat-assignment$ a$ nlit clause)))
(cond
(flg (let ((a$ (pop-literals old-ptr a$)))
(RATp1$ (cdr alist) formula nlit (cdr drat-hints) a$)))
(t (mv-let
(flg a$)
(unit-propagation$ formula (cdar drat-hints) a$)
(let ((a$ (pop-literals old-ptr a$)))
(cond
(flg (RATp1$ (cdr alist) formula nlit (cdr drat-hints) a$))
(t ; error
(mv (list 'unit-propagation-failure index clause nlit)
a$)))))))))
((or (not (member nlit clause))
(deleted-clause-p (cdr (my-hons-get index formula))))
(RATp1$ (cdr alist) formula nlit drat-hints a$))
(t ; error
(mv (list 'index-failure index clause nlit)
a$))))))
(defun RATp$ (formula literal drat-hints a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(formula-p$ formula a$)
(literalp$ literal a$)
(drat-hint-listp drat-hints))))
(RATp1$ formula formula (negate literal) drat-hints a$))
(defun verify-clause$ (formula add-step ncls ndel a$)
; In the normal case this function returns (mv ncls ndel formula a$) for new
; ncls, ndel, formula, and a$. Otherwise it returns (mv nil _ formula a$)
; where formula is unchanged, but in that case a hard error occurs. Note that
; a$ptr is the same for the input and output a$.
(declare (xargs :stobjs a$
:guard
(and (a$p a$)
; The following is necessary for proof that follows below, but perhaps it can
; be eliminated if we also eliminate the corresponding conjunct from other
; guards below. It will then likely be necessary to prove that a$ptr can only
; increase with various operations, in particular, unit-propagation$ and
; negate-clause. That shouldn't be difficult (in fact, such lemmas might be
; somewhere in this book already). However, it seems harmless to leave this
; conjunct in guards.
(= (a$ptr a$) 0)
(formula-p$ formula a$)
(add-step-p add-step)
(clausep$ (access add-step add-step :clause) a$)
(integerp ncls) ; really natp; see verify-proof-rec
(natp ndel))))
(b* ((proof-clause (access add-step add-step :clause))
(old-ptr (a$ptr a$))
((mv flg0 a$) (negate-clause proof-clause a$))
((when flg0) ; Shouldn't happen
(prog2$ (er hard? 'verify-clause$
"Implementation error?! Note that a$ptr is ~x0."
(a$ptr a$))
(let ((a$ (pop-literals old-ptr a$)))
(mv nil nil nil a$))))
(rup-indices (access add-step add-step :rup-indices))
((mv flg a$) (unit-propagation$ formula rup-indices a$)))
(cond
((eq flg t)
(b* ((a$ (pop-literals old-ptr a$))
((mv ncls nlit new-formula)
(maybe-shrink-formula ncls ndel formula
; shrink when ndel > 10 * ncls; factor can be changed
10)))
(mv ncls nlit new-formula a$)))
((consp proof-clause)
(b* (((mv ncls ndel formula)
(maybe-shrink-formula ncls ndel formula
; shrink when ndel > 1/3 * ncls; factor can be changed
1/3))
((mv flg a$)
(RATp$ formula (car proof-clause)
(access add-step add-step :drat-hints)
a$))
(a$ (pop-literals old-ptr a$)))
(cond
((eq flg t)
(mv ncls ndel formula a$))
(t (prog2$
(b* ((current-index (access add-step add-step :index))
((list er-type earlier-index & nlit) flg))
(case er-type
(unit-propagation-failure
(er hard? 'verify-clause$
"Unit propagation failure has caused the RAT check to ~
fail when attempting to add proof clause #~x0 for ~
earlier RAT clause #~x1."
current-index earlier-index))
(index-failure
(er hard? 'verify-clause$
"The RAT check has failed for proof clause #~x0, ~
because literal ~x1 belongs to earlier proof clause ~
#~x2 but no hint for that clause is given with proof ~
clause #~x0."
current-index nlit earlier-index))
(otherwise ; surprising; RATp1$ and this function are out of sync
(er hard? 'verify-clause$
"Unexpected error for RAT check, proof clause #~x0; the ~
error is probably a true error but the checker needs ~
to be fixed to print a more useful error in this case."
current-index))))
(mv nil nil nil a$))))))
(t (prog2$
(er hard? 'verify-clause$
"The unit-propagation$ check failed at proof clause #~x0, which ~
is the empty clause."
(access add-step add-step :index))
(b* ((a$ (pop-literals old-ptr a$)))
(mv nil nil nil a$)))))))
(defun proofp$ (proof a$)
; We could make this predicate more efficient by folding the clausep$ check
; into a suitable strengthening proof-entry-p$ of proof-entry-p. But this
; predicate is only for proof purposes, not to be executed.
(declare (xargs :stobjs a$
:guard (a$p a$)))
(if (atom proof)
(null proof)
(and (proof-entry-p (car proof))
(or (proof-entry-deletion-p (car proof))
(clausep$ (access add-step (car proof) :clause) a$))
(proofp$ (cdr proof) a$))))
(defund initialize-a$ (max-var a$)
(declare (xargs :stobjs a$
:guard (varp max-var)))
(let* ((a$ (update-a$ptr 0 a$))
(a$ (resize-a$stk 0 a$))
(a$ (resize-a$stk max-var a$))
(a$ (resize-a$arr 0 a$))
(a$ (resize-a$arr (1+ max-var) a$)))
a$))
(defun clause-max-var (clause acc)
(declare (xargs :guard (and (literal-listp clause)
(natp acc))))
(cond ((endp clause) acc)
(t (clause-max-var (cdr clause)
(max (abs (car clause))
acc)))))
(defun formula-max-var (fal acc)
; We only apply this function to formulas with no deleted clauses, so there is
; a slight opportunity for optimization. But that seems really minor.
(declare (xargs :guard (and (formula-p fal)
(natp acc))))
(cond ((atom fal) acc)
(t (formula-max-var (cdr fal)
(if (deleted-clause-p (cdar fal))
acc
(clause-max-var (cdar fal) acc))))))
(defun proof-max-var (proof acc)
(declare (xargs :guard (and (proofp proof)
(natp acc))))
(cond
((endp proof) acc)
(t (proof-max-var (cdr proof)
(let ((entry (car proof)))
(if (proof-entry-deletion-p entry)
acc
(clause-max-var (access add-step entry :clause)
acc)))))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Include code from lrat-checker-support.lisp:
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun ordered-literalsp (x)
(declare (xargs :guard (literal-listp x)))
(cond ((endp x) t)
(t (or (null (cdr x))
(and (< (abs (car x)) (abs (cadr x)))
(ordered-literalsp (cdr x)))))))
(defun lrat-clausep (x)
(declare (xargs :guard t))
(or (null x)
(and (literal-listp x)
(not (member (car x) (cdr x)))
(not (member (negate (car x)) (cdr x)))
(ordered-literalsp (cdr x)))))
(defun lrat-add-step-p (x)
(declare (xargs :guard t))
(and (weak-add-step-p x)
(posp (access add-step x :index))
(lrat-clausep (access add-step x :clause))
(index-listp (access add-step x :rup-indices))
(drat-hint-listp (access add-step x :drat-hints))))
(defun lrat-proof-entry-p (entry)
(declare (xargs :guard t))
(cond ((and (consp entry)
(eq (car entry) t)) ; deletion
(index-listp (cdr entry)))
(t (lrat-add-step-p entry))))
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Include code from incremental.lisp:
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
(defun incl-verify-proof$-rec (ncls ndel formula proof a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
; See the comment in verify-clause$ about perhaps eliminating the next
; conjunct (which is perhaps not necessary).
(= (a$ptr a$) 0)
(integerp ncls) ; really natp; see comment below
(natp ndel)
(formula-p$ formula a$)
(proofp$ proof a$))))
(cond
((atom proof) (mv t formula a$))
(t
(let* ((entry (car proof))
(delete-flg (proof-entry-deletion-p entry)))
(cond
(delete-flg
(let* ((indices (proof-entry-deletion-indices entry))
(new-formula (delete-clauses indices formula))
(len (length indices))
(ncls
; We expect that (<= len ncls). It is tempting to assert that here (with
; assert$), but it's not necessary so we avoid the overhead (mostly in proof,
; but perhaps also a bit in execution).
(- ncls len))
(ndel (+ ndel len)))
(incl-verify-proof$-rec ncls ndel new-formula (cdr proof) a$)))
(t ; addition
(mv-let (ncls ndel new-formula a$)
(verify-clause$ formula entry ncls ndel a$)
(cond (ncls ; success
(let* ((entry-clause (access add-step entry :clause))
(newest-formula
(add-proof-clause (access add-step entry :index)
entry-clause
new-formula)))
(cond
((null entry-clause)
(mv :complete newest-formula a$))
(t
(incl-verify-proof$-rec
(1+ ncls)
ndel
newest-formula
(cdr proof)
a$)))))
(t (mv nil formula a$))))))))))
(defun incl-verify-proof$ (formula proof a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
; See the comment in verify-clause$ about perhaps eliminating the next
; conjunct (which is perhaps not necessary).
(= (a$ptr a$) 0)
(formula-p$ formula a$)
(proofp$ proof a$))))
(incl-verify-proof$-rec (fast-alist-len formula)
0
formula
proof
a$))
(defun incl-initialize-a$ (max-var a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(equal (a$ptr a$) 0)
(natp max-var))))
(cond ((<= max-var (a$stk-length a$))
a$)
(t (let* ((new-max-var (* 2 max-var)))
(initialize-a$ new-max-var a$)))))
(defun check-proofp (proof) ; primitive
; This variant of proofp causes an error when false, printing the offending
; entry.
(declare (xargs :guard t))
(if (atom proof)
(null proof)
(and (or (lrat-proof-entry-p (car proof))
(er hard? 'check-proof
"Illegal proof entry: ~X01"
(car proof)
nil))
(check-proofp (cdr proof)))))
(defun incl-valid-proofp$ (formula proof old-max-var a$)
(declare (xargs :stobjs a$
:guard (and (a$p a$)
(eql (a$ptr a$) 0)
(formula-p formula)
(natp old-max-var)
(<= (formula-max-var formula 0)
old-max-var))))
(let* ((formula (shrink-formula formula))
(max-var (and (check-proofp proof)
(proof-max-var proof old-max-var))))
(cond ((natp max-var)
(let ((a$ (incl-initialize-a$ max-var a$)))
(mv-let (v new-formula a$)
(incl-verify-proof$ formula proof a$)
(mv v
new-formula
max-var
a$))))
(t (mv nil formula old-max-var a$)))))
(defun incl-valid-proofp$-top-rec (formula clrat-file posn chunk-size
clrat-file-length old-suffix debug
old-max-var a$ ctx state)
(declare (xargs :guard (and (a$p a$)
(eql (a$ptr a$) 0)
(formula-p formula)
(stringp clrat-file)
(natp posn)
(< posn *2^56*)
(posp chunk-size)
(posp clrat-file-length)
(stringp old-suffix)
(natp old-max-var)
(<= (formula-max-var formula 0)
old-max-var))
:ruler-extenders (:lambdas mv-list return-last) ; ugly bug work-around
:measure (nfix (- clrat-file-length posn))
:stobjs (state a$)))
(cond
((and (mbt (natp posn))
(mbt (posp clrat-file-length))
(mbt (posp chunk-size))
(<= posn clrat-file-length))
(prog2$
(and debug
(cw "; Note: Reading from position ~x0~|" posn))
(mv-let (proof suffix new-posn)
(clrat-read-next clrat-file posn chunk-size old-suffix
clrat-file-length state)
(cond
((null suffix) ; error (normally a string, possibly even "")
(mv (er hard? ctx "Implementation error: Null suffix!")
a$))
((null proof)
(mv :incomplete a$))
((stringp proof) ; impossible
(mv (er hard? ctx
"Implementation error: ~x0 returned a string for its proof, ~
which we thought was impossible!"
'clrat-read-next)
a$))
(t
(mv-let (v new-formula new-max-var a$)
(prog2$
(cw "; Note: Checking next proof segment.~|")
(incl-valid-proofp$ formula proof old-max-var a$))
(cond
((>= new-posn *2^56*)
(mv (er hard? ctx
"Attempted to read at position ~x0, but the maximum ~
legal such position is 2^56 = ~x1."
new-posn *2^56*)
a$))
((not v) (mv nil a$))
((eq v :complete)
(mv :complete a$))
((> new-posn clrat-file-length)
; If new-posn is exactly clrat-file-length, then as per the discussion of the
; "truncation case" in :doc read-file-into-string, we iterate. But if
; new-posn exceeds clrat-file-length, then we have a valid proof that does not
; include the empty clause.
(mv :incomplete a$))
(t
(incl-valid-proofp$-top-rec new-formula clrat-file new-posn
chunk-size clrat-file-length suffix
debug new-max-var a$ ctx
state)))))))))
(t ; impossible
(mv nil a$))))
(defun incl-valid-proofp$-top-aux (formula clrat-file incomplete-okp chunk-size
clrat-file-length debug a$ ctx state)
(declare (xargs :stobjs (a$ state)
:guard (and (a$p a$)
(eql (a$ptr a$) 0)
(formula-p formula)
(stringp clrat-file)
(posp chunk-size)
(posp clrat-file-length))))
(mv-let (val a$)
(incl-valid-proofp$-top-rec formula clrat-file 0 chunk-size
clrat-file-length "" debug
(formula-max-var formula 0)
a$ ctx state)
(case val
(:complete (mv t a$))
(:incomplete (mv (or incomplete-okp
(er hard? ctx
"The proof is valid but does not contain the ~
empty clause."))
a$))
(t
; We do not expect to reach the following case. If nil is returned as the
; first value, it is ultimately because an error occurred. In particular,
; verify-clause$ either succeeds or causes an error.
(mv (er hard? ctx
"Invalid proof!")
a$)))))
(defun ordered-formula-p1 (formula index)
(declare (xargs :guard (posp index)))
(if (atom formula)
(null formula)
(let ((pair (car formula)))
(and (consp pair)
(posp (car pair))
(clause-or-assignment-p (cdr pair))
(< (car pair) index)
(ordered-formula-p1 (cdr formula) (car pair))))))
(defund ordered-formula-p (formula)
; It is important that the formula produced by the cnf parser does not have
; duplicate indices, since otherwise the first call of shrink-formula will
; change its semantics. Fortunately, the cnf parser presents the formula in
; reverse order; so we can check for duplicate-free indices in linear time.
(declare (xargs :guard t))
(if (atom formula)
(null formula)
(let ((pair (car formula)))
(and (consp pair)
(posp (car pair))
(clause-or-assignment-p (cdr pair))
(ordered-formula-p1 (cdr formula) (car pair))))))
(defun incl-valid-proofp$-top (cnf-file clrat-file incomplete-okp chunk-size
debug ctx state)
(declare (xargs :guard t :stobjs state))
(let* ((formula (ec-call (cnf-read-file cnf-file state)))
(state (increment-file-clock state)))
(cond
((not (stringp clrat-file))
(er-soft-logic
ctx
"The filename argument is not a string:~|~x0"
clrat-file))
((not (posp chunk-size))
(er-soft-logic
ctx
"The supplied :chunk-size must be a positive integer, but ~x0 is ~
not.~|~x0"
clrat-file))
((not (ordered-formula-p formula))
(er-soft-logic ctx
"An invalid formula was supplied by the parser from ~
input file ~x0."
cnf-file))
(t
(mv-let (clrat-file-length state)
(file-length$ clrat-file state)
(cond
((posp clrat-file-length)
(prog2$
(and debug
(cw "Length of file ~x0: ~x1~|" clrat-file clrat-file-length))
(value
(with-fast-alist
formula
(with-local-stobj a$
(mv-let
(val a$)
(let ((a$ (resize-a$arr 2 a$))) ; to get a$p to hold
(incl-valid-proofp$-top-aux formula
clrat-file
incomplete-okp chunk-size
clrat-file-length debug a$
ctx state))
(cons val formula)))))))
((eql clrat-file-length 0)
(er-soft-logic
ctx
"Zero-length file: ~x0."
clrat-file))
(t (er-soft-logic
ctx
"Sorry, Lisp cannot determine the file-length of file ~x0."
clrat-file))))))))
(defun incl-verify-proof-fn (cnf-file clrat-file incomplete-okp chunk-size
debug state)
; This is just a little interface to incl-valid-proofp$-top.
(declare (xargs :guard t
:stobjs state))
(er-let* ((val/formula
(time$ (incl-valid-proofp$-top cnf-file clrat-file incomplete-okp
chunk-size debug 'incl-verify-proof
state))))
(value (car val/formula))))
(defconst *256mb*
(expt 2 28))
(defconst *default-chunk-size*
*256mb*)
(defmacro incl-verify-proof (cnf-file clrat-file
&key
incomplete-okp
chunk-size
(debug 't))
`(incl-verify-proof-fn ,cnf-file ,clrat-file ,incomplete-okp
,(or chunk-size *default-chunk-size*)
,debug
state))
; Soundness
(local (include-book "soundness"))
(defun proved-formula (cnf-file clrat-file chunk-size debug incomplete-okp ctx
state)
(declare (xargs :stobjs state))
(mv-let (erp val/formula state)
(incl-valid-proofp$-top cnf-file clrat-file
incomplete-okp
chunk-size debug ctx state)
(value (and (null erp)
(consp val/formula)
(eq (car val/formula) t)
; The formula returned by incl-valid-proofp$-top is in reverse order. We
; prefer to return a formula that we expect to agree with the formula
; represented in the input cnf-file.
(reverse (cdr val/formula))))))
(defthm soundness
(let ((formula (mv-nth 1
(proved-formula cnf-file clrat-file chunk-size debug
nil ; incomplete-okp
ctx state))))
(implies formula
(not (satisfiable formula)))))
|