/usr/share/acl2-8.0dfsg/books/misc/multi-v-uni.lisp is in acl2-books-source 8.0dfsg-1.
This file is owned by root:root, with mode 0o644.
The actual contents of the file can be viewed below.
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 | ; A Mechanically Checked Proof of a
; Multiprocessor Result via a Uniprocessor View
; J Strother Moore
; Febuary 27, 1998
; This script proves the theorem described in the paper ``A Mechanically
; Checked Proof of a Multiprocessor Result via a Uniprocessor View'', J Moore,
; Formal Methods in System Design, 14(2), March, 1999, pp. 213-228. See also
; http://www.cs.utexas.edu/users/moore/publications/multi-v-uni.pdf
(in-package "ACL2")
(defun make-mstate (alist mem code)
(list alist mem code))
(defun pida (s) (nth 0 s))
(defun mem (s) (nth 1 s))
(defun code (s) (nth 2 s))
(defthm pida-make-mstate
(equal (pida (make-mstate alist mem code)) alist))
(defthm mem-make-mstate
(equal (mem (make-mstate alist mem code)) mem))
(defthm code-make-mstate
(equal (code (make-mstate alist mem code)) code))
(in-theory (disable make-mstate pida mem code))
(defun make-ls (pcn regs)
(list pcn regs))
(defun pcn (ls) (nth 0 ls))
(defun regs (ls) (nth 1 ls))
(defthm pcn-make-ls
(equal (pcn (make-ls pcn regs)) pcn))
(defthm regs-make-ls
(equal (regs (make-ls pcn regs)) regs))
(in-theory (disable make-ls pcn regs))
(defmacro modify (s &key
(pcn 'nil pcnp)
(regs 'nil regsp)
(pida 'nil pidap)
(mem 'nil memp)
(code 'nil codep))
; This is a weird macro which implements an overloaded form of modify.
; (modify s :pcn x) treats s as though it were a local state, while
; (modify s :mem x) treats s as though it were a mstate.
(if (or pcnp regsp)
`(make-ls
,(if pcnp pcn `(pcn ,s))
,(if regsp regs `(regs ,s)))
`(make-mstate
,(if pidap
pida
`(pida ,s))
,(if memp mem `(mem ,s))
,(if codep code `(code ,s)))))
(defun bind (x y alist)
(cond ((endp alist) (list (cons x y)))
((equal x (car (car alist)))
(cons (cons x y) (cdr alist)))
(t (cons (car alist) (bind x y (cdr alist))))))
(defun bound? (x alist)
(cond ((endp alist) nil)
((equal x (caar alist)) (car alist))
(t (bound? x (cdr alist)))))
(defmacro binding (x alist)
`(cdr (bound? ,x ,alist)))
(defthm bound?-bind
(equal (bound? var1 (bind var2 val alist))
(if (equal var1 var2)
(cons var1 val)
(bound? var1 alist))))
(defun ls (pid s)
(binding pid (pida s)))
(defun current-instruction (pid s)
(nth (pcn (ls pid s)) (code s)))
; The code in question is:
#| ((LOAD OLD CTR)
(INCR NEW OLD)
(CAS CTR OLD NEW)
(BR NEW 1)
(JUMP 0))|#
(defun load-step (ins ls mem)
(let ((reg (cadr ins)) ; (LOAD reg loc): reg := mem(loc)
(loc (caddr ins)))
(mv (modify ls
:pcn (1+ (pcn ls))
:regs (bind reg (binding loc mem) (regs ls)))
mem)))
(defun incr-step (ins ls mem)
(let ((reg1 (cadr ins)) ; (INCR reg1 reg2): reg1 := reg2+1
(reg2 (caddr ins)))
(mv (modify ls
:pcn (1+ (pcn ls))
:regs (bind reg1
(1+ (binding reg2 (regs ls)))
(regs ls)))
mem)))
(defun cas-step (ins ls mem)
; (CAS loc reg1 reg2): If mem(loc)=reg1
; then mem(loc):=reg2 ; reg2:=nil
; else reg1:=mem(loc) ; reg2:=t
(let ((loc (cadr ins))
(reg1 (caddr ins))
(reg2 (cadddr ins)))
(cond
((equal (binding reg1 (regs ls))
(binding loc mem))
(mv (modify ls
:pcn (1+ (pcn ls))
:regs (bind reg2 nil (regs ls)))
(bind loc (binding reg2 (regs ls)) mem)))
(t (mv (modify ls
:pcn (1+ (pcn ls))
:regs (bind reg1 (binding loc mem)
(bind reg2 t (regs ls))))
mem)))))
(defun br-step (ins ls mem)
(let ((reg (cadr ins)) ; (BR reg adr): if reg, go to adr
(adr (caddr ins)))
(mv (modify ls
:pcn (if (binding reg (regs ls))
adr
(1+ (pcn ls))))
mem)))
(defun jump-step (ins ls mem)
(let ((adr (cadr ins)) ; (JUMP adr): goto to adr
)
(mv (modify ls
:pcn adr)
mem)))
(defun execute (pid ins s)
(let ((ls (ls pid s)))
(mv-let (new-ls new-mem)
(case (car ins)
(load (load-step ins ls (mem s)))
(incr (incr-step ins ls (mem s)))
(cas (cas-step ins ls (mem s)))
(br (br-step ins ls (mem s)))
(jump (jump-step ins ls (mem s)))
(otherwise (mv ls (mem s))))
(modify s
:pida
(bind pid new-ls (pida s))
:mem new-mem))))
(defun mstep (pid s)
(cond ((bound? pid (pida s))
(execute pid
(current-instruction pid s)
s))
(t s)))
(defun mrun (oracle s)
(if (endp oracle)
s
(mrun (cdr oracle)
(mstep (car oracle) s))))
; We now develop the basic rules for dealing with mrun.
(defthm multrun-opener
(equal (mrun (cons pid L) s)
(mrun L (mstep pid s)))
:hints (("goal" :in-theory (disable mstep))))
(defthm mstep-opener
(implies (consp (current-instruction pid s))
(equal (mstep pid s)
(if (bound? pid (pida s))
(execute pid
(current-instruction pid s)
s)
s))))
(in-theory (disable mstep))
(defthm mrun-append
(equal (mrun (append L1 L2) s)
(mrun L2 (mrun L1 s))))
(defthm code-mstep
(equal (code (mstep pid s))
(code s))
:hints (("Goal" :in-theory (enable mstep))))
(defthm code-mrun
(equal (code (mrun L s))
(code s)))
(defthm bound?-mstep
(iff (bound? pid (pida (mstep active-pid s)))
(bound? pid (pida s)))
:hints (("Goal" :in-theory (enable mstep))))
(defthm bound?-mrun
(iff (bound? pid (pida (mrun L s)))
(bound? pid (pida s))))
(defthm bound?-pida
(implies (not (member-equal pid L))
(equal (bound? pid (pida (mrun L s)))
(bound? pid (pida s))))
:hints (("Goal" :in-theory (enable mstep))))
; ----------------------------------------------------------------------------
; We now develop the single processor view of the system.
(defun make-ustate (uls umem ucode)
(list uls umem ucode))
(defun uls (us) (nth 0 us))
(defun umem (us) (nth 1 us))
(defun ucode (us) (nth 2 us))
(defthm uls-make-ustate
(equal (uls (make-ustate uls umem ucode)) uls))
(defthm umem-make-ustate
(equal (umem (make-ustate uls umem ucode)) umem))
(defthm ucode-make-ustate
(equal (ucode (make-ustate uls umem ucode)) ucode))
(in-theory (disable make-ustate uls umem ucode))
(defun current-uinstruction (us)
(nth (pcn (uls us)) (ucode us)))
(defun uexecute (ins us actual-mem)
(LET
((LS (uls us)))
(MV-LET (NEW-LS NEW-MEM)
(CASE (CAR INS)
(LOAD (LOAD-STEP INS LS actual-mem))
(INCR (INCR-STEP INS LS actual-mem))
(CAS (CAS-STEP INS LS actual-mem))
(BR (BR-STEP INS LS actual-mem))
(JUMP (JUMP-STEP INS LS actual-mem))
(OTHERWISE (MV LS actual-mem)))
(make-ustate new-ls
new-mem
(ucode us)))))
(defun ustep (us actual-mem)
(uexecute (current-uinstruction us) us actual-mem))
(defun urun (us M)
(cond ((endp M) us)
((endp (cdr M))
(make-ustate (uls us)
(car M)
(ucode us)))
(t (urun (ustep us (car M)) (cdr M)))))
; Originally, I defined (ascendingp M) to just check that the sequence
; of 'ctr bindings was weakly ascending. That was insufficient to
; make our main theorem a theorem! Imagine that ctr is fixed in
; M1-M6. When the CAS is executed, it will do the write,
; increasing the CTR, say on the step applied to M3. But the step
; applied to M4 will use M4, not the new memory! The ascending
; predicate must be stronger.
(defthm urun-opener
(equal (urun us (cons mem M))
(if (endp M)
(make-ustate (uls us) mem (ucode us))
(urun (ustep us mem) M)))
:hints (("goal" :in-theory (disable ustep))))
(defthm ustep-opener
(implies (consp (current-uinstruction us))
(equal (ustep us mem)
(uexecute
(current-uinstruction us)
us
mem))))
(in-theory (disable ustep))
(defun proj (pid s)
(make-ustate (ls pid s)
(mem s)
(code s)))
(defun initial-subsequence (L pid)
(cond ((endp L) nil)
((equal (car L) pid) nil)
(t (cons (car L) (initial-subsequence (cdr L) pid)))))
(defthm acl2-count-member-equal
(implies (member-equal pid L)
(<= (acl2-count (member-equal pid L))
(acl2-count L)))
:rule-classes :linear)
(defun partition-oracle (pid L)
(cond ((member-equal pid L)
(cons (initial-subsequence L pid)
(partition-oracle pid (cdr (member-equal pid L)))))
(t (list L))))
(defun new-oracle (pid s partitions)
(cond ((endp partitions) nil)
(t (cons (mem (mrun (car partitions) s))
(new-oracle pid
(mstep pid (mrun (car partitions) s))
(cdr partitions))))))
(defun proj-oracle (pid s L)
(new-oracle pid s (partition-oracle pid L)))
(defthm new-oracle-opener
(equal (new-oracle pid s (cons part L))
(cons (mem (mrun part s))
(new-oracle pid
(mstep pid (mrun part s))
L))))
(defun mrun-is-urun-hint (pid L s)
(cond ((member-equal pid L)
(mrun-is-urun-hint
pid
(cdr (member-equal pid L))
(mstep pid (mrun (initial-subsequence L pid) s))))
(t s)))
(defthm consp-new-oracle
(iff (consp (new-oracle pid s L))
(consp L)))
(defthm initial-subsequence-elim
(implies (member-equal pid L)
(equal (append (initial-subsequence L pid)
(cons pid (cdr (member-equal pid L))))
L))
:rule-classes :elim)
(defthm ustep-is-mstep
(implies (and (not (member-equal pid L))
(bound? pid (pida s))
(equal code (code s)))
(equal (ustep (make-ustate (binding pid (pida s))
(mem s)
code)
(mem (mrun L s)))
(make-ustate
(binding pid
(pida
(mstep pid (mrun L s))))
(mem (mstep pid (mrun L s)))
(code s))))
:hints (("Goal" :do-not-induct t
:in-theory (set-difference-theories (enable ustep mstep)
'(load-step
incr-step
cas-step
br-step
jump-step)))))
(defthm member-equal-initial-subsequence
(not (member-equal pid (initial-subsequence L pid))))
(defmacro processp (pid s)
`(bound? ,pid (pida ,s)))
(defthm commutative-diagram
(implies (processp pid s)
(equal (urun (proj pid s)
(proj-oracle pid s L))
(proj pid (mrun L s))))
:hints (("Goal" :induct (mrun-is-urun-hint pid L s))))
(defthm umem-proj
(equal (umem (proj pid s)) (mem s)))
(defthm commutative-diagram-corollary
(implies (processp pid s)
(equal (mem (mrun L s))
(umem
(urun (proj pid s)
(proj-oracle pid s L)))))
:hints (("Goal" :in-theory (disable proj proj-oracle))))
(in-theory (disable commutative-diagram-corollary))
; if you enable this rule, (disable ustep-is-mstep commutative-diagram)!
(defthm len-new-oracle
(equal (len (new-oracle pid s L))
(len L)))
(defthm uls-proj (equal (uls (proj pid s)) (ls pid s)))
(defthm ucode-proj (equal (ucode (proj pid s)) (code s)))
; ----------------------------------------------------------------------------
; We now develop the Preorder Projection Theorem.
(encapsulate
((P (s) t)
(R (mem1 mem2) t))
(local
(defun P (s) (declare (ignore s)) t))
(local
(defun R (mem1 mem2) (declare (ignore mem1 mem2)) t))
(defthm R-reflexive
(implies (P s)
(R (mem s) (mem s))))
(defthm R-transitive
(implies (and (R mem1 mem2)
(R mem2 mem3))
(R mem1 mem3)))
(defthm P-mstep
(implies (P s) (P (mstep pid s))))
(defthm R-mstep
(implies (P s)
(R (mem s) (mem (mstep pid s))))))
(defun R-urunp (us M)
(declare (xargs :measure (acl2-count M)))
(cond ((endp M) t)
((R (umem us) (car M))
(R-urunp (ustep us (car M))
(cdr M)))
(t nil)))
(defthm R-urunp-opener
(equal (R-urunp us (cons mem M))
(and (R (umem us) mem)
(R-urunp (ustep us mem) M))))
(defthm P-mrun
(implies (P s) (P (mrun L s))))
(defthm R-mrun
(implies (P s)
(R (mem s) (mem (mrun L s))))
:hints (("Subgoal *1/3"
:use
(:instance R-mstep
(pid (car L))
(s s))
:in-theory (disable R-mstep))))
(defthm preorder-projection
(implies (and (P s)
(bound? pid (pida s)))
(R-urunp (proj pid s) (proj-oracle pid s L)))
:hints (("Goal" :induct (mrun-is-urun-hint pid L s))))
; The above theorem is interesting because it lets us conclude
; something about a urun of a projection from an analogous thing about
; a mrun. I show how to use this in the example proof below.
; -----------------------------------------------------------------------------
; Utilities about occurrences and oracles...
(defun occurrences (pid L)
(cond ((endp L) 0)
((equal pid (car L))
(1+ (occurrences pid (cdr L))))
(t (occurrences pid (cdr L)))))
(defthm equal-len-n
(implies (and (syntaxp (quotep n))
(< 0 n))
(equal (equal (len x) n)
(and (consp x)
(equal (len (cdr x)) (1- n))))))
(in-theory (disable equal-len-n))
(defthm equal-len-0
(equal (equal (len x) 0) (endp x)))
(defthm alt-occurrences
(equal (occurrences pid L)
(if (member-equal pid L)
(1+ (occurrences pid (cdr (member-equal pid L))))
0))
:rule-classes
((:definition :clique (occurrences)
:controller-alist ((occurrences nil t)))))
(defthm occurrences-non-0-implies-member-equal
(implies (< 0 (occurrences pid L))
(member-equal pid L)))
(defthm len-partition-oracle
(equal (len (partition-oracle pid L))
(1+ (occurrences pid L))))
(defthm len-proj-oracle
(equal (len (proj-oracle pid s L))
(1+ (occurrences pid L))))
; -----------------------------------------------------------------------------
; We now develop the proof of the program per se.
(defmacro ctr (s)
`(binding 'ctr (mem ,s)))
(defun good-local-statep (ls)
(and (integerp (pcn ls))
(<= 0 (pcn ls))
(<= (pcn ls) 4)
(integerp (binding 'old (regs ls)))
(implies (equal (pcn ls) 2)
(equal (1+ (binding 'old (regs ls)))
(binding 'new (regs ls))))))
(defun pida-invariantp (alist)
(cond ((endp alist)
(equal alist nil))
(t (and (consp (car alist))
(good-local-statep (cdar alist))
(pida-invariantp (cdr alist))))))
(defun good-statep (s)
(and (pida-invariantp (pida s))
(integerp (ctr s))
(equal (code s)
'((LOAD OLD CTR)
(INCR NEW OLD)
(CAS CTR OLD NEW)
(BR NEW 1)
(JUMP 0)))))
; See the essay at the bottom for some explanation of the next two
; lemmas.
(defthm good-local-statep!
(implies (good-local-statep ls)
(and (integerp (pcn ls))
(<= 0 (pcn ls))
(<= (pcn ls) 4)
(integerp (binding 'old (regs ls)))
(implies (equal (pcn ls) 2)
(equal (1+ (binding 'old (regs ls)))
(binding 'new (regs ls))))))
:rule-classes
((:type-prescription
:corollary
(implies (good-local-statep ls)
(integerp (pcn ls))))
(:type-prescription
:corollary
(implies (good-local-statep ls)
(integerp (binding 'old (regs ls)))))
(:rewrite
:corollary
(implies (and (good-local-statep ls)
(equal (pcn ls) 2))
(equal (binding 'new (regs ls)) ; note orientation
(1+ (binding 'old (regs ls))))))
(:linear
:corollary
(implies (good-local-statep ls)
(and (<= 0 (pcn ls))
(<= (pcn ls) 4))))))
(defthm pida-invariantp!
(implies (and (pida-invariantp alist)
(bound? pid alist))
(good-local-statep (binding pid alist)))
:rule-classes :type-prescription)
(defthm pida-invariantp-bind
(implies (pida-invariantp alist)
(equal (pida-invariantp (bind pid ls alist))
(good-local-statep ls))))
; The following theorem was not needed by ACL2 Version 1.9, which existed in
; 1998 when this proof was first done. The reason is that Version 1.9
; opened recursive functions more agressively than subsequent ACL2 releases.
; Basically (nth pcn '((LOAD OLD CTR) ...)) is the expansion of the
; (current-instruction s) under our good-statep hypothesis. When trying to
; prove that good-statep is preserved by mstep, one must consider which
; instruction mstep is executing. When, for example, executing a LOAD into
; OLD, we must be able to show that the memory location being laoded is an
; integer. But the only such knowledge about mem is for CTR. Thus, the only
; legal LOAD to OLD is (LOAD OLD CTR). Any other LOAD to OLD would break the
; good-statep invariant. So the only way to prove the invariance of
; good-statep is to make sure that each of the five instructions in our program
; preserves it. Thus rule forces ACL2 to case split on the pcn and consider
; each of the possbilities.
(defthm pcn-case-split
(equal (nth pcn
'((LOAD OLD CTR) ; 0
(INCR NEW OLD) ; 1
(CAS CTR OLD NEW) ; 2
(BR NEW 1) ; 3
(JUMP 0))) ; 4
(case pcn
(0 '(LOAD OLD CTR))
(1 '(INCR NEW OLD))
(2 '(CAS CTR OLD NEW))
(3 '(BR NEW 1))
(4 '(JUMP 0))
(otherwise (if (zp pcn)
'(LOAD OLD CTR)
nil)))))
; So it is easy to see now that good-statep is preserved by mstep,
(defthm good-statep-mstep
(implies (good-statep s)
(good-statep (mstep pid s)))
:hints
(("Goal" :in-theory (enable mstep))))
; and hence by mrun.
(defthm good-statep-mrun
(implies (good-statep s)
(good-statep (mrun L s))))
; Now we define the predicate on M that insures that memory values are
; legal. They dominate the value seen by the selected process.
(defun ascendingp (us M)
(declare (xargs :measure (acl2-count M)))
(cond ((endp M) t)
(t (and (integerp (binding 'ctr (car M)))
(<= (binding 'ctr (umem us))
(binding 'ctr (car M)))
(ascendingp (ustep us (car M)) (cdr M))))))
; To prove that by our preorder-projection we really just need two
; facts:
(defthm ctr-<=-ctr-mstep
(implies (good-statep s)
(<= (ctr s) (ctr (mstep pid s))))
:rule-classes :linear
:hints (("Goal" :in-theory (enable mstep))))
(defthm integerp-ctr-mstep
(implies (good-statep s)
(integerp (ctr (mstep pid s))))
:hints (("Goal" :in-theory (enable mstep)))
:rule-classes (:rewrite :type-prescription))
; So the projection property tells us that the projected state/oracle
; ascend.
(defthm preorder-projection-corollary
(implies (and (good-statep s)
(processp pid s))
(ascendingp (proj pid s)
(proj-oracle pid s L)))
:hints (("Goal"
:use
((:functional-instance
preorder-projection
(P good-statep)
(R (lambda (mem1 mem2)
(and (integerp (binding 'ctr mem2))
(<= (binding 'ctr mem1) (binding 'ctr mem2)))))
(R-urunp ascendingp))))))
(defthm ascendingp-opener
(equal (ascendingp us (cons mem M))
(and (integerp (binding 'ctr mem))
(<= (binding 'ctr (umem us))
(binding 'ctr mem))
(ascendingp (ustep us mem) M)))
:hints (("goal" :in-theory (disable ustep))))
(defthm crux-lemma
(let ((M (list* M1 M2 M3 M4 M5 M6 M7 end)))
(implies (and (endp end)
(ascendingp us M)
(good-local-statep (uls us))
(equal (ucode us)
'((LOAD OLD CTR)
(INCR NEW OLD)
(CAS CTR OLD NEW)
(BR NEW 1)
(JUMP 0))))
(< (binding 'ctr (umem us)) (binding 'ctr (umem (urun us M))))))
:rule-classes nil
:hints
(("Goal" :cases ((equal (pcn (uls us)) 0)
(equal (pcn (uls us)) 1)
(equal (pcn (uls us)) 2)
(equal (pcn (uls us)) 3)
(equal (pcn (uls us)) 4)))))
(defun good-u-statep (us)
(and (good-local-statep (uls us))
(equal (ucode us)
'((LOAD OLD CTR)
(INCR NEW OLD)
(CAS CTR OLD NEW)
(BR NEW 1)
(JUMP 0)))))
(defthm crux
(implies (and (equal (len M) 7)
(ascendingp us M)
(good-u-statep us))
(< (binding 'ctr (umem us))
(binding 'ctr (umem (urun us M)))))
:rule-classes
((:rewrite :corollary
(implies (and (equal (len M) 7)
(equal mem (umem us))
(ascendingp us M)
(good-u-statep us))
(< (binding 'ctr mem)
(binding 'ctr (umem (urun us M)))))))
:hints (("Goal" :do-not-induct t
:use
(:instance crux-lemma
(m1 (car M))
(m2 (cadr M))
(m3 (caddr M))
(m4 (cadddr M))
(m5 (car (cddddr M)))
(m6 (cadr (cddddr M)))
(m7 (caddr (cddddr M)))
(end (cdddr (cddddr M))))
:in-theory
(set-difference-theories
(enable equal-len-n)
'(len umem ascendingp good-local-statep uls
ucode urun
urun-opener
ascendingp-opener)))))
(defthm lemma
(implies (and (good-statep s)
(processp pid s)
(equal 6 (occurrences pid L)))
(< (ctr s)
(ctr (mrun L s))))
:rule-classes nil
:hints (("Goal"
:do-not-induct t
:in-theory
(union-theories '(commutative-diagram-corollary)
(disable proj proj-oracle
good-local-statep
ustep-is-mstep
commutative-diagram)))))
; -----------------------------------------------------------------------------
; Essay on good-local-statep! and pida-invariantp!
; Our goal formulas will have a hypothesis of (good-statep s). This
; will expand to (pida-invariantp (pida s)). From this we will have
; to be able to conclude such things as (integerp (pcn (binding pid
; (pida s)))). How do we arrange for this to happen?
; Since the integerp fact is "type-like" we build it in at the
; :type-prescription level. The lemma good-local-statep! causes us to
; backchain from (integerp (pcn ...)) to (good-local-statep ls). But in
; general we won't have (good-local-statep ls) in the goal. Instead we
; have to continue to backchain to (pida-invariantp ...).
; Furthermore, we have to do this backchain at the :type-prescription
; level for it to have the desired effect.
; We could reduce these two lemmas to one, eliminating the talk about
; good-local-statep and going directly from pida-invariantp down to
; integerp. I chose this way because it is more illustrative of a
; hierarchical decomposition.
; -----------------------------------------------------------------------------
; I now finish the proof off. We're here concerned with the Pigeon
; Hold Principle and the problem that the actual oracle may contain
; more than 6 occurrences of pid.
(defun cardinality (x)
(if (endp x)
0
(if (member-equal (car x) (cdr x))
(cardinality (cdr x))
(1+ (cardinality (cdr x))))))
(defun delete-all (x L)
(cond ((endp L) L)
((equal x (car L)) (delete-all x (cdr L)))
(t (cons (car L) (delete-all x (cdr L))))))
(defthm acl2-count-delete-all
(<= (acl2-count (delete-all x L))
(acl2-count L))
:rule-classes :linear)
(defthm not-member-equal-delete-all
(implies (not (member-equal x L))
(equal (delete-all x L) L)))
(defthm member-equal-delete-all
(iff (member-equal x (delete-all y L))
(if (equal x y)
nil
(member-equal x L))))
(defun choose (n L)
(cond ((endp L) nil)
((<= n (occurrences (car L) (cdr L))) (car L))
(t (choose n
(delete-all (car L) (cdr L))))))
(defthm cardinality<=len
(<= (cardinality x) (len x))
:rule-classes :linear)
(defun weird-induction-hint (L)
(cond ((endp L) t)
(t (weird-induction-hint (delete-all (car L) (cdr L))))))
(in-theory (disable initial-subsequence-elim))
(defthm cardinality-delete-all
(equal (cardinality (delete-all x L))
(if (member-equal x L)
(1- (cardinality L))
(cardinality L))))
(defthm alt-cardinality
(equal (cardinality L)
(if (endp L)
0
(1+ (cardinality (delete-all (car L) (cdr L))))))
:rule-classes
((:definition :clique (cardinality)
:controller-alist ((cardinality t)))))
(in-theory (disable cardinality-delete-all))
(defthm len-delete-all
(equal (len (delete-all x L))
(- (len L) (occurrences x L))))
(defthm occurrence-delete-all
(equal (occurrences x (delete-all y L))
(if (equal x y)
0
(occurrences x L))))
; Here is the crucial fact:
(defthm pigeon-hole-principle
(implies (and (integerp i)
(<= 0 i)
(< (* i (cardinality L)) (len L)))
(< i (occurrences (choose i L) L)))
:rule-classes :linear)
; Given (< (* 5 (cardinality L)) (len L)) we know (<= 6 (occurrences
; (choose 5 L) L)). Thus we can partition L into two bits:
(defun first-n-occurrences (n x L)
(cond ((zp n) nil)
((endp L) L)
((equal x (car L))
(cons (car L) (first-n-occurrences (1- n) x (cdr L))))
(t (cons (car L) (first-n-occurrences n x (cdr L))))))
(defun all-but-first-n-occurrences (n x L)
(cond ((zp n) L)
((endp L) L)
((equal x (car L))
(all-but-first-n-occurrences (1- n) x (cdr L)))
(t (all-but-first-n-occurrences n x (cdr L)))))
(defthm partition-after-first-n
(implies (<= n (occurrences x L))
(equal L (append (first-n-occurrences n x L)
(all-but-first-n-occurrences n x L))))
:rule-classes nil)
; Of course, if everything in L is bound in (pida s), then choose
; returns something bound in (pida s).
(defun all-bound? (L alist)
(cond ((endp L) t)
(t (and (bound? (car L) alist)
(all-bound? (cdr L) alist)))))
(defthm all-bound?-implies-choose-bound
(implies (and (all-bound? L alist)
(integerp n)
(<= 0 n)
(< n (occurrences (choose n L) L)))
(bound? (choose n L) alist)))
; Given that, we can apply lemma to get past the first 6
; steps of the chosen pid, and then appeal to the fact
(defthm pida-invariantp-mstep
(implies (good-statep s)
(pida-invariantp (pida (mstep pid s))))
:hints (("Goal" :in-theory (disable good-statep-mstep)
:use good-statep-mstep)))
(defthm ctr-<=-ctr-mrun
(implies (good-statep s)
(<= (ctr s) (ctr (mrun L s))))
:rule-classes :linear)
; Finally, to appeal to lemma we need to know:
(defthm occurrences-first-n-occurrences
(implies (and (integerp n)
(<= 0 n)
(<= n (occurrences x L)))
(equal (occurrences x (first-n-occurrences n x L))
n)))
(defun every-element-a-processp (L s)
(all-bound? L (pida s)))
; So here is our main goal, stated simply:
(defthm theorem
(implies (and (good-statep s)
(every-element-a-processp L s)
(< (* 5 (cardinality L)) (len L)))
(< (ctr s) (ctr (mrun L s))))
:hints (("Goal"
:in-theory (disable good-statep)
:use ((:instance lemma (s s)
(pid (choose 5 L))
(L (first-n-occurrences 6 (choose 5 L) L)))
(:instance partition-after-first-n
(n 6)
(x (choose 5 L))
(L L))))))
; -----------------------------------------------------------------------------
; I finish by proving that good states exist. In particular, the
; function n-processor-s0 constructs one for n processors,
; provided you give it a memory with a integerp ctr.
(defun pida0 (n)
(cond ((zp n) nil)
(t (cons (cons n (make-ls 0 '((old . 0)(new . 0))))
(pida0 (1- n))))))
(defun n-processor-s0 (n mem)
(make-mstate (pida0 n)
mem
'((LOAD OLD CTR)
(INCR NEW OLD)
(CAS CTR OLD NEW)
(BR NEW 1)
(JUMP 0))))
(defthm good-states-exist
(implies (integerp (binding 'ctr mem))
(good-statep (n-processor-s0 n mem))))
|