package codex
sectionYPositions = computeSectionYPositions($el), 10)"
x-init="setTimeout(() => sectionYPositions = computeSectionYPositions($el), 10)"
>
The Codex library for building static analysers based on abstract interpretation
Install
dune-project
Dependency
Authors
Maintainers
Sources
1.0-rc4.tar.gz
md5=bc7266a140c6886add673ede90e335d3
sha512=8da42c0ff2c1098c5f9cb2b5b43b306faf7ac93b8f5ae00c176918cee761f249ff45b29309f31a05bbcf6312304f86a0d5a000eb3f1094d3d3c2b9b4c7f5c386
doc/src/codex.framac_ival/float_interval.ml.html
Source file float_interval.ml
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(**************************************************************************) (* This file is part of the Codex semantics library. *) (* *) (* Copyright (C) 2013-2025 *) (* CEA (Commissariat à l'énergie atomique et aux énergies *) (* alternatives) *) (* *) (* you can redistribute it and/or modify it under the terms of the GNU *) (* Lesser General Public License as published by the Free Software *) (* Foundation, version 2.1. *) (* *) (* It is distributed in the hope that it will be useful, *) (* but WITHOUT ANY WARRANTY; without even the implied warranty of *) (* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the *) (* GNU Lesser General Public License for more details. *) (* *) (* See the GNU Lesser General Public License version 2.1 *) (* for more details (enclosed in the file LICENSE). *) (* *) (**************************************************************************) module Comp = Abstract_interp.Comp open Bottom.Type type round = Float_sig.round = Up | Down | Near | Zero type prec = Float_sig.prec = Single | Double | Long_Double | Real module Make (F: Float_sig.S) = struct (* Definitions of useful floating-point constants. *) module Cst = struct let pos_zero prec = F.of_float Near prec 0. let neg_zero prec = F.of_float Near prec (- 0.) let pos_infinity prec = F.of_float Near prec infinity let neg_infinity prec = F.of_float Near prec neg_infinity (* Maximum value M such that all integers 0 < n <= M are exactly representable in the given precision. It is 2^53 for double and 2^24 for single precision. *) let max_precise_integer prec = let f = match prec with | Single -> 2. ** 24. | Double | Long_Double -> 2. ** 53. | Real -> infinity in F.of_float Near prec f (* Maximum odd integer representable in the given precision. It is 2^53-1 for double and 2^24-1 for single precision. *) let max_odd_integer prec = let f = match prec with | Single -> 2. ** 24. -. 1. | Double -> 2. ** 53. -. 1. | Real | Long_Double -> infinity in F.of_float Near prec f (* Successor of the maximum non-integer representable in the given precision. It is 2^52 in double and 2^23 in single precision. *) let ceil_of_max_non_integer prec = let f = match prec with | Single -> 2. ** 23. | Double -> 2. ** 52. | Real | Long_Double -> infinity in F.of_float Near prec f (* Returns 1.0 if [f] is positive, and -1.0 if [t] is negative. [f] can be infinite or a (positive or negative) zero, but not NaN. *) let sign prec f = let s = if F.is_negative f then -1. else 1. in F.of_float Near prec s end (* IEEE and non-IEEE comparisons. *) module Cmp = struct (* Constants in single precision: should be used only for comparisons, and not to create intervals of arbitrary precisions. *) let pos_zero = Cst.pos_zero Float_sig.Single let neg_zero = Cst.neg_zero Float_sig.Single let one = F.of_float Near Float_sig.Single 1. let pos_infinity = Cst.pos_infinity Float_sig.Single let neg_infinity = Cst.neg_infinity Float_sig.Single let equal_ieee f1 f2 = F.cmp_ieee f1 f2 = 0 let le_ieee f1 f2 = F.cmp_ieee f1 f2 <= 0 let lt_ieee f1 f2 = F.cmp_ieee f1 f2 < 0 let ge_ieee f1 f2 = F.cmp_ieee f1 f2 >= 0 let gt_ieee f1 f2 = F.cmp_ieee f1 f2 > 0 let is_a_zero f = equal_ieee pos_zero f let is_pos_infinity f = equal_ieee pos_infinity f let is_neg_infinity f = equal_ieee neg_infinity f let equal f1 f2 = F.compare f1 f2 = 0 let le f1 f2 = F.compare f1 f2 <= 0 let lt f1 f2 = F.compare f1 f2 < 0 let ge f1 f2 = F.compare f1 f2 >= 0 let gt f1 f2 = F.compare f1 f2 > 0 let min f1 f2 = if le f1 f2 then f1 else f2 let max f1 f2 = if le f1 f2 then f2 else f1 (* next_float_ieee -0. = next_float_ieee 0. *) let next_float_ieee prec f = let f = if equal f neg_zero then Cst.pos_zero prec else f in F.next_float prec f (* prev_float_ieee 0. = next_float_ieee -0. *) let prev_float_ieee prec f = let f = if equal f pos_zero then Cst.neg_zero prec else f in F.prev_float prec f end module FRange = struct (* invariants for intervals Itv (b, e, nan): - b and e are not NaN; - b <= e *) type t = | Itv of F.t * F.t * bool | NaN let nan = NaN let inject ?(nan=false) b e = if F.is_nan b || F.is_nan e || F.compare b e > 0 then Codex_log.fatal "Invalid bounds for float interval:@ %a .. %a@." F.pretty b F.pretty e; Itv (b, e, nan) let inject_after_tighten prec ~nan b e = let b = F.round_to_precision Up prec b and e = F.round_to_precision Down prec e in if Cmp.le b e then `Value (inject ~nan b e) else if nan then `Value NaN else `Bottom let add_nan = function | NaN as f -> f | Itv (_, _, true) as f -> f | Itv (b, e, false) -> Itv (b, e, true) let add_pos_infinity prec t = let pos_inf = Cst.pos_infinity prec in match t with | NaN -> Itv (pos_inf, pos_inf, true) | Itv (b, _e, nan) -> Itv (b, pos_inf, nan) let add_neg_infinity prec t = let neg_inf = Cst.neg_infinity prec in match t with | NaN -> Itv (neg_inf, neg_inf, true) | Itv (_b, e, nan) -> Itv (neg_inf, e, nan) end (* ------------------------------------------------------------------------ Datatype ------------------------------------------------------------------------ *) type t = FRange.t (* let structural_descr = * Structural_descr.t_sum [| * [| F.packed_descr; F.packed_descr; Structural_descr.p_bool |]; * [| |] * |] * let packed_descr = Structural_descr.pack structural_descr *) let compare x y = match x, y with | FRange.Itv (b1, e1, n1), FRange.Itv (b2, e2, n2) -> let c = Stdlib.compare n1 n2 in if c <> 0 then c else let r = F.compare b1 b2 in if r <> 0 then r else F.compare e1 e2 | FRange.Itv _, FRange.NaN -> -1 | FRange.NaN, FRange.Itv _ -> 1 | FRange.NaN, FRange.NaN -> 0 let equal x y = match x,y with | FRange.Itv (b1, e1, n1), FRange.Itv (b2, e2, n2) -> Cmp.equal b1 b2 && Cmp.equal e1 e2 && n1 = n2 | FRange.NaN, FRange.NaN -> true | _ -> false let pretty_or_zero fmt f = if Cmp.(equal f pos_zero) then Format.fprintf fmt "0" else F.pretty fmt f let pretty fmt = function | FRange.Itv (b, e, nan) -> if Cmp.equal b e then Format.fprintf fmt "{%a%t}" pretty_or_zero b (fun fmt -> if nan then Format.pp_print_string fmt ";NaN") else let print_nan fmt nan = if nan then Format.fprintf fmt " %s {NaN}" ("U"(* Unicode.union_string ()*) ) in Format.fprintf fmt "[%a .. %a]%a" F.pretty b F.pretty e print_nan nan | FRange.NaN -> Format.fprintf fmt "NaN" let hash = function | FRange.Itv (b, e, n) -> (2 * F.hash b) + (5 * F.hash e) + (7 * Hashtbl.hash n) | FRange.NaN -> 3 (* ------------------------------------------------------------------------ Constants, injections, projections ------------------------------------------------------------------------ *) let min_and_max = function | FRange.Itv (b, e, nan) -> Some (b, e), nan | FRange.NaN -> None, true let nan = FRange.nan let inject = FRange.inject let singleton x = FRange.inject x x let pos_zero prec = singleton (Cst.pos_zero prec) let one prec = singleton (F.of_float Near prec 1.) let pos_infinity prec = singleton (Cst.pos_infinity prec) let _neg_infinity prec = singleton (Cst.neg_infinity prec) let minus_one_one prec ~nan = FRange.inject ~nan (F.of_float Near prec (-1.)) (F.of_float Near prec 1.) let m_pi = 3.1415929794311523 (* single-precision *) let m_minus_pi = -. m_pi let minus_pi_pi prec ~nan = FRange.inject ~nan (F.of_float Near prec m_minus_pi) (F.of_float Near prec m_pi) let top_f prec ~nan = FRange.inject ~nan (Cst.neg_infinity prec) (Cst.pos_infinity prec) let top prec = top_f prec ~nan:true let top_finite prec = if F.is_exact prec then let b = F.next_float prec (Cst.neg_infinity prec) in let e = F.prev_float prec (Cst.pos_infinity prec) in FRange.inject ~nan:false b e else top_f prec ~nan:false (* Splits the interval [b; e] into two intervals (neg, pos) such that [neg] (resp. [pos]) contains only negative (resp. positive) floats. *) let split_by_sign prec ((b, e) as x) = if Cmp.(le e neg_zero) then `Value x, `Bottom else if Cmp.(ge b pos_zero) then `Bottom, `Value x else `Value (b, Cst.neg_zero prec), `Value (Cst.pos_zero prec, e) (* Returns true iff [f] represents an odd integer. *) let is_odd prec f = let two = F.of_float Near prec 2. in Cmp.equal (F.abs (F.fmod Near prec f two)) Cmp.one (* [split_by_parity (b, e)] returns [even_bounds, odd_bounds], where: - [even_bounds] are the min and max even integers enclosed between b and e (or None if there is no such even integer). - [odd_bounds] are the min and max odd integers (representable as floating- point values) enclosed between b and e (or None if there is no such odd integer). These bounds cannot be infinite. *) let split_by_parity prec (b, e) = let b = F.ceil b and e = F.floor e in if Cmp.gt_ieee b e then None, None else if Cmp.equal_ieee b e then let x = (b, e) in if is_odd prec b then None, Some x else Some x, None else let one = F.of_float Near prec 1. in let min_even, min_odd = if is_odd prec b (* No rounding errors may happen below because odd numbers are bounded. *) then F.add Near prec b one, b else b, Cmp.max (F.neg (Cst.max_odd_integer prec)) (F.add Near prec b one) and max_even, max_odd = if is_odd prec e (* No rounding errors may happen below because odd numbers are bounded. *) then F.sub Near prec e one, e else e, Cmp.min (Cst.max_odd_integer prec) (F.sub Near prec e one) in let even = Some (min_even, max_even) and odd = if Cmp.lt_ieee max_odd min_odd then None else Some (min_odd, max_odd) in even, odd (* ----------------------------------------------------------------------- Lattice ----------------------------------------------------------------------- *) let is_included x1 x2 = match x1, x2 with | FRange.Itv (b1, e1, n1), FRange.Itv (b2, e2, n2) -> F.compare b2 b1 <= 0 && F.compare e1 e2 <= 0 && (not n1 || n2) | FRange.NaN, FRange.Itv (_, _, true) -> true | FRange.NaN, FRange.NaN -> true | _ -> false let join f1 f2 = match f1, f2 with | FRange.Itv (b1, e1, n1), FRange.Itv (b2, e2, n2) -> FRange.inject ~nan:(n1 || n2) (Cmp.min b1 b2) (Cmp.max e1 e2) | (FRange.Itv (b1, e1, _), FRange.NaN) | (FRange.NaN, FRange.Itv (b1, e1, _)) -> FRange.inject ~nan:true b1 e1 | FRange.NaN, FRange.NaN -> FRange.nan let widen_down f = F.neg (F.widen_up (F.neg f)) let widen f1 f2 = assert (is_included f1 f2); match f1, f2 with | FRange.Itv (b1, e1, _), FRange.Itv (b2, e2, nan) -> let b = if Cmp.equal b2 b1 then b2 else widen_down b2 in let e = if Cmp.equal e2 e1 then e2 else F.widen_up e2 in (** widen_up and down produce double only if the input is a double *) FRange.inject ~nan b e | FRange.NaN, f2 -> f2 | FRange.Itv _, FRange.NaN -> assert false let meet f1 f2 = match f1, f2 with | FRange.Itv (b1, e1, n1), FRange.Itv (b2, e2, n2) -> let is_finite = F.compare b2 e1 <= 0 && F.compare b1 e2 <= 0 in let is_nan = n1 && n2 in if is_finite || is_nan then let v = if is_finite then FRange.inject ~nan:is_nan (Cmp.max b1 b2) (Cmp.min e1 e2) else FRange.nan in `Value v else `Bottom | (FRange.Itv (_, _, true) | FRange.NaN), (FRange.Itv (_, _, true) | FRange.NaN) -> `Value FRange.nan | _ -> `Bottom let narrow = meet (* ------------------------------------------------------------------------ Tests ------------------------------------------------------------------------ *) let contains cst = function | FRange.NaN -> false | FRange.Itv (b, e, _) -> Cmp.le b cst && Cmp.ge e cst let contains_pos_zero t = contains Cmp.pos_zero t let contains_neg_zero t = contains Cmp.neg_zero t let contains_a_zero = function | FRange.Itv (b, e, _) -> Cmp.(le_ieee b pos_zero) && Cmp.(ge_ieee e pos_zero) | FRange.NaN -> false let contains_non_zero = function | FRange.Itv (b, e, nan) -> nan || Cmp.(lt_ieee b pos_zero) || Cmp.(gt_ieee e pos_zero) | FRange.NaN -> true let contains_strictly_pos = function | FRange.Itv (_, e, _) -> Cmp.(gt_ieee e pos_zero) | FRange.NaN -> false let contains_strictly_neg = function | FRange.Itv (b, _, _) -> Cmp.(lt_ieee b neg_zero) | FRange.NaN -> false let contains_strict_neg_finite (b, e) = Cmp.(gt_ieee e neg_infinity) && Cmp.(lt b neg_zero) let contains_finite_noninteger prec (b, e) = let ib = F.ceil b in not ((Cmp.equal_ieee ib b && Cmp.equal_ieee b e) || Cmp.le_ieee e (F.neg (Cst.ceil_of_max_non_integer prec)) || Cmp.ge_ieee b (Cst.ceil_of_max_non_integer prec)) let contains_pos_infinity = function | FRange.Itv (_, e, _) -> Cmp.is_pos_infinity e | FRange.NaN -> false let contains_neg_infinity = function | FRange.Itv (b, _, _) -> Cmp.is_neg_infinity b | FRange.NaN -> false let contains_infinity f = contains_pos_infinity f || contains_neg_infinity f let contains_nan = function | FRange.NaN -> true | FRange.Itv (_, _, nan) -> nan let is_singleton = function | FRange.NaN -> true | FRange.Itv (b, e, nan) -> Cmp.equal b e && not nan let is_one = function | FRange.NaN -> false | FRange.Itv (b, e, nan) -> Cmp.(equal b one) && Cmp.(equal e one) && not nan let is_a_zero = function | FRange.NaN -> false | FRange.Itv (b, e, nan) -> not nan && Cmp.is_a_zero b && Cmp.is_a_zero e let if_not_nan = function | FRange.NaN -> assert false | FRange.Itv (b, e, _) -> b, e let is_not_nan = function | FRange.NaN -> Comp.False | FRange.Itv (_b, _e, nan) -> if nan then Comp.Unknown else Comp.True let is_finite = function | FRange.NaN -> Comp.False | FRange.Itv (b, e, nan) -> if Cmp.is_neg_infinity e || Cmp.is_pos_infinity b then Comp.False else if nan || Cmp.is_neg_infinity b || Cmp.is_pos_infinity e then Comp.Unknown else Comp.True let is_negative = function | FRange.Itv (b, e, false) -> if F.is_negative e then Comp.True else if not (F.is_negative b) then Comp.False else Comp.Unknown | FRange.Itv (_, _, true) | FRange.NaN -> Comp.Unknown let backward_is_not_nan = function | FRange.NaN -> `Bottom | FRange.Itv (b, e, _) -> `Value (FRange.inject ~nan:false b e) let backward_is_finite prec = function | FRange.NaN -> `Bottom | FRange.Itv (b, e, _) as f -> if Cmp.equal b e && F.is_infinite b then `Bottom (* [f] is exactly an infinite, we can return `Bottom even in the [Real] case *) else narrow (top_finite prec) f let has_greater_min_bound t1 t2 = match t1, t2 with | FRange.Itv (m1, _, _), FRange.Itv (m2, _, _) -> F.compare m1 m2 | FRange.NaN, FRange.Itv _ -> 1 | FRange.Itv _, FRange.NaN -> -1 | FRange.NaN, FRange.NaN -> 0 let has_smaller_max_bound t1 t2 = match t1, t2 with | FRange.Itv (_, m1, _), FRange.Itv (_, m2, _) -> F.compare m2 m1 | FRange.NaN, FRange.Itv _ -> 1 | FRange.Itv _, FRange.NaN -> -1 | FRange.NaN, FRange.NaN -> 0 (* ------------------------------------------------------------------------ Comparisons ------------------------------------------------------------------------ *) let forward_eq (b1,e1) (b2,e2) = let not_intersects = Cmp.lt_ieee e2 b1 || Cmp.lt_ieee e1 b2 in if not_intersects then Comp.False else if Cmp.equal_ieee b1 e1 && Cmp.equal_ieee b2 e2 then Comp.True else Comp.Unknown let forward_le (b1, e1) (b2, e2) = if Cmp.le_ieee e1 b2 then Comp.True else if Cmp.lt_ieee e2 b1 then Comp.False else Comp.Unknown let forward_lt (b1, e1) (b2, e2) = if Cmp.lt_ieee e1 b2 then Comp.True else if Cmp.le_ieee e2 b1 then Comp.False else Comp.Unknown let forward_comp op f1 f2 = match f1, f2 with | FRange.NaN, _ | _, FRange.NaN -> if op = Comp.Ne then Comp.True else Comp.False | FRange.Itv (b1, e1, nan1), FRange.Itv (b2, e2, nan2) -> let r = match op with | Comp.Le -> forward_le (b1, e1) (b2, e2) | Comp.Ge -> forward_le (b2, e2) (b1, e1) | Comp.Lt -> forward_lt (b1, e1) (b2, e2) | Comp.Gt -> forward_lt (b2, e2) (b1, e1) | Comp.Eq -> forward_eq (b1, e1) (b2, e2) | Comp.Ne -> Abstract_interp.inv_truth (forward_eq (b1, e1) (b2, e2)) in if nan1 || nan2 then if op = Comp.Ne then (match r with Comp.True -> Comp.True | _ -> Comp.Unknown) else (match r with Comp.False -> Comp.False | _ -> Comp.Unknown) else r (* This function intentionally returns different results with [e2 = -0.] and [e2 = 0.] *) let backward_le_aux prec (b1, e1) e2 = if not (Cmp.le b1 e2) then `Bottom else if Cmp.le e1 e2 then `Value (FRange.inject b1 e1) else FRange.inject_after_tighten prec ~nan:false b1 e2 (* This is the "real" backward transformer for [le], which does not distinguish [0.] and [-0.]. Thus we enlarge the bound in the "worst" direction. *) let backward_le prec (b1, e1) e2 = let e2 = if Cmp.is_a_zero e2 then Cst.pos_zero prec else e2 in backward_le_aux prec (b1, e1) e2 let backward_lt prec ((b1, e1) as f1) e2 = if Cmp.le_ieee e2 b1 then `Bottom else if F.is_exact prec || Cmp.equal b1 e1 then backward_le prec f1 (Cmp.prev_float_ieee prec e2) else (* On real we cannot be more precise than [le], except on zeros: at least get rid of the "bad" zero *) let e2 = if Cmp.is_a_zero e2 then Cst.neg_zero prec else e2 in backward_le_aux prec f1 e2 (* see comments in {!backward_le_aux} *) let backward_ge_aux prec (b1, e1) b2 = if not (Cmp.le b2 e1) then `Bottom else if Cmp.le b2 b1 then `Value (FRange.inject b1 e1) else FRange.inject_after_tighten prec ~nan:false b2 e1 (* see comments in {!backward_le} *) let backward_ge prec (b1, e1) b2 = let b2 = if Cmp.is_a_zero b2 then Cst.neg_zero prec else b2 in backward_ge_aux prec (b1, e1) b2 (* see comments in {!backward_gt} *) let backward_gt prec ((b1, e1) as f1) b2 = if Cmp.le_ieee e1 b2 then `Bottom else if F.is_exact prec || Cmp.equal b1 e1 then backward_ge prec f1 (Cmp.next_float_ieee prec b2) else let b2 = if Cmp.is_a_zero b2 then Cst.pos_zero prec else b2 in backward_ge_aux prec f1 b2 (** The operands cannot be {!Nan} *) let backward_comp_left_true_finite op prec f1' f2' = let f1 = if_not_nan f1' in let (b2, e2) = if_not_nan f2' in match op with | Comp.Le -> backward_le prec f1 e2 | Comp.Ge -> backward_ge prec f1 b2 | Comp.Lt -> backward_lt prec f1 e2 | Comp.Gt -> backward_gt prec f1 b2 | Comp.Eq -> (* -0 and +0 must not be distinguished here *) let f2 = if contains_a_zero f2' then join f2' (FRange.inject (Cst.neg_zero prec) (Cst.pos_zero prec)) else f2' in narrow f1' f2 | Comp.Ne -> (* compute (f1 ∩ [-infty,min[ ) ∪ (f1 ∩ ]max,infty]) *) let before_or_after min max = Bottom.join join (backward_lt prec f1 min) (backward_gt prec f1 max) in (* As usual, we cannot reduce if [f2] is not a singleton, except that the two zeros are a kind of singleton. Checking whether [f2] is on a frontier of [f1] is not obvious because of the multiple cases (and [allmodes]) so we use the transformers for [lt] instead. *) if is_a_zero f2' then before_or_after (Cst.neg_zero prec) (Cst.pos_zero prec) else if is_singleton f2' then before_or_after b2 b2 else `Value f1' (* Applies [backward f1 f2] and removes NaN from [f1] and [f2]. *) let backward_comp_no_nan backward_finite f1 f2 = match f1, f2 with | FRange.NaN, _ | _, FRange.NaN -> `Bottom | FRange.Itv (b, e, nan), FRange.Itv _ -> let f1 = if nan then FRange.inject ~nan:false b e else f1 in backward_finite f1 f2 (* Applies [backward f1 f2] but preserves NaN from [f1] and [f2]. *) let backward_comp_with_nan backward_finite f1 f2 = if contains_nan f2 then `Value f1 else match f1 with | FRange.NaN -> `Value f1 | FRange.Itv (_, _, nan) -> let nan = if nan then `Value FRange.nan else `Bottom in Bottom.join join (backward_finite f1 f2) nan let backward_comp_left_true op prec = let backward_finite = backward_comp_left_true_finite op prec in if op = Comp.Ne then backward_comp_with_nan backward_finite else backward_comp_no_nan backward_finite let backward_comp_left_false op prec = let backward_finite = backward_comp_left_true_finite (Comp.inv op) prec in if op = Comp.Ne then backward_comp_no_nan backward_finite else backward_comp_with_nan backward_finite (* ------------------------------------------------------------------------ Simple arithmetic operations ------------------------------------------------------------------------ *) (* The functions defined using [exact_aux] below are, among other properties, (1) exact (the result as a real can always be represented exactly, in the good type), and (2) total. In particular, given a float 'x', 'ff x == (float)(f (double)x)'. Thus, in this module, the 'f' functions are also the non-f (since float32 are represented using double) *) let (>>) t f = match t with | FRange.NaN -> t | FRange.Itv (b, e, nan) -> FRange.inject ~nan (f b) (f e) let floor t = t >> F.floor let ceil t = t >> F.ceil let trunc t = t >> F.trunc let fround t = t >> F.fround let neg = function | FRange.Itv (b, e, nan) -> (* do not round because exact operation *) FRange.inject ~nan (F.neg e) (F.neg b) | FRange.NaN -> FRange.nan let abs prec = function | FRange.Itv (b, e, nan) as f -> if contains Cmp.pos_zero f then let zero = Cst.pos_zero prec in FRange.inject ~nan zero (Cmp.max (F.abs b) (F.abs e)) else (* f is either strictly positive or strictly negative *) if F.compare e Cmp.pos_zero < 0 then neg f else f | FRange.NaN as f -> f (* This monad returns a NaN if one operand can only be NaN, and lets the second function perform the computation if both operands contain a non-empty floating-point interval. *) let ( >>% ) = fun (x,y) f -> match x, y with | FRange.NaN, _ | _, FRange.NaN -> FRange.nan | FRange.Itv (b1, e1, nan1), FRange.Itv (b2, e2, nan2) -> let nan = nan1 || nan2 in f ~nan (b1, e1) (b2, e2) (* Auxiliary function used for the forward semantics of add, mul and div. For a monotonic function [op], the bounds of [[b1..e1] op [b2..e2]] are the minimum and maximum of [b1 op b2], [b1 op e2], [e1 op b2] and [e1 op e2]. NaN can be created from \infty - \infty, 0 * \infty, 0/0 and \infty / \infty, in which case the result contains NaN, and new operations are performed to take into account the results of values near \infty and 0. Beware that NaN and discontinuities occuring between the bounds of the arguments (i.e. on zeros, as an infinity is always a bound) should be checked and processed by the caller. *) let monotonic op prec x y = (x, y) >>% fun ~nan (b1, e1) (b2, e2) -> let nan = ref nan in (* Results of [op] applied to the bounds of the intervals, excluding NaN. *) let results = ref [] in (* When [a op b = NaN], performs new operations to take into account values near [a] (and near [b] with the same reasoning). For such a NaN from add, mul or div, [c op b] is constant for all values c <> a of the same sign. Thus, we can replace [a] by any of these values. - if [x] is a singleton or [-0 .. +0], there are no values other than the bounds to take into account; - otherwise, if [a] is infty, replace it by 1 with the sign of [a] (no risk of NaN with 1 on add, mul and div); - otherwise, if [a] is zero, the other bound [c] of the interval has the same sign as the values near [a] in the interval; as [c op b] is also computed, no need to perform a new operation. *) let treat_nan_result rnd a b = nan := true; if F.is_infinite a && not (Cmp.equal b1 e1) then results := op rnd prec (Cst.sign prec a) b :: !results; if F.is_infinite b && not (Cmp.equal b2 e2) then results := op rnd prec a (Cst.sign prec b) :: !results; in let op rnd x y = let r = op rnd prec x y in if F.is_nan r then treat_nan_result rnd x y else results := r :: !results in let compute rnd = results := []; op rnd e1 e2; op rnd e1 b2; op rnd b1 e2; op rnd b1 b2; in let rounding_mode = if F.is_exact prec then Near else Down in compute rounding_mode; let pos_inf = Cst.pos_infinity prec in let min = List.fold_left Cmp.min pos_inf !results in if not (F.is_exact prec) then compute Up; let neg_inf = Cst.neg_infinity prec in let max = List.fold_left Cmp.max neg_inf !results in if min > max then (assert !nan; FRange.nan) else FRange.inject ~nan:!nan min max let add prec = monotonic F.add prec let sub prec = monotonic F.sub prec let mul prec x y = let r = monotonic F.mul prec x y in (* A NaN may occur between the bounds of the intervals, on 0 * \infty. *) if (contains_infinity x && contains_a_zero y) || (contains_infinity y && contains_a_zero x) then FRange.add_nan r else r let div prec x y = let r = monotonic F.div prec x y in (* A NaN may occur between the bounds of the intervals, on 0/0. *) let nan = (contains_a_zero x && contains_a_zero y) in (* Treat the discontinuity around 0: divisions by 0 produce infinites. *) let pos_inf = contains Cmp.pos_zero y && contains_strictly_pos x || contains Cmp.neg_zero y && contains_strictly_neg x and neg_inf = contains Cmp.pos_zero y && contains_strictly_neg x || contains Cmp.neg_zero y && contains_strictly_pos x in let r = if pos_inf then FRange.add_pos_infinity prec r else r in let r = if neg_inf then FRange.add_neg_infinity prec r else r in if nan then FRange.add_nan r else r (* Could be improved a lot, cf [Marre10]. *) let backward_add_one prec ~other ~result = (* No reduction when the result contains an infinity, and when the result and the other operand contain NaN (as x + NaN = NaN for any x). *) if contains_infinity result || (contains_nan other && contains_nan result) then `Value (top prec) else (* Values that can lead to NaN in the result. *) let reduce_for_nan t = let t = if contains_pos_infinity other then FRange.add_neg_infinity prec t else t in if contains_neg_infinity other then FRange.add_pos_infinity prec t else t in let reduced_for_nan = if contains_nan result then `Value (reduce_for_nan FRange.nan) else `Bottom in (* Values that can lead to finite values in the result. *) let reduced_for_finite_values = match result, other with | FRange.NaN, _ | _, FRange.NaN -> `Bottom | FRange.Itv (bres, eres, _), FRange.Itv (bother, eother, _) -> let bres = Cmp.prev_float_ieee prec bres in let eres = Cmp.next_float_ieee prec eres in let round = if F.is_exact prec then Up else Near in let b = F.sub round prec bres eother in let round = if F.is_exact prec then Down else Near in let e = F.sub round prec eres bother in if Cmp.le b e then `Value (FRange.inject ~nan:false b e) else `Bottom in Bottom.join join reduced_for_finite_values reduced_for_nan let backward_add fkind ~left ~right ~result = backward_add_one fkind ~other:right ~result >>- fun left' -> backward_add_one fkind ~other:left ~result >>- fun right' -> `Value (left', right') let backward_sub fk ~left ~right ~result = let right = neg right in backward_add fk ~left ~right ~result >>-: fun (left, right) -> (left, neg right) (* ------------------------------------------------------------------------ Exp Log Sqrt Pow Fmod ------------------------------------------------------------------------ *) let (>>:) t f = match t with | FRange.NaN -> t | FRange.Itv (b, e, nan) -> f ~nan b e let approx f prec ~nan b e = let round = if F.is_exact prec then Near else Down in let min = f round prec b in let round = if F.is_exact prec then Near else Up in let max = f round prec e in FRange.inject ~nan min max let exp prec t = t >>: approx F.exp prec let log_aux log prec t = t >>: fun ~nan b e -> if Cmp.(lt e neg_zero) then FRange.nan else let nan = nan || Cmp.(lt b neg_zero) in let b = Cmp.max (Cst.neg_zero prec) b in approx log prec ~nan b e let log prec = log_aux F.log prec let log10 prec = log_aux F.log10 prec (* [sqrt_f] is the actual function computing the (exact) square root, in single precision (sqrtf) or double precision (sqrt). *) let sqrt prec t = t >>: fun ~nan b e -> if Cmp.(lt_ieee e neg_zero) then FRange.nan else let nan, b = if Cmp.(ge_ieee b neg_zero) then nan, b else true, Cst.neg_zero prec in approx F.sqrt prec ~nan b e let value_if prec condition f = if condition then `Value (f prec) else `Bottom (* Returns the minimal or maximal (according to [min_or_max]) results of the binary operation [op] applied to the bounds of the intervals [b1..e1] and [b2..e2]. *) let extremum min_or_max op (b1, e1) (b2, e2) = let extremum4 a b c d = min_or_max a (min_or_max b (min_or_max c d)) in extremum4 (op b1 b2) (op b1 e2) (op e1 b2) (op e1 e2) (* Returns the minimum and maximum results of the binary operation [op] applied to the bounds of the intervals [b1..e1] and [b2..e2]. *) let extrema op (b1, e1) (b2, e2) = let a = op b1 b2 and b = op b1 e2 and c = op e1 b2 and d = op e1 e2 in Cmp.min a (Cmp.min b (Cmp.min c d)), Cmp.max a (Cmp.max b (Cmp.max c d)) (* Computes [pow_f] on a negative [bx; ex] interval (including infinites). Processes by disjunction over even and odd integers enclosed within [by; ey]. [pow] is then monotonic on even integers (including zeros and infinities), and on odd integers (except on infinities). *) let pow_negative_x pow_f prec (bx, ex as x) (by, ey as y) = let even, odd = split_by_parity prec y in (* Even integers [y] lead to positive results, while odd ones lead to negative results. When [y] contains both even and odd integers, the minimum result is in odd integers, and the maximum in even integers. *) let min, max = match even, odd with | None, None -> Cst.pos_infinity prec, Cst.neg_infinity prec | Some even, None -> extrema pow_f x even | None, Some odd -> extrema pow_f x odd | Some even, Some odd -> extremum Cmp.min pow_f x odd, extremum Cmp.max pow_f x even in let nonint_y = contains_finite_noninteger prec y in (* pow creates NaN when [x] is a negative non-zero finite value, and [y] a non integer value. *) let nan = contains_strict_neg_finite x && nonint_y in (* Special cases of neg_infinity and neg_zero for [x], that do not produce a NaN on non integer [y], unlike strictly negative finite values [x]. *) let neg_nonint_y = nonint_y && Cmp.(lt by neg_zero) in let pos_nonint_y = nonint_y && Cmp.(gt ey pos_zero) in let neg_infinity_x = Cmp.is_neg_infinity bx in let zero_x = Cmp.(equal ex neg_zero) in Bottom.join_list join [ if Cmp.le min max then `Value (FRange.inject min max) else `Bottom; value_if prec nan (fun _ -> FRange.nan); value_if prec (neg_infinity_x && neg_nonint_y) pos_zero; value_if prec (neg_infinity_x && pos_nonint_y) pos_infinity; value_if prec (zero_x && neg_nonint_y) pos_infinity; value_if prec (zero_x && pos_nonint_y) pos_zero ] (* Computes pow on a positive [bx; ex] interval (including infinites): the function is continuous and monotonic. *) let pow_positive_x pow_f _prec x y = let min, max = extrema pow_f x y in FRange.inject min max let pow prec x y = if is_one x || is_a_zero y then one prec else (x, y) >>% fun ~nan itv_x itv_y -> let pow_f = F.pow Near prec in (* Split the x interval around zeros, as pow is discontinuous on zeros. *) let neg_x, pos_x = split_by_sign prec itv_x in let pos_x_res = pos_x >>-: fun x -> pow_positive_x pow_f prec x itv_y in let neg_x_res = neg_x >>- fun x -> pow_negative_x pow_f prec x itv_y in let nan_res = if nan then `Value FRange.nan else `Bottom in Bottom.non_bottom (Bottom.join_list join [ pos_x_res; neg_x_res; nan_res ]) (* Is [fmod] continuous on positive intervals [b1..e1] and [b2..e2]? This is the case if for all x, y in these intervals, the rounded quotient [floor(x/y)] is constant, as [fmod x y = x - floor(x/y) * y] when [x] and [y] are positive. Also checks that [x/y < 2^53], otherwise truncation to an integer may return an incorrect result. Note: to avoid issues with rounding, we conservatively set the limit to 2^51 instead of 2^53 (and to 2^21 instead of 2^23 in single precision). *) let is_fmod_continuous prec (b1, e1) (b2, e2) = (* Discontinuity of [fmod x y] on infinite [x] and on zero [y]. *) F.is_finite e1 && Cmp.(gt_ieee b2 pos_zero) && let four = F.of_float Near prec 4. in let max_i = F.div Near prec (Cst.max_precise_integer prec) four in let f1 = F.floor (F.div Zero prec b1 e2) in let f2 = F.floor (F.div Zero prec e1 b2) in Cmp.equal f1 f2 && Cmp.le f1 max_i && F.is_exact prec (* Forward semantics of fmod on positive intervals. [x] must contain finite values, and [y] must contain non-zero values, in which case finite values are produced. This function does not check the creation of NaN. *) let positive_fmod prec (b1, e1 as x) (b2, e2 as y) = let fmod = F.fmod Near prec in (* Singleton case. [x] cannot be infinite, and [y] cannot be zero. *) if Cmp.equal b1 e1 && Cmp.equal b2 e2 then let c = fmod b1 b2 in c, c (* If all values of x are smaller than all values of [y], [x] is unchanged. *) else if Cmp.lt_ieee e1 b2 then x (* If fmod is continuous on the intervals [x] and [y], it is also monotonic, and the bounds of the result are the remainders of the bounds. *) else if is_fmod_continuous prec x y then fmod b1 e2, fmod e1 b2 (* Otherwise, fmod always satisfies 0 <= [fmod x y] <= x and [fmod x y] < y. *) else let max = Cmp.min e1 (F.prev_float prec e2) in (Cst.pos_zero prec, max) let fmod prec x y = (* [fmod x (-y)] = [fmod x y], so use only positive [y]. *) let y = abs prec y in (x, y) >>% fun ~nan (b1, e1) (b2, e2) -> (* If [x] is an infinite singleton, or if [y] contains only zero, only NaN can be created. *) if (Cmp.equal b1 e1 && F.is_infinite b1) || (Cmp.equal b2 e2 && Cmp.is_a_zero b2) then FRange.nan else let nan = nan || contains_infinity x || contains_a_zero y in let positive_fmod x = let b, e = positive_fmod prec x (b2, e2) in FRange.inject ~nan b e in (* Process by disjunction on the sign of [x], and join the results for negative x and for positive x. *) let neg_x, pos_x = split_by_sign prec (b1, e1) in let neg_itv (b, e) = F.neg e, F.neg b in let neg_r = neg_x >>-: neg_itv >>-: positive_fmod >>-: neg in let pos_r = pos_x >>-: positive_fmod in Bottom.non_bottom (Bottom.join join neg_r pos_r) (* ------------------------------------------------------------------------ Trigonometry ------------------------------------------------------------------------ *) (* It is wrong to use m_pi as the local minimum as was previously done, because: A = 3.14159265358979312 and A < m_pi, cos A = 1. and cos m_pi = 0.999 Moreover it is not due to the imprecision in the value of pi: A < pred m_pi in simple. So we use a quarter of the interval [-pi:pi] to be safe. (But still, nothing proves that cos or sin are monotonic on those ranges.) *) let m_pi = 3.1415929794311523 (* single-precision *) let cos_sin_security = F.of_float Near Float_sig.Single (m_pi /. 4.) let cos_sin op prec t = t >>: fun ~nan b e -> (* Special case when at least a bound is infinite. *) if F.is_infinite b || F.is_infinite e then if Cmp.equal b e then FRange.nan else minus_one_one prec ~nan:true (* [b] and [e] are finite. Precise case for a singleton. *) else if Cmp.equal b e then let c = op Near prec b in FRange.inject ~nan c c else if Cmp.le_ieee b (F.neg cos_sin_security) || Cmp.ge_ieee e cos_sin_security then minus_one_one prec ~nan else let xl = if Cmp.(lt b pos_zero) && Cmp.(lt pos_zero e) then [b; e; Cst.pos_zero prec] else [b; e] in let l = List.map (op Near prec) xl in let min_f = List.fold_left Cmp.min (Cst.pos_infinity prec) l in let max_f = List.fold_left Cmp.max (Cst.neg_infinity prec) l in FRange.inject ~nan min_f max_f let cos prec = cos_sin F.cos prec let sin prec = cos_sin F.sin prec let atan2 prec x y = (x, y) >>% fun ~nan (b1, e1) (b2, e2) -> let op = F.atan2 Near prec in if Cmp.equal b1 e1 && Cmp.equal b2 e2 then let c = op b1 b2 in FRange.inject ~nan c c else (* Unless y ([b1,e1]) crosses the x-axis, atan2 is continuous, and its minimum/maximum are at the ends of the intervals of x and y. Otherwise, the result is [-pi,+pi]. *) if Cmp.(lt b1 pos_zero) && Cmp.(gt e1 neg_zero) then minus_pi_pi prec ~nan else let a1, a2, a3, a4 = op b1 b2, op b1 e2, op e1 b2, op e1 e2 in let b = Cmp.min a1 (Cmp.min a2 (Cmp.min a3 a4)) in let e = Cmp.max a1 (Cmp.max a2 (Cmp.max a3 a4)) in match prec with | Float_sig.Single -> (* Rounding of atan2f in single-precision may go against monotony and reach the next (previous) float after (before) the bounds. *) FRange.inject ~nan (Cmp.prev_float_ieee prec b) (Cmp.next_float_ieee prec e) | _ -> FRange.inject ~nan b e (* ------------------------------------------------------------------------ Casts ------------------------------------------------------------------------ *) let forward_cast ~dst = function | FRange.NaN -> nan | FRange.Itv (b, e, nan) -> let round = F.round_to_precision Near dst in inject ~nan (round b) (round e) (* This function must make sure to return a result with float 32 bounds *) let backward_cast ~src = function | FRange.NaN -> `Value nan | FRange.Itv (b, e, nan) -> FRange.inject_after_tighten src ~nan b e let cast_int_to_float prec min max = let min = match min with | None -> Cst.neg_infinity prec | Some v -> let round = if F.is_exact prec then Near else Down in F.of_float round prec (Integer.to_float v) in let max = match max with | None -> Cst.neg_infinity prec | Some v -> let round = if F.is_exact prec then Near else Up in F.of_float round prec (Integer.to_float v) in FRange.inject min max (* Bitwise reinterpretation of a floating-point value into consecutive ranges of integer. This operation is exact in terms of concretization. 'Parametric' in the number of bits. *) let bits_of_float_list ~prec ~bits_of_float ~max_int = let neg_infinity = Cst.neg_infinity prec and pos_infinity = Cst.pos_infinity prec in let itvs_nan = let smallest_neg_nan = Integer.succ (bits_of_float neg_infinity) in let biggest_neg_nan = Integer.minus_one in let smallest_pos_nan = Integer.succ (bits_of_float pos_infinity) in let biggest_pos_nan = max_int in [(smallest_neg_nan, biggest_neg_nan); (smallest_pos_nan, biggest_pos_nan)] in function | FRange.NaN -> itvs_nan | FRange.Itv (b, e, nan) -> let nan = if nan then itvs_nan else [] in let neg, pos = split_by_sign prec (b, e) in let neg = neg >>-: fun (b, e) -> bits_of_float e, bits_of_float b in let pos = pos >>-: fun (b, e) -> bits_of_float b, bits_of_float e in Bottom.add_to_list pos (Bottom.add_to_list neg nan) let bits_of_float64_list = let bits_of_float f = Integer.of_int64 (Int64.bits_of_float (F.to_float f)) in let max_int = Integer.of_int64 Int64.max_int in bits_of_float_list ~prec:Double ~bits_of_float ~max_int let bits_of_float32_list = let bits_of_float f = Integer.of_int32 (Int32.bits_of_float (F.to_float f)) in let max_int = Integer.of_int32 Int32.max_int in bits_of_float_list ~prec:Single ~bits_of_float ~max_int (* ------------------------------------------------------------------------ Subdivision ------------------------------------------------------------------------ *) (* [avg] and [split] implement two different strategies for cutting a floating-point interval in half: [avg] computes the mathematical average of the two bounds, while [split] balances the number of representable values of the given precision in each resulting intervals. *) (* Computes the average between two ocaml doubles. *) let avg x y = let fx = F.to_float x and fy = F.to_float y in if F.is_negative x && F.is_negative y then fy +. (fx -. fy) /. 2. else (fx +. fy) /. 2. (* assumption: [0. <= x <= y]. returns the median of the range [x..y] in number of double values. *) let split_positive prec x y = let ix = Int64.bits_of_float (F.to_float x) in let iy = Int64.bits_of_float (F.to_float y) in let f = Int64.(float_of_bits (add ix (div (sub iy ix) 2L))) in F.of_float Near prec f (* assumption: [x <= y] *) let _split prec x y = match F.is_negative x, F.is_negative y with | false, false -> split_positive prec x y | true, true -> F.neg (split_positive prec (F.neg x) (F.neg y)) | true, false -> Cst.neg_zero prec | false, true -> assert false exception Can_not_subdiv = Abstract_interp.Can_not_subdiv let subdivide prec t = assert (prec = Single || prec = Double); (* See Value/Value#105 *) match t with | FRange.NaN -> raise Can_not_subdiv | FRange.Itv (b, e, nan) -> if nan then FRange.inject ~nan:false b e, FRange.nan else if Cmp.equal b e then raise Can_not_subdiv else let midpoint, smidpoint = (* Infinities are interesting points to consider specially. *) if Cmp.is_neg_infinity b then Cst.neg_infinity prec, F.next_float prec (Cst.neg_infinity prec) else if Cmp.is_pos_infinity e then F.prev_float prec (Cst.pos_infinity prec), Cst.pos_infinity prec else if Cmp.equal (F.next_float prec b) e then b, e else let midpoint = avg b e in let midpoint = F.of_float Down prec midpoint in let smidpoint = if F.is_exact prec then F.next_float prec midpoint else midpoint in if Cmp.le smidpoint e then midpoint, smidpoint else midpoint, e in if Cmp.le smidpoint b || Cmp.le e midpoint then raise Can_not_subdiv; let i1 = FRange.inject ~nan b midpoint in assert (is_included i1 t); let i2 = FRange.inject ~nan smidpoint e in assert (is_included i2 t); i1, i2 end
sectionYPositions = computeSectionYPositions($el), 10)"
x-init="setTimeout(() => sectionYPositions = computeSectionYPositions($el), 10)"
>