/usr/include/trilinos/Kokkos_ArithTraits.hpp is in libtrilinos-kokkos-kernels-dev 12.12.1-5.
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 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 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 | /*
//@HEADER
// ************************************************************************
//
// KokkosKernels 0.9: Linear Algebra and Graph Kernels
// Copyright 2017 Sandia Corporation
//
// Under the terms of Contract DE-AC04-94AL85000 with Sandia Corporation,
// the U.S. Government retains certain rights in this software.
//
// Redistribution and use in source and binary forms, with or without
// modification, are permitted provided that the following conditions are
// met:
//
// 1. Redistributions of source code must retain the above copyright
// notice, this list of conditions and the following disclaimer.
//
// 2. Redistributions in binary form must reproduce the above copyright
// notice, this list of conditions and the following disclaimer in the
// documentation and/or other materials provided with the distribution.
//
// 3. Neither the name of the Corporation nor the names of the
// contributors may be used to endorse or promote products derived from
// this software without specific prior written permission.
//
// THIS SOFTWARE IS PROVIDED BY SANDIA CORPORATION "AS IS" AND ANY
// EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
// IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
// PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL SANDIA CORPORATION OR THE
// CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
// EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
// PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
// PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
// LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
// NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
// SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
//
// Questions? Contact Siva Rajamanickam (srajama@sandia.gov)
//
// ************************************************************************
//@HEADER
*/
#ifndef KOKKOS_ARITHTRAITS_HPP
#define KOKKOS_ARITHTRAITS_HPP
/// \file Kokkos_ArithTraits.hpp
/// \brief Declaration and definition of Kokkos::Details::ArithTraits
#include <KokkosKernels_config.h>
#include <Kokkos_Complex.hpp>
#ifdef HAVE_KOKKOSKERNELS_QUADMATH
# include <quadmath.h>
#endif // HAVE_KOKKOSKERNELS_QUADMATH
#include <cfloat>
#include <climits>
#include <cmath>
#include <complex> // std::complex
#include <limits> // std::numeric_limits
#ifdef __CUDACC__
# include <math_constants.h>
#endif
//
// mfh 24 Dec 2013: Temporary measure for testing; will go away.
//
#ifndef KOKKOS_FORCEINLINE_FUNCTION
# ifdef __CUDA_ARCH__
# define KOKKOS_FORCEINLINE_FUNCTION inline __host__ __device__
# else
# define KOKKOS_FORCEINLINE_FUNCTION
# endif // __CUDA_ARCH__
#endif // KOKKOS_FORCEINLINE_FUNCTION
namespace { // anonymous
/// \fn intPowImpl
/// \tparam IntType A built-in integer type.
/// \brief Implementation of intPowSigned and intPowUnsigned.
///
/// \pre x != 0
/// \pre y > 0
///
/// Use intPowSigned or intPowUnsigned for general y.
template<class IntType>
KOKKOS_FORCEINLINE_FUNCTION IntType
intPowImpl (const IntType x, const IntType y)
{
// Recursion (unrolled into while loop): pow(x, 2y) = (x^y)^2
IntType prod = x;
IntType y_cur = 1;
// If y == 1, then prod stays x.
while (y_cur < y) {
prod = prod * prod;
y_cur = y_cur << 1;
}
// abs(y - y_cur) < floor(log2(y)), so it won't hurt asymptotic run
// time to finish the remainder in a linear iteration.
if (y > y_cur) {
const IntType left = y - y_cur;
for (IntType k = 0; k < left; ++k) {
prod = prod * x;
}
}
else if (y < y_cur) {
// There's probably a better way to do this in order to avoid the
// (expensive) integer division, but I'm not motivated to think of
// it at the moment.
const IntType left = y_cur - y;
for (IntType k = 0; k < left; ++k) {
prod = prod / x;
}
}
return prod;
// y = 8:
//
// x,1 -> x^2,2
// x^2,2 -> x^4,4
// x^4,4 -> x^8,8
//
// y = 9:
//
// x,1 -> x^2,2
// x^2,2 -> x^4,4
// x^4,4 -> x^8,8
//
// y - y_cur is what's left over. Just do it one at a time.
//
// y = 3:
// x,1 -> x^2,2
// x^2,2 -> x^4,4
}
/// \fn intPowSigned
/// \tparam IntType A built-in signed integer type.
/// \brief Compute x raised to the power y.
///
/// If the arguments are invalid (e.g., if x and y are both zero), the
/// result of this function is undefined. However, this function will
/// not throw an exception in that case.
template<class IntType>
KOKKOS_FORCEINLINE_FUNCTION IntType
intPowSigned (const IntType x, const IntType y)
{
// It's not entirely clear what to return if x and y are both zero.
// In the case of floating-point numbers, 0^0 is NaN. Here, though,
// I think it's safe to return 0.
if (x == 0) {
return 0;
} else if (y == 0) {
return 1;
} else if (y < 0) {
if (x == 1) {
return 1;
}
else if (x == -1) {
return (y % 2 == 0) ? 1 : -1;
}
else {
return 0; // round the fraction to zero
}
} else {
return intPowImpl<IntType> (x, y);
}
}
/// \fn intPowUnsigned
/// \tparam IntType A built-in unsigned integer type.
/// \brief Compute x raised to the power y.
///
/// If the arguments are invalid (e.g., if x and y are both zero), the
/// result of this function is undefined. However, this function will
/// not throw an exception in that case.
template<class IntType>
KOKKOS_FORCEINLINE_FUNCTION IntType
intPowUnsigned (const IntType x, const IntType y)
{
// It's not entirely clear what to return if x and y are both zero.
// In the case of floating-point numbers, 0^0 is NaN. Here, though,
// I think it's safe to return 0.
if (x == 0) {
return 0;
} else if (y == 0) {
return 1;
} else {
return intPowImpl<IntType> (x, y);
}
}
// It might make sense to use special sqrt() approximations for
// integer arguments, like those presented on the following web site:
//
// http://www.azillionmonkeys.com/qed/sqroot.html#implementations
//
// Note that some of the implementations on the above page break ANSI
// C(++) aliasing rules (by assigning to the results of
// reinterpret_cast-ing between int and float). It's also just a
// performance optimization and not required for a reasonable
// implementation.
} // namespace (anonymous)
namespace Kokkos {
namespace Details {
/// \class ArithTraits
/// \brief Traits class for arithmetic on type T.
/// \tparam T "Scalar" type of interest
///
/// This is a traits class for the "arithmetic" type T. "Arithmetic
/// types" include built-in signed and unsigned integer types,
/// floating-point types, complex-valued types, and anything else that
/// looks like these. This class is useful for implementing numerical
/// algorithms that are generic on the data type. You may also use
/// this class to query attributes of T, like whether it is signed or
/// complex, or its precision.
///
/// We really did not want to implement this class or expose it to
/// users. It would be much better to use existing traits classes
/// like std::numeric_limits. We decided to implement and expose this
/// class for the following reasons:
/// <ol>
/// <li> std::numeric_limits class methods cannot be used in CUDA
/// device functions, since they themselves are not device
/// functions </li>
/// <li> Existing traits classes like std::numeric_limits do not
/// provide enough information to implement algorithms that are
/// agnostic of whether T is real-valued or complex-valued. </li>
/// </ol>
///
/// All class methods must be suitable for parallel kernels, if the
/// type T itself is suitable for parallel kernels. In particular,
/// specializations for types T that make sense to use on a CUDA
/// device must mark all class methods as device (and host) functions,
/// using the KOKKOS_FORCEINLINE_FUNCTION macro. All class methods must be
/// callable both inside and outside a parallel kernel (for CUDA, this
/// means they must be marked as both device and host functions).
///
/// \section Kokkos_ArithTraits_compat Compatibility
///
/// Whenever possible, class methods in ArithTraits use the same names
/// as their equivalents in the C++ Standard Library. If this was not
/// possible, for example with isInf and isNan, we explain why in
/// their documentation.
///
/// This class has redundant typedefs and methods in order to maintain
/// backwards compatibility with Teuchos::ScalarTraits, while
/// preferring forwards (partial) compatibility with
/// std::numeric_limits. Users should prefer typedefs, \c bool
/// constants, and class methods compatible with std::numeric_limits,
/// to those from Teuchos::ScalarTraits. The latter may go away at
/// any time. Furthermore, Teuchos::ScalarTraits contains methods
/// that do not make sense for use as parallel device functions, in
/// particular those relating to pseudorandom number generation that
/// refer to hidden state, so we will never include all class methods
/// from Teuchos::ScalarTraits in ArithTraits.
///
/// \section Kokkos_ArithTraits_unsupp Unsupported types on CUDA devices
///
/// CUDA does not support long double or std::complex<T> in device
/// functions. ArithTraits does have specializations for these types,
/// but the class methods therein are not marked as device functions.
///
/// \section Kokkos_ArithTraits_whyNotC99 What about C99 integer types?
///
/// C99 and C++11 include typedefs int${N}_t and uint${N}_t, where N
/// is the number of bits in the integer. These typedefs are useful
/// because they make the length of the type explicit. Users are
/// welcome to use these types as the template parameter of
/// ArithTraits.
///
/// We chose not to use these types when <i>defining</i> full
/// specializations of ArithTraits. This is because the C99 integer
/// types are typedefs, not types in themselves. This makes it
/// impossible to avoid duplicate or missing full specializations of
/// ArithTraits. For example, on my Mac, for CUDA 5.5, gcc 4.2.1, and
/// Clang 3.2, <tt>int64_t</tt> is a typedef of <tt>long long</tt>,
/// but <tt>long long</tt> and <tt>long</tt> are separate types, even
/// though they have the same length (64 bits). In contrast, on
/// Windows (even Win64), <tt>long</tt> is a 32-bit type (but a
/// distinct type from <tt>int</tt>), and <tt>long long</tt> is a
/// 64-bit type. Thus, if we define full specializations of
/// ArithTraits using <i>only</i> the C99 integer types, we will be
/// missing a specialization for <tt>long</tt> on at least one
/// platform.
///
/// Rather than trouble ourselves with trying to figure this out for
/// each platform, we decided to provide specializations only for the
/// integer types in the C89 and C++03 language standards. This
/// includes signed and unsigned versions of <tt>char</tt>,
/// <tt>short</tt>, <tt>int</tt>, and <tt>long</tt>. We also include
/// <tt>long long</tt> if your platform supports it. We may thus have
/// left out some C99 integer type, but this is only possible if the
/// C89 / C++03 integer types do not have complete coverage of all
/// powers of two bits from 8 up to the longest provided length (e.g.,
/// 64 on a 64-bit system). On all platforms I have encountered,
/// <tt>char</tt> has 8 bits and <tt>short</tt> has 16 bits, so I am
/// not worried about missing specializations for <tt>int16_t</tt> or
/// <tt>uint16_t</tt>. If you should find that either of these
/// specializations are missing, though, please let us know.
///
/// Note that <tt>char</tt>, <tt>signed char</tt>, and <tt>unsigned
/// char</tt> are distinct types, whether <tt>char</tt> is signed or
/// unsigned. (The language standards do not specify whether
/// <tt>char</tt> is signed or unsigned.) That is, <tt>char</tt> is
/// <i>not</i> a typedef of <tt>signed char</tt> or <tt>unsigned
/// char</tt>. This is why we provide full specializations of
/// ArithTraits for each of these types. Interestingly enough, on my
/// system, <tt>char</tt> and <tt>int8_t</tt> are different types, but
/// <tt>signed char</tt> and <tt>int8_t</tt> are the same.
///
/// \section Kokkos_ArithTraits_impl Implementation notes
///
/// This section contains notes to developers who which to add a
/// partial specialization of this class for a new type T. If you
/// decide to write a default templated implementation, it must not
/// declare any methods as device functions. This ensures correct
/// behavior for arbitrary T, but does require specializations for
/// common types like T = float and double, as well as for other types
/// T that make sense to use on a CUDA device.
template<class T>
class ArithTraits {
public:
/// \brief A type that acts like T and works with Kokkos.
///
/// This is usually just an alias for T. However, some types T do
/// not work well with Kokkos. In that case, we use a mostly
/// equivalent type here. For example, ArithTraits<std::complex<R>
/// >::val_type is Kokkos::complex<R>.
typedef T val_type;
/// \brief The type of the magnitude (absolute value) of T.
///
/// We define this as the type returned by abs() in this class. If
/// T is real (not complex), then \c val_type and \c mag_type are
/// usually the same. If T is <tt>std::complex<R></tt> for some R,
/// then R and \c mag_type are usually the same.
typedef T mag_type;
//! Whether ArithTraits has a specialization for T.
static const bool is_specialized = false;
//! Whether T is a signed type (has negative values).
static const bool is_signed = false;
//! Whether T is an integer type.
static const bool is_integer = false;
/// \brief Whether T "uses exact representations."
///
/// The opposite of is_exact is "is approximate," that is, "may
/// commit rounding error."
static const bool is_exact = false;
//! Whether T is a complex-valued type.
static const bool is_complex = false;
/// \brief Whether x is Inf.
///
/// This can only be true for floating-point types T that support
/// Inf. If T is a complex type, we say that a T instance x is Inf
/// if and only if <tt>isinf(real(x)) || isinf(imag(x))</tt>.
///
/// Unfortunately we can't call this "isinf" (the equivalent C99
/// function), because CUDA appears to implement that function using
/// a macro, rather than using a function (as C++11 requires).
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const T& x);
/// \brief Whether x is NaN (not a number).
///
/// This can only be true for floating-point types T that support
/// NaN. If T is a complex type, we say that a T instance x is NaN
/// if and only if <tt>isNan(real(x)) || isNan(imag(x))</tt>.
///
/// Unfortunately we can't call this "isnan" (the equivalent C99
/// function), because CUDA appears to implement that function using
/// a macro, rather than using a function (as C++11 requires).
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const T& x);
//! The absolute value (magnitude) of x.
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const T& x);
//! The zero value of T; the arithmetic identity.
static KOKKOS_FORCEINLINE_FUNCTION T zero ();
//! The one value of T; the multiplicative identity.
static KOKKOS_FORCEINLINE_FUNCTION T one ();
/// \brief The minimum possible value of T.
///
/// If T is a real floating-point type, then this is the minimum
/// <i>positive</i> value, as with std::numeric_limits<T>::min().
static KOKKOS_FORCEINLINE_FUNCTION T min ();
//! The maximum possible value of T.
static KOKKOS_FORCEINLINE_FUNCTION T max ();
/// \brief The real part of x.
///
/// If \c is_complex is false, then this just returns x.
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const T& x);
/// \brief The imaginary part of x.
///
/// If \c is_complex is false, then this just returns zero().
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const T&);
/// \brief The complex conjugate of x.
///
/// If \c is_complex is false, then this just returns x.
static KOKKOS_FORCEINLINE_FUNCTION T conj (const T&);
//! x raised to the power y.
static KOKKOS_FORCEINLINE_FUNCTION T pow (const T& x, const T& y);
/// \brief The square root of x.
///
/// If T is an integer type, this is the floor of the square root.
/// If T is a complex-valued type, then this method returns the
/// principal branch of the square root.
///
/// If T is real-valued and x is negative, the result of the square
/// root is undefined in general. (CUDA does not allow throwing
/// exceptions in device functions.) Implementations should return
/// NaN if the type T supports this. Of course, in that case, the
/// square of the result will not equal x.
static KOKKOS_FORCEINLINE_FUNCTION T sqrt (const T& x);
/// \brief The natural (base e) logarithm of x.
///
/// If T is an integer type, this is the floor of the logarithm. If
/// T is a complex-valued type, then this method returns the
/// principal branch of the logarithm.
///
/// If T is real-valued and x is negative, the result of the
/// logarithm is undefined in general. (CUDA does not allow
/// throwing exceptions in device functions.) Implementations
/// should return NaN if the type T supports this. Of course, in
/// that case, if y is the result, \f$e^y\f$ will not equal x.
static KOKKOS_FORCEINLINE_FUNCTION T log (const T& x);
/// \brief The base ten logarithm of the input.
///
/// If T is an integer type, this is the floor of the logarithm. If
/// T is a complex-valued type, then this method returns the
/// principal branch of the logarithm.
///
/// If T is real-valued and x is negative, the result of the
/// logarithm is undefined in general. (CUDA does not allow
/// throwing exceptions in device functions.) Implementations
/// should return NaN if the type T supports this. Of course, in
/// that case, if y is the result, \f$10^y\f$ will not equal x.
static KOKKOS_FORCEINLINE_FUNCTION T log10 (const T& x);
/// \brief Return a silent NaN, if appropriate for T.
///
/// If T does <i>not</i> implement a silent NaN, the return value is
/// undefined, but calling this method is still allowed.
static KOKKOS_FORCEINLINE_FUNCTION T nan ();
/// \brief Machine epsilon.
///
/// If T is an integer type (std::numeric_traits<T>::is_exact is
/// true), then epsilon() returns 0. Otherwise, if T is a
/// floating-point type, it returns machine epsilon that T.
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon ();
//@{
/// \name Traits defined for backwards compatibility with Teuchos::ScalarTraits
///
/// All of the typedefs, \c bool constants, and class methods in
/// this section are defined in order that one may replace most uses
/// of Teuchos::ScalarTraits with ArithTraits. Users who do not
/// have this backwards compatibility requirement should prefer
/// equivalents in other sections. Those class methods which have
/// the same name and meaning in both Teuchos::ScalarTraits and this
/// class, such as log() and pow(), are not in this section.
//! Same as mag_type; the type of the absolute value (magnitude) of T.
typedef T magnitudeType;
/// \brief The type with "half the precision" of T.
///
/// This typedef only makes sense if T is a floating-point type.
typedef T halfPrecision;
/// \brief The type with "twice the the precision" of T.
///
/// This typedef only makes sense if T is a floating-point type.
typedef T doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = false;
/// \brief True if this type T has floating-point parameters.
///
/// This is true if and only if this specialization of ArithTraits
/// has "machine-specific" parameters eps(), sfmin(), base(),
/// prec(), t(), rnd(), emin(), rmin(), emax(), and rmax(), relating
/// to floating-point types.
static const bool hasMachineParameters = false;
//! Return relative machine precision.
static KOKKOS_FORCEINLINE_FUNCTION mag_type eps ();
//! Return safe minimum (sfmin), such that 1/sfmin does not overflow.
static KOKKOS_FORCEINLINE_FUNCTION mag_type sfmin ();
//! Return the base of the scalar type T.
static KOKKOS_FORCEINLINE_FUNCTION int base ();
//! Return <tt>eps*base</tt>.
static KOKKOS_FORCEINLINE_FUNCTION mag_type prec ();
//! Returns the number of (base) digits in the significand.
static KOKKOS_FORCEINLINE_FUNCTION int t ();
//! 1.0 when rounding occurs in addition, else 0.0.
static KOKKOS_FORCEINLINE_FUNCTION mag_type rnd ();
//! Returns the minimum exponent before (gradual) underflow.
static KOKKOS_FORCEINLINE_FUNCTION int emin ();
//! Returns the underflow threshold: <tt>base^(emin-1)</tt>
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmin ();
//! Returns the largest exponent before overflow.
static KOKKOS_FORCEINLINE_FUNCTION int emax ();
//! Overflow theshold: <tt>(base^emax)*(1-eps)</tt>
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmax ();
//! Same as abs(); return the magnitude of x.
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const T& x);
//! Same as conj(); return the complex conjugate of x.
static KOKKOS_FORCEINLINE_FUNCTION T conjugate (const T& x);
/// \brief Whether x is (silent) NaN or Inf.
///
/// This is the same as <tt>isNan(x) || isInf(x)</tt>.
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const T& x);
/// \brief The string name of T.
///
/// Note that this is not a device function.
static std::string name ();
//! Same as sqrt(x); the square root of x.
static KOKKOS_FORCEINLINE_FUNCTION T squareroot (const T& x);
//@}
};
template<>
class ArithTraits<float> {
public:
typedef float val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const float x) {
#ifndef __CUDA_ARCH__
using std::isinf;
#endif
return isinf (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const float x) {
#ifndef __CUDA_ARCH__
using std::isnan;
#endif
return isnan (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const float x) {
return ::fabs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION float zero () {
return 0.0;
}
static KOKKOS_FORCEINLINE_FUNCTION float one () {
return 1.0;
}
static KOKKOS_FORCEINLINE_FUNCTION float min () {
return FLT_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION float max () {
return FLT_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const float x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const float) {
return 0.0;
}
static KOKKOS_FORCEINLINE_FUNCTION float conj (const float x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION float pow (const float x, const float y) {
return ::pow (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION float sqrt (const float x) {
return ::sqrt (x);
}
static KOKKOS_FORCEINLINE_FUNCTION float log (const float x) {
return ::log (x);
}
static KOKKOS_FORCEINLINE_FUNCTION float log10 (const float x) {
return ::log10 (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return FLT_EPSILON;
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
// C++ doesn't have a standard "half-float" type.
typedef float halfPrecision;
typedef double doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = true;
static const bool hasMachineParameters = true;
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const float x) {
return isNan (x) || isInf (x);
}
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const float x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION float conjugate (const float x) {
return conj (x);
}
static std::string name () {
return "float";
}
static KOKKOS_FORCEINLINE_FUNCTION float squareroot (const float x) {
return sqrt (x);
}
static KOKKOS_FORCEINLINE_FUNCTION float nan () {
#ifdef __CUDA_ARCH__
return CUDART_NAN_F;
//return nan (); //this returns 0???
#else
return std::numeric_limits<float>::quiet_NaN();
#endif // __CUDA_ARCH__
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type eps () {
return epsilon ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type sfmin () {
return FLT_MIN; // ???
}
static KOKKOS_FORCEINLINE_FUNCTION int base () {
return FLT_RADIX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type prec () {
return eps () * static_cast<mag_type> (base ());
}
static KOKKOS_FORCEINLINE_FUNCTION int t () {
return FLT_MANT_DIG;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rnd () {
return 1.0;
}
static KOKKOS_FORCEINLINE_FUNCTION int emin () {
return FLT_MIN_EXP;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmin () {
return FLT_MIN; // ??? // should be base^(emin-1)
}
static KOKKOS_FORCEINLINE_FUNCTION int emax () {
return FLT_MAX_EXP;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmax () {
return FLT_MAX; // ??? // should be (base^emax)*(1-eps)
}
};
/// \brief Partial specialization for std::complex<RealFloatType>.
///
/// The C++ Standard Library (with C++03 at least) only allows
/// std::complex<RealFloatType> for RealFloatType = float, double, or
/// long double.
template<class RealFloatType>
class ArithTraits<std::complex<RealFloatType> > {
public:
//! Kokkos internally replaces std::complex with Kokkos::complex.
typedef ::Kokkos::complex<RealFloatType> val_type;
typedef RealFloatType mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = true;
static bool isInf (const std::complex<RealFloatType>& x) {
#ifndef __CUDA_ARCH__
using std::isinf;
#endif
return isinf (real (x)) || isinf (imag (x));
}
static bool isNan (const std::complex<RealFloatType>& x) {
#ifndef __CUDA_ARCH__
using std::isnan;
#endif
return isnan (real (x)) || isnan (imag (x));
}
static mag_type abs (const std::complex<RealFloatType>& x) {
return std::abs (x);
}
static std::complex<RealFloatType> zero () {
return std::complex<RealFloatType> (ArithTraits<mag_type>::zero (), ArithTraits<mag_type>::zero ());
}
static std::complex<RealFloatType> one () {
return std::complex<RealFloatType> (ArithTraits<mag_type>::one (), ArithTraits<mag_type>::zero ());
}
static std::complex<RealFloatType> min () {
return std::complex<RealFloatType> (ArithTraits<mag_type>::min (), ArithTraits<mag_type>::zero ());
}
static std::complex<RealFloatType> max () {
return std::complex<RealFloatType> (ArithTraits<mag_type>::max (), ArithTraits<mag_type>::zero ());
}
static mag_type real (const std::complex<RealFloatType>& x) {
return std::real (x);
}
static mag_type imag (const std::complex<RealFloatType>& x) {
return std::imag (x);
}
static std::complex<RealFloatType> conj (const std::complex<RealFloatType>& x) {
return std::conj (x);
}
static std::complex<RealFloatType>
pow (const std::complex<RealFloatType>& x, const std::complex<RealFloatType>& y) {
// Fix for some weird gcc 4.2.1 inaccuracy.
if (y == one ()) {
return x;
} else if (y == one () + one ()) {
return x * x;
} else {
return std::pow (x, y);
}
}
static std::complex<RealFloatType> sqrt (const std::complex<RealFloatType>& x) {
return std::sqrt (x);
}
static std::complex<RealFloatType> log (const std::complex<RealFloatType>& x) {
return std::log (x);
}
static std::complex<RealFloatType> log10 (const std::complex<RealFloatType>& x) {
return std::log10 (x);
}
static std::complex<RealFloatType> nan () {
const mag_type mag_nan = ArithTraits<mag_type>::nan ();
return std::complex<RealFloatType> (mag_nan, mag_nan);
}
static mag_type epsilon () {
return ArithTraits<mag_type>::epsilon ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef std::complex<typename ArithTraits<mag_type>::halfPrecision> halfPrecision;
typedef std::complex<typename ArithTraits<mag_type>::doublePrecision> doublePrecision;
static const bool isComplex = true;
static const bool isOrdinal = false;
static const bool isComparable = false;
static const bool hasMachineParameters = true;
static bool isnaninf (const std::complex<RealFloatType>& x) {
return isNan (x) || isInf (x);
}
static mag_type magnitude (const std::complex<RealFloatType>& x) {
return abs (x);
}
static std::complex<RealFloatType> conjugate (const std::complex<RealFloatType>& x) {
return conj (x);
}
static std::string name () {
return std::string ("std::complex<") + ArithTraits<mag_type>::name () + ">";
}
static std::complex<RealFloatType> squareroot (const std::complex<RealFloatType>& x) {
return sqrt (x);
}
static mag_type eps () {
return epsilon ();
}
static mag_type sfmin () {
return ArithTraits<mag_type>::sfmin ();
}
static int base () {
return ArithTraits<mag_type>::base ();
}
static mag_type prec () {
return ArithTraits<mag_type>::prec ();
}
static int t () {
return ArithTraits<mag_type>::t ();
}
static mag_type rnd () {
return ArithTraits<mag_type>::one ();
}
static int emin () {
return ArithTraits<mag_type>::emin ();
}
static mag_type rmin () {
return ArithTraits<mag_type>::rmin ();
}
static int emax () {
return ArithTraits<mag_type>::emax ();
}
static mag_type rmax () {
return ArithTraits<mag_type>::rmax ();
}
};
template<>
class ArithTraits<double> {
public:
typedef double val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type x) {
#ifndef __CUDA_ARCH__
using std::isinf;
#endif
return isinf (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type x) {
#ifndef __CUDA_ARCH__
using std::isnan;
#endif
return isnan (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return ::fabs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0.0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1.0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return DBL_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return DBL_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type) {
return 0.0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type pow (const val_type x, const val_type y) {
return ::pow (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
return ::sqrt (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return ::log (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return ::log10 (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
#ifdef __CUDA_ARCH__
return CUDART_NAN;
//return nan (); // this returns 0 ???
#else
return std::numeric_limits<val_type>::quiet_NaN();
#endif // __CUDA_ARCH__
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return DBL_EPSILON;
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef float halfPrecision;
#ifdef __CUDA_ARCH__
typedef double doublePrecision; // CUDA doesn't support long double, unfortunately
#else
typedef long double doublePrecision;
#endif // __CUDA_ARCH__
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = true;
static const bool hasMachineParameters = true;
static bool isnaninf (const val_type& x) {
return isNan (x) || isInf (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static std::string name () {
return "double";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type eps () {
return epsilon ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type sfmin () {
return DBL_MIN; // ???
}
static KOKKOS_FORCEINLINE_FUNCTION int base () {
return FLT_RADIX; // same for float as for double
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type prec () {
return eps () * static_cast<mag_type> (base ());
}
static KOKKOS_FORCEINLINE_FUNCTION int t () {
return DBL_MANT_DIG;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rnd () {
return 1.0;
}
static KOKKOS_FORCEINLINE_FUNCTION int emin () {
return DBL_MIN_EXP;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmin () {
return DBL_MIN; // ??? // should be base^(emin-1)
}
static KOKKOS_FORCEINLINE_FUNCTION int emax () {
return DBL_MAX_EXP;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmax () {
return DBL_MAX; // ??? // should be (base^emax)*(1-eps)
}
};
// CUDA does not support long double in device functions, so none of
// the class methods in this specialization are marked as device
// functions.
template<>
class ArithTraits<long double> {
public:
typedef long double val_type;
typedef long double mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = false;
static bool isInf (const val_type& x) {
#ifndef __CUDA_ARCH__
using std::isinf;
#endif
return isinf (x);
}
static bool isNan (const val_type& x) {
#ifndef __CUDA_ARCH__
using std::isnan;
#endif
return isnan (x);
}
static mag_type abs (const val_type& x) {
return ::fabsl (x);
}
static val_type zero () {
return 0.0;
}
static val_type one () {
return 1.0;
}
static val_type min () {
return LDBL_MIN;
}
static val_type max () {
return LDBL_MAX;
}
static mag_type real (const val_type& x) {
return x;
}
static mag_type imag (const val_type&) {
return zero ();
}
static val_type conj (const val_type& x) {
return x;
}
static val_type pow (const val_type& x, const val_type& y) {
return ::pow (x, y);
}
static val_type sqrt (const val_type& x) {
return ::sqrt (x);
}
static val_type log (const val_type& x) {
return ::log (x);
}
static val_type log10 (const val_type& x) {
return ::log10 (x);
}
static val_type nan () {
return std::numeric_limits<val_type>::quiet_NaN();
}
static mag_type epsilon () {
return LDBL_EPSILON;
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef double halfPrecision;
// It might be appropriate to use QD's qd_real here.
// For now, long double is the most you get.
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = true;
static const bool hasMachineParameters = true;
static bool isnaninf (const val_type& x) {
return isNan (x) || isInf (x);
}
static mag_type magnitude (const val_type& x) {
return abs (x);
}
static val_type conjugate (const val_type& x) {
return conj (x);
}
static std::string name () {
return "long double";
}
static val_type squareroot (const val_type& x) {
return sqrt (x);
}
static mag_type eps () {
return epsilon ();
}
static mag_type sfmin () {
return LDBL_MIN; // ???
}
static int base () {
return FLT_RADIX; // same for float as for double or long double
}
static mag_type prec () {
return eps () * static_cast<mag_type> (base ());
}
static int t () {
return LDBL_MANT_DIG;
}
static mag_type rnd () {
return one ();
}
static int emin () {
return LDBL_MIN_EXP;
}
static mag_type rmin () {
return LDBL_MIN;
}
static int emax () {
return LDBL_MAX_EXP;
}
static mag_type rmax () {
return LDBL_MAX;
}
};
#ifdef HAVE_KOKKOSKERNELS_QUADMATH
// CUDA does not support __float128 in device functions, so none of
// the class methods in this specialization are marked as device
// functions.
template<>
class ArithTraits<__float128> {
public:
typedef __float128 val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = false;
static bool isInf (const __float128 x) {
return isinfq (x);
}
static bool isNan (const __float128 x) {
return isnanq (x);
}
static mag_type abs (const __float128 x) {
return fabsq (x);
}
static __float128 zero () {
return 0.0;
}
static __float128 one () {
return 1.0;
}
static __float128 min () {
return FLT128_MIN;
}
static __float128 max () {
return FLT128_MAX;
}
static mag_type real (const __float128 x) {
return x;
}
static mag_type imag (const __float128 /* x */) {
return 0.0;
}
static __float128 conj (const __float128 x) {
return x;
}
static __float128 pow (const __float128 x, const __float128 y) {
return powq (x, y);
}
static __float128 sqrt (const __float128 x) {
return sqrtq (x);
}
static __float128 log (const __float128 x) {
return logq (x);
}
static __float128 log10 (const __float128 x) {
return log10q (x);
}
static mag_type epsilon () {
return FLT128_EPSILON;
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef double halfPrecision;
// Unfortunately, we can't rely on a standard __float256 type.
typedef __float128 doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = true;
static const bool hasMachineParameters = true;
static bool isnaninf (const __float128 x) {
return isNan (x) || isInf (x);
}
static magnitudeType magnitude (const __float128 x) {
return abs (x);
}
static __float128 conjugate (const __float128 x) {
return conj (x);
}
static std::string name () {
return "__float128";
}
static __float128 squareroot (const __float128 x) {
return sqrt (x);
}
static __float128 nan () {
return strtoflt128 ("NAN()", NULL); // ???
}
static mag_type eps () {
return epsilon ();
}
static mag_type sfmin () {
return FLT128_MIN; // ???
}
static int base () {
return 2;
}
static mag_type prec () {
return eps () * static_cast<mag_type> (base ());
}
static int t () {
return FLT_MANT_DIG;
}
static mag_type rnd () {
return 1.0;
}
static int emin () {
return FLT128_MIN_EXP;
}
static mag_type rmin () {
return FLT128_MIN; // ??? // should be base^(emin-1)
}
static int emax () {
return FLT128_MAX_EXP;
}
static mag_type rmax () {
return FLT128_MAX; // ??? // should be (base^emax)*(1-eps)
}
};
#endif // HAVE_KOKKOSKERNELS_QUADMATH
template<>
class ArithTraits< ::Kokkos::complex<float> > {
public:
typedef ::Kokkos::complex<float> val_type;
typedef float mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = true;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type x) {
return ArithTraits<mag_type>::isInf (x.real ()) ||
ArithTraits<mag_type>::isInf (x.imag ());
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type x) {
return ArithTraits<mag_type>::isNan (x.real ()) ||
ArithTraits<mag_type>::isNan (x.imag ());
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return std::sqrt (::Kokkos::real (x) * ::Kokkos::real (x) +
::Kokkos::imag (x) * ::Kokkos::imag (x));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return val_type (ArithTraits<mag_type>::zero (), ArithTraits<mag_type>::zero ());
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return val_type (ArithTraits<mag_type>::one (), ArithTraits<mag_type>::zero ());
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return val_type (ArithTraits<mag_type>::min (), ArithTraits<mag_type>::min ()); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return val_type (ArithTraits<mag_type>::max (), ArithTraits<mag_type>::max ()); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x.real ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type x) {
return x.imag ();
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return ::Kokkos::conj (x);
}
// static KOKKOS_FORCEINLINE_FUNCTION val_type pow (const val_type x, const val_type y) {
// return ::pow (x, y);
// }
// static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// return ::sqrt (x);
// }
// static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
// return ::log (x);
// }
// static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
// return ::log10 (x);
// }
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// ???
return val_type (ArithTraits<mag_type>::nan (), ArithTraits<mag_type>::nan ());
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return ArithTraits<mag_type>::epsilon (); // ???
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef ::Kokkos::complex<ArithTraits<mag_type>::halfPrecision> halfPrecision;
typedef ::Kokkos::complex<ArithTraits<mag_type>::doublePrecision> doublePrecision;
static const bool isComplex = true;
static const bool isOrdinal = false;
static const bool isComparable = false;
static const bool hasMachineParameters = ArithTraits<mag_type>::hasMachineParameters;
static bool isnaninf (const val_type& x) {
return isNan (x) || isInf (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static std::string name () {
return "Kokkos::complex<float>";
}
// static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
// return sqrt (x);
// }
static KOKKOS_FORCEINLINE_FUNCTION mag_type eps () {
return epsilon ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type sfmin () {
return ArithTraits<mag_type>::sfmin (); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION int base () {
return ArithTraits<mag_type>::base ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type prec () {
return ArithTraits<mag_type>::prec (); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION int t () {
return ArithTraits<mag_type>::t ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rnd () {
return ArithTraits<mag_type>::rnd ();
}
static KOKKOS_FORCEINLINE_FUNCTION int emin () {
return ArithTraits<mag_type>::emin ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmin () {
return ArithTraits<mag_type>::rmin ();
}
static KOKKOS_FORCEINLINE_FUNCTION int emax () {
return ArithTraits<mag_type>::emax ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmax () {
return ArithTraits<mag_type>::rmax ();
}
};
template<>
class ArithTraits< ::Kokkos::complex<double> > {
public:
typedef ::Kokkos::complex<double> val_type;
typedef double mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = true;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type x) {
return ArithTraits<mag_type>::isInf (x.real ()) ||
ArithTraits<mag_type>::isInf (x.imag ());
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type x) {
return ArithTraits<mag_type>::isNan (x.real ()) ||
ArithTraits<mag_type>::isNan (x.imag ());
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return ::Kokkos::abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return val_type (ArithTraits<mag_type>::zero (), ArithTraits<mag_type>::zero ());
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return val_type (ArithTraits<mag_type>::one (), ArithTraits<mag_type>::zero ());
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return val_type (ArithTraits<mag_type>::min (), ArithTraits<mag_type>::min ()); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return val_type (ArithTraits<mag_type>::max (), ArithTraits<mag_type>::max ()); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x.real ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type x) {
return x.imag ();
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return ::Kokkos::conj (x);
}
// static KOKKOS_FORCEINLINE_FUNCTION val_type pow (const val_type x, const val_type y) {
// return ::pow (x, y);
// }
// static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// return ::sqrt (x);
// }
// static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
// return ::log (x);
// }
// static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
// return ::log10 (x);
// }
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// ???
return val_type (ArithTraits<mag_type>::nan (), ArithTraits<mag_type>::nan ());
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return ArithTraits<mag_type>::epsilon (); // ???
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef ::Kokkos::complex<ArithTraits<mag_type>::halfPrecision> halfPrecision;
typedef ::Kokkos::complex<ArithTraits<mag_type>::doublePrecision> doublePrecision;
static const bool isComplex = true;
static const bool isOrdinal = false;
static const bool isComparable = false;
static const bool hasMachineParameters = ArithTraits<mag_type>::hasMachineParameters;
static bool isnaninf (const val_type& x) {
return isNan (x) || isInf (x);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static std::string name () {
return "Kokkos::complex<double>";
}
// static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
// return sqrt (x);
// }
static KOKKOS_FORCEINLINE_FUNCTION mag_type eps () {
return epsilon ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type sfmin () {
return ArithTraits<mag_type>::sfmin (); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION int base () {
return ArithTraits<mag_type>::base ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type prec () {
return ArithTraits<mag_type>::prec (); // ???
}
static KOKKOS_FORCEINLINE_FUNCTION int t () {
return ArithTraits<mag_type>::t ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rnd () {
return ArithTraits<mag_type>::rnd ();
}
static KOKKOS_FORCEINLINE_FUNCTION int emin () {
return ArithTraits<mag_type>::emin ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmin () {
return ArithTraits<mag_type>::rmin ();
}
static KOKKOS_FORCEINLINE_FUNCTION int emax () {
return ArithTraits<mag_type>::emax ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type rmax () {
return ArithTraits<mag_type>::rmax ();
}
};
template<>
class ArithTraits<char> {
public:
typedef char val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
// The C(++) standard does not require that char be signed. In
// fact, signed char, unsigned char, and char are distinct types.
// We can use std::numeric_limits here because it's a const bool,
// not a class method.
static const bool is_signed = std::numeric_limits<char>::is_signed;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
// This may trigger a compiler warning if char is unsigned. On
// all platforms I have encountered, char is signed, but the C(++)
// standard does not require this.
return x >= 0 ? x : -x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return CHAR_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return CHAR_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
if (is_signed) {
return intPowSigned<val_type> (x, y);
} else {
return intPowUnsigned<val_type> (x, y);
}
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// C++11 defines std::sqrt for integer arguments. However, we
// currently can't assume C++11.
//
// This cast will result in no loss of accuracy, though it might
// be more expensive than it should, if we were clever about using
// bit operations.
//
// We take the absolute value first to avoid negative arguments.
// Negative real arguments to sqrt(float) return (float) NaN, but
// built-in integer types do not have an equivalent to NaN.
// Casting NaN to an integer type will thus result in some integer
// value which appears valid, but is not. We cannot raise an
// exception in device functions. Thus, we prefer to take the
// absolute value of x first, to avoid issues. Another
// possibility would be to test for a NaN output and convert it to
// some reasonable value (like 0), though this might be more
// expensive than the absolute value interpreted using the ternary
// operator.
return static_cast<val_type> ( ::sqrt (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "char";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<signed char> {
public:
typedef signed char val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x >= 0 ? x : -x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return SCHAR_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return SCHAR_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowSigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
return static_cast<val_type> ( ::sqrt (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "signed char";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<unsigned char> {
public:
typedef unsigned char val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = false;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x; // it's unsigned, so it's positive
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return UCHAR_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowUnsigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// This will result in no loss of accuracy, though it might be
// more expensive than it should, if we were clever about using
// bit operations.
return static_cast<val_type> ( ::sqrt (static_cast<float> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<float> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<float> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "unsigned char";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<short> {
public:
typedef short val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
// std::abs appears to work with CUDA 5.5 at least, but I'll use
// the ternary expression for maximum generality. Note that this
// expression does not necessarily obey the rules for fabs() with
// NaN input, so it should not be used for floating-point types.
// It's perfectly fine for signed integer types, though.
return x >= 0 ? x : -x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
// Macros like this work with CUDA, but
// std::numeric_limits<val_type>::min() does not, because it is
// not marked as a __device__ function.
return SHRT_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return SHRT_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type pow (const val_type x, const val_type y) {
return intPowSigned<val_type> (x, y);
}
//! Integer square root returns a lower bound.
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// This will result in no loss of accuracy, though it might be
// more expensive than it should, if we were clever about using
// bit operations.
return static_cast<val_type> ( ::sqrt (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<float> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// short doesn't implement a NaN value, but we can still have it
// return some "flag" value that can help users find use of
// uninitialized data.
return static_cast<val_type> (-1);
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "short";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<unsigned short> {
public:
typedef unsigned short val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = false;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x; // it's unsigned, so it's positive
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return USHRT_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowUnsigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// This will result in no loss of accuracy, though it might be
// more expensive than it should, if we were clever about using
// bit operations.
return static_cast<val_type> ( ::sqrt (static_cast<float> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<float> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<float> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// unsigned short doesn't implement a NaN value, but we can still
// have it return some "flag" value that can help users find use
// of uninitialized data.
return max ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "unsigned short";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<int> {
public:
typedef int val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
// std::abs appears to work with CUDA 5.5 at least, but I'll use
// the ternary expression for maximum generality. Note that this
// expression does not necessarily obey the rules for fabs() with
// NaN input, so it should not be used for floating-point types.
// It's perfectly fine for signed integer types, though.
return x >= 0 ? x : -x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
// Macros like INT_MIN work with CUDA, but
// std::numeric_limits<val_type>::min() does not, because it is
// not marked as a __device__ function.
return INT_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return INT_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowSigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// This will result in no loss of accuracy, though it might be
// more expensive than it should, if we were clever about using
// bit operations.
return static_cast<val_type> ( ::sqrt (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// int doesn't implement a NaN value, but we can still have it
// return some "flag" value that can help users find use of
// uninitialized data.
return -1;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "int";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<unsigned int> {
public:
typedef unsigned int val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = false;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x; // it's unsigned, so it's positive
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return UINT_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowUnsigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
// This will result in no loss of accuracy, though it might be
// more expensive than it should, if we were clever about using
// bit operations.
return static_cast<val_type> ( ::sqrt (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// unsigned int doesn't implement a NaN value, but we can still
// have it return some "flag" value that can help users find use
// of uninitialized data.
return max ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "unsigned int";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<long> {
public:
typedef long val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x >= 0 ? x : -x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return LONG_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return LONG_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowSigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
#ifdef __CUDA_ARCH__
return static_cast<val_type> ( ::sqrt (static_cast<double> (abs (x))));
#else
return static_cast<val_type> ( ::sqrt (static_cast<long double> (abs (x))));
#endif // __CUDA_ARCH__
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// long doesn't implement a NaN value, but we can still have it
// return some "flag" value that can help users find use of
// uninitialized data.
return -1;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "long";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<unsigned long> {
public:
typedef unsigned long val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = false;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return ULONG_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type pow (const val_type x, const val_type y) {
return intPowUnsigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
#ifdef __CUDA_ARCH__
return static_cast<val_type> ( ::sqrt (static_cast<double> (x)));
#else
return static_cast<val_type> ( ::sqrt (static_cast<long double> (x)));
#endif // __CUDA_ARCH__
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<long> ( ::log (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<long> ( ::log10 (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// unsigned long doesn't implement a NaN value, but we can still
// have it return some "flag" value that can help users find use
// of uninitialized data.
return max ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "unsigned long";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<long long> {
public:
typedef long long val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x >= 0 ? x : -x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return LLONG_MIN;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return LLONG_MAX;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowSigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
#ifdef __CUDA_ARCH__
// Casting from a 64-bit integer type to double does result in a
// loss of accuracy. However, it gives us a good first
// approximation. For very large numbers, we may lose some
// significand bits, but will always get within a factor of two
// (assuming correct rounding) of the exact double-precision
// number. We could then binary search between half the result
// and twice the result (assuming the latter is <= INT64_MAX,
// which it has to be, so we don't have to check) to ensure
// correctness. It actually should suffice to check numbers
// within 1 of the result.
return static_cast<val_type> ( ::sqrt (static_cast<double> (abs (x))));
#else
// IEEE 754 promises that long double has at least 64 significand
// bits, so we can use it to represent any signed or unsigned
// 64-bit integer type exactly. However, CUDA does not implement
// long double for device functions.
return static_cast<val_type> ( ::sqrt (static_cast<long double> (abs (x))));
#endif // __CUDA_ARCH__
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<double> (abs (x))));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// long long doesn't implement a NaN value, but we can still have
// it return some "flag" value that can help users find use of
// uninitialized data.
return -1;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "long long";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
template<>
class ArithTraits<unsigned long long> {
public:
typedef unsigned long long val_type;
typedef val_type mag_type;
static const bool is_specialized = true;
static const bool is_signed = false;
static const bool is_integer = true;
static const bool is_exact = true;
static const bool is_complex = false;
static KOKKOS_FORCEINLINE_FUNCTION bool isInf (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION bool isNan (const val_type ) {
return false;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type abs (const val_type x) {
return x; // unsigned integers are always nonnegative
}
static KOKKOS_FORCEINLINE_FUNCTION val_type zero () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type one () {
return 1;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type min () {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type max () {
return ULLONG_MAX ;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type real (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type imag (const val_type ) {
return 0;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conj (const val_type x) {
return x;
}
static KOKKOS_FORCEINLINE_FUNCTION val_type
pow (const val_type x, const val_type y) {
return intPowUnsigned<val_type> (x, y);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type sqrt (const val_type x) {
#ifdef __CUDA_ARCH__
return static_cast<val_type> ( ::sqrt (static_cast<double> (x)));
#else
return static_cast<val_type> ( ::sqrt (static_cast<long double> (x)));
#endif // __CUDA_ARCH__
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log (const val_type x) {
return static_cast<val_type> ( ::log (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type log10 (const val_type x) {
return static_cast<val_type> ( ::log10 (static_cast<double> (x)));
}
static KOKKOS_FORCEINLINE_FUNCTION val_type nan () {
// unsigned long long doesn't implement a NaN value, but we can
// still have it return some "flag" value that can help users find
// use of uninitialized data.
return max ();
}
static KOKKOS_FORCEINLINE_FUNCTION mag_type epsilon () {
return zero ();
}
// Backwards compatibility with Teuchos::ScalarTraits.
typedef mag_type magnitudeType;
typedef val_type halfPrecision;
typedef val_type doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = true;
static const bool isComparable = true;
static const bool hasMachineParameters = false;
static KOKKOS_FORCEINLINE_FUNCTION magnitudeType magnitude (const val_type x) {
return abs (x);
}
static KOKKOS_FORCEINLINE_FUNCTION val_type conjugate (const val_type x) {
return conj (x);
}
static KOKKOS_FORCEINLINE_FUNCTION bool isnaninf (const val_type) {
return false;
}
static std::string name () {
return "unsigned long long";
}
static KOKKOS_FORCEINLINE_FUNCTION val_type squareroot (const val_type x) {
return sqrt (x);
}
};
// dd_real and qd_real are floating-point types provided by the QD
// library of David Bailey (LBNL):
//
// http://crd-legacy.lbl.gov/~dhbailey/mpdist/
//
// dd_real uses two doubles (128 bits), and qd_real uses four doubles
// (256 bits).
//
// Kokkos does <i>not</i> currently support these types in device
// functions. It should be possible to use Kokkos' support for
// aggregate types to implement device function support for dd_real
// and qd_real, but we have not done this yet (as of 09 Jan 2015).
// Hence, the class methods of the ArithTraits specializations for
// dd_real and qd_real are not marked as device functions.
#ifdef HAVE_KOKKOS_QD
template<>
struct ArithTraits<dd_real>
{
typedef dd_real val_type;
typedef dd_real mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = false;
static inline bool isInf (const val_type& x) {
return isinf (x);
}
static inline bool isNan (const val_type& x) {
return isnan (x);
}
static inline mag_type abs (const val_type& x) {
return ::abs (x);
}
static inline val_type zero () {
return val_type (0.0);
}
static inline val_type one () {
return val_type (1.0);
}
static inline val_type min () {
return std::numeric_limits<val_type>::min ();
}
static inline val_type max () {
return std::numeric_limits<val_type>::max ();
}
static inline mag_type real (const val_type& x) {
return x;
}
static inline mag_type imag (const val_type&) {
return zero ();
}
static inline val_type conj (const val_type& x) {
return x;
}
static inline val_type pow (const val_type& x, const val_type& y) {
return ::pow(x,y);
}
static inline val_type sqrt (const val_type& x) {
return ::sqrt (x);
}
static inline val_type log (const val_type& x) {
// dd_real puts its transcendental functions in the global namespace.
return ::log (x);
}
static inline val_type log10 (const val_type& x) {
return ::log10 (x);
}
static inline val_type nan () {
return val_type::_nan;
}
static val_type epsilon () {
return std::numeric_limits<val_type>::epsilon ();
}
typedef dd_real magnitudeType;
typedef double halfPrecision;
typedef qd_real doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = true;
static const bool hasMachineParameters = true;
static mag_type eps () {
return epsilon ();
}
static mag_type sfmin () {
return min ();
}
static int base () {
return std::numeric_limits<val_type>::radix;
}
static mag_type prec () {
return eps () * base ();
}
static int t () {
return std::numeric_limits<val_type>::digits;
}
static mag_type rnd () {
return std::numeric_limits<val_type>::round_style == std::round_to_nearest ?
one () :
zero ();
}
static int emin () {
return std::numeric_limits<val_type>::min_exponent;
}
static mag_type rmin () {
return std::numeric_limits<val_type>::min ();
}
static int emax () {
return std::numeric_limits<val_type>::max_exponent;
}
static mag_type rmax () {
return std::numeric_limits<val_type>::max ();
}
static mag_type magnitude (const val_type& x) {
return ::abs (x);
}
static val_type conjugate (const val_type& x) {
return conj (x);
}
static bool isnaninf (const val_type& x) {
return isNan (x) || isInf (x);
}
static std::string name () { return "dd_real"; }
static val_type squareroot (const val_type& x) {
return ::sqrt (x);
}
};
template<>
struct ArithTraits<qd_real>
{
typedef qd_real val_type;
typedef qd_real mag_type;
static const bool is_specialized = true;
static const bool is_signed = true;
static const bool is_integer = false;
static const bool is_exact = false;
static const bool is_complex = false;
static inline bool isInf (const val_type& x) {
return isinf (x);
}
static inline bool isNan (const val_type& x) {
return isnan (x);
}
static inline mag_type abs (const val_type& x) {
return ::abs (x);
}
static inline val_type zero () {
return val_type (0.0);
}
static inline val_type one () {
return val_type (1.0);
}
static inline val_type min () {
return std::numeric_limits<val_type>::min ();
}
static inline val_type max () {
return std::numeric_limits<val_type>::max ();
}
static inline mag_type real (const val_type& x) {
return x;
}
static inline mag_type imag (const val_type&) {
return zero ();
}
static inline val_type conj (const val_type& x) {
return x;
}
static inline val_type pow (const val_type& x, const val_type& y) {
return ::pow (x, y);
}
static inline val_type sqrt (const val_type& x) {
return ::sqrt (x);
}
static inline val_type log (const val_type& x) {
// val_type puts its transcendental functions in the global namespace.
return ::log (x);
}
static inline val_type log10 (const val_type& x) {
return ::log10 (x);
}
static inline val_type nan () {
return val_type::_nan;
}
static inline val_type epsilon () {
return std::numeric_limits<val_type>::epsilon ();
}
typedef qd_real magnitudeType;
typedef dd_real halfPrecision;
// The QD library does not have an "oct-double real" class. One
// could use an arbitrary-precision library like MPFR or ARPREC,
// with the precision set appropriately, to get an
// extended-precision type for qd_real.
typedef qd_real doublePrecision;
static const bool isComplex = false;
static const bool isOrdinal = false;
static const bool isComparable = true;
static const bool hasMachineParameters = true;
static mag_type eps () {
return epsilon ();
}
static mag_type sfmin () {
return min ();
}
static int base () {
return std::numeric_limits<val_type>::radix;
}
static mag_type prec () {
return eps () * base ();
}
static int t () {
return std::numeric_limits<val_type>::digits;
}
static mag_type rnd () {
return std::numeric_limits<val_type>::round_style == std::round_to_nearest ?
one () :
zero ();
}
static int emin () {
return std::numeric_limits<val_type>::min_exponent;
}
static mag_type rmin () {
return std::numeric_limits<val_type>::min ();
}
static int emax () {
return std::numeric_limits<val_type>::max_exponent;
}
static mag_type rmax () {
return std::numeric_limits<val_type>::max ();
}
static mag_type magnitude (const val_type& x) {
return ::abs (x);
}
static val_type conjugate (const val_type& x) {
return conj (x);
}
static bool isnaninf (const val_type& x) {
return isNan (x) || isInf (x);
}
static std::string name () { return "qd_real"; }
static val_type squareroot (const val_type& x) {
return ::sqrt (x);
}
};
#endif // HAVE_KOKKOS_QD
} // namespace Details
// Promote ArithTraits into Kokkos namespace. At some point, we
// will remove it from the Details namespace completely. We leave
// it there for now, because a lot of code depends on it being
// there.
using Details::ArithTraits;
} // namespace Kokkos
#endif // KOKKOS_ARITHTRAITS_HPP
|