Path: blob/master/test/jdk/java/lang/Math/HyperbolicTests.java
41149 views
/*1* Copyright (c) 2003, 2012, Oracle and/or its affiliates. All rights reserved.2* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.3*4* This code is free software; you can redistribute it and/or modify it5* under the terms of the GNU General Public License version 2 only, as6* published by the Free Software Foundation.7*8* This code is distributed in the hope that it will be useful, but WITHOUT9* ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or10* FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License11* version 2 for more details (a copy is included in the LICENSE file that12* accompanied this code).13*14* You should have received a copy of the GNU General Public License version15* 2 along with this work; if not, write to the Free Software Foundation,16* Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA.17*18* Please contact Oracle, 500 Oracle Parkway, Redwood Shores, CA 94065 USA19* or visit www.oracle.com if you need additional information or have any20* questions.21*/2223/*24* @test25* @bug 4851625 4900189 493944126* @summary Tests for {Math, StrictMath}.{sinh, cosh, tanh}27* @author Joseph D. Darcy28*/2930public class HyperbolicTests {31private HyperbolicTests(){}3233static final double NaNd = Double.NaN;3435/**36* Test accuracy of {Math, StrictMath}.sinh. The specified37* accuracy is 2.5 ulps.38*39* The defintion of sinh(x) is40*41* (e^x - e^(-x))/242*43* The series expansion of sinh(x) =44*45* x + x^3/3! + x^5/5! + x^7/7! +...46*47* Therefore,48*49* 1. For large values of x sinh(x) ~= signum(x)*exp(|x|)/250*51* 2. For small values of x, sinh(x) ~= x.52*53* Additionally, sinh is an odd function; sinh(-x) = -sinh(x).54*55*/56static int testSinh() {57int failures = 0;58/*59* Array elements below generated using a quad sinh60* implementation. Rounded to a double, the quad result61* *should* be correctly rounded, unless we are quite unlucky.62* Assuming the quad value is a correctly rounded double, the63* allowed error is 3.0 ulps instead of 2.5 since the quad64* value rounded to double can have its own 1/2 ulp error.65*/66double [][] testCases = {67// x sinh(x)68{0.0625, 0.06254069805219182172183988501029229},69{0.1250, 0.12532577524111545698205754229137154},70{0.1875, 0.18860056562029018382047025055167585},71{0.2500, 0.25261231680816830791412515054205787},72{0.3125, 0.31761115611357728583959867611490292},73{0.3750, 0.38385106791361456875429567642050245},74{0.4375, 0.45159088610312053032509815226723017},75{0.5000, 0.52109530549374736162242562641149155},76{0.5625, 0.59263591611468777373870867338492247},77{0.6250, 0.66649226445661608227260655608302908},78{0.6875, 0.74295294580567543571442036910465007},79{0.7500, 0.82231673193582998070366163444691386},80{0.8125, 0.90489373856606433650504536421491368},81{0.8750, 0.99100663714429475605317427568995231},82{0.9375, 1.08099191569306394011007867453992548},83{1.0000, 1.17520119364380145688238185059560082},84{1.0625, 1.27400259579739321279181130344911907},85{1.1250, 1.37778219077984075760379987065228373},86{1.1875, 1.48694549961380717221109202361777593},87{1.2500, 1.60191908030082563790283030151221415},88{1.3125, 1.72315219460596010219069206464391528},89{1.3750, 1.85111856355791532419998548438506416},90{1.4375, 1.98631821852425112898943304217629457},91{1.5000, 2.12927945509481749683438749467763195},92{1.5625, 2.28056089740825247058075476705718764},93{1.6250, 2.44075368098794353221372986997161132},94{1.6875, 2.61048376261693140366028569794027603},95{1.7500, 2.79041436627764265509289122308816092},96{1.8125, 2.98124857471401377943765253243875520},97{1.8750, 3.18373207674259205101326780071803724},98{1.9375, 3.39865608104779099764440244167531810},99{2.0000, 3.62686040784701876766821398280126192},100{2.0625, 3.86923677050642806693938384073620450},101{2.1250, 4.12673225993027252260441410537905269},102{2.1875, 4.40035304533919660406976249684469164},103{2.2500, 4.69116830589833069188357567763552003},104{2.3125, 5.00031440855811351554075363240262157},105{2.3750, 5.32899934843284576394645856548481489},106{2.4375, 5.67850746906785056212578751630266858},107{2.5000, 6.05020448103978732145032363835040319},108{2.5625, 6.44554279850040875063706020260185553},109{2.6250, 6.86606721451642172826145238779845813},110{2.6875, 7.31342093738196587585692115636603571},111{2.7500, 7.78935201149073201875513401029935330},112{2.8125, 8.29572014785741787167717932988491961},113{2.8750, 8.83450399097893197351853322827892144},114{2.9375, 9.40780885043076394429977972921690859},115{3.0000, 10.01787492740990189897459361946582867},116{3.0625, 10.66708606836969224165124519209968368},117{3.1250, 11.35797907995166028304704128775698426},118{3.1875, 12.09325364161259019614431093344260209},119{3.2500, 12.87578285468067003959660391705481220},120{3.3125, 13.70862446906136798063935858393686525},121{3.3750, 14.59503283146163690015482636921657975},122{3.4375, 15.53847160182039311025096666980558478},123{3.5000, 16.54262728763499762495673152901249743},124{3.5625, 17.61142364906941482858466494889121694},125{3.6250, 18.74903703113232171399165788088277979},126{3.6875, 19.95991268283598684128844120984214675},127{3.7500, 21.24878212710338697364101071825171163},128{3.8125, 22.62068164929685091969259499078125023},129{3.8750, 24.08097197661255803883403419733891573},130{3.9375, 25.63535922523855307175060244757748997},131{4.0000, 27.28991719712775244890827159079382096},132{4.0625, 29.05111111351106713777825462100160185},133{4.1250, 30.92582287788986031725487699744107092},134{4.1875, 32.92137796722343190618721270937061472},135{4.2500, 35.04557405638942942322929652461901154},136{4.3125, 37.30671148776788628118833357170042385},137{4.3750, 39.71362570500944929025069048612806024},138{4.4375, 42.27572177772344954814418332587050658},139{4.5000, 45.00301115199178562180965680564371424},140{4.5625, 47.90615077031205065685078058248081891},141{4.6250, 50.99648471383193131253995134526177467},142{4.6875, 54.28608852959281437757368957713936555},143{4.7500, 57.78781641599226874961859781628591635},144{4.8125, 61.51535145084362283008545918273109379},145{4.8750, 65.48325905829987165560146562921543361},146{4.9375, 69.70704392356508084094318094283346381},147{5.0000, 74.20321057778875897700947199606456364},148{5.0625, 78.98932788987998983462810080907521151},149{5.1250, 84.08409771724448958901392613147384951},150{5.1875, 89.50742798369883598816307922895346849},151{5.2500, 95.28051047011540739630959111303975956},152{5.3125, 101.42590362176666730633859252034238987},153{5.3750, 107.96762069594029162704530843962700133},154{5.4375, 114.93122359426386042048760580590182604},155{5.5000, 122.34392274639096192409774240457730721},156{5.5625, 130.23468343534638291488502321709913206},157{5.6250, 138.63433897999898233879574111119546728},158{5.6875, 147.57571121692522056519568264304815790},159{5.7500, 157.09373875244884423880085377625986165},160{5.8125, 167.22561348600435888568183143777868662},161{5.8750, 178.01092593829229887752609866133883987},162{5.9375, 189.49181995209921964640216682906501778},163{6.0000, 201.71315737027922812498206768797872263},164{6.0625, 214.72269333437984291483666459592578915},165{6.1250, 228.57126288889537420461281285729970085},166{6.1875, 243.31297962030799867970551767086092471},167{6.2500, 259.00544710710289911522315435345489966},168{6.3125, 275.70998400700299790136562219920451185},169{6.3750, 293.49186366095654566861661249898332253},170{6.4375, 312.42056915013535342987623229485223434},171{6.5000, 332.57006480258443156075705566965111346},172{6.5625, 354.01908521044116928437570109827956007},173{6.6250, 376.85144288706511933454985188849781703},174{6.6875, 401.15635576625530823119100750634165252},175{6.7500, 427.02879582326538080306830640235938517},176{6.8125, 454.56986017986077163530945733572724452},177{6.8750, 483.88716614351897894746751705315210621},178{6.9375, 515.09527172439720070161654727225752288},179{7.0000, 548.31612327324652237375611757601851598},180{7.0625, 583.67953198942753384680988096024373270},181{7.1250, 621.32368116099280160364794462812762880},182{7.1875, 661.39566611888784148449430491465857519},183{7.2500, 704.05206901515336623551137120663358760},184{7.3125, 749.45957067108712382864538206200700256},185{7.3750, 797.79560188617531521347351754559776282},186{7.4375, 849.24903675279739482863565789325699416},187{7.5000, 904.02093068584652953510919038935849651},188{7.5625, 962.32530605113249628368993221570636328},189{7.6250, 1024.38998846242707559349318193113614698},190{7.6875, 1090.45749701500081956792547346904792325},191{7.7500, 1160.78599193425808533255719118417856088},192{7.8125, 1235.65028334242796895820912936318532502},193{7.8750, 1315.34290508508890654067255740428824014},194{7.9375, 1400.17525781352742299995139486063802583},195{8.0000, 1490.47882578955018611587663903188144796},196{8.0625, 1586.60647216744061169450001100145859236},197{8.1250, 1688.93381781440241350635231605477507900},198{8.1875, 1797.86070905726094477721128358866360644},199{8.2500, 1913.81278009067446281883262689250118009},200{8.3125, 2037.24311615199935553277163192983440062},201{8.3750, 2168.63402396170125867037749369723761636},202{8.4375, 2308.49891634734644432370720900969004306},203{8.5000, 2457.38431841538268239359965370719928775},204{8.5625, 2615.87200310986940554256648824234335262},205{8.6250, 2784.58126450289932429469130598902487336},206{8.6875, 2964.17133769964321637973459949999057146},207{8.7500, 3155.34397481384944060352507473513108710},208{8.8125, 3358.84618707947841898217318996045550438},209{8.8750, 3575.47316381333288862617411467285480067},210{8.9375, 3806.07137963459383403903729660349293583},211{9.0000, 4051.54190208278996051522359589803425598},212{9.0625, 4312.84391255878980330955246931164633615},213{9.1250, 4590.99845434696991399363282718106006883},214{9.1875, 4887.09242236403719571363798584676797558},215{9.2500, 5202.28281022453561319352901552085348309},216{9.3125, 5537.80123121853803935727335892054791265},217{9.3750, 5894.95873086734181634245918412592155656},218{9.4375, 6275.15090986233399457103055108344546942},219{9.5000, 6679.86337740502119410058225086262108741},220{9.5625, 7110.67755625726876329967852256934334025},221{9.6250, 7569.27686218510919585241049433331592115},222{9.6875, 8057.45328194243077504648484392156371121},223{9.7500, 8577.11437549816065709098061006273039092},224{9.8125, 9130.29072986829727910801024120918114778},225{9.8750, 9719.14389367880274015504995181862860062},226{9.9375, 10345.97482346383208590278839409938269134},227{10.0000, 11013.23287470339337723652455484636420303},228};229230for(int i = 0; i < testCases.length; i++) {231double [] testCase = testCases[i];232failures += testSinhCaseWithUlpDiff(testCase[0],233testCase[1],2343.0);235}236237double [][] specialTestCases = {238{0.0, 0.0},239{NaNd, NaNd},240{Double.longBitsToDouble(0x7FF0000000000001L), NaNd},241{Double.longBitsToDouble(0xFFF0000000000001L), NaNd},242{Double.longBitsToDouble(0x7FF8555555555555L), NaNd},243{Double.longBitsToDouble(0xFFF8555555555555L), NaNd},244{Double.longBitsToDouble(0x7FFFFFFFFFFFFFFFL), NaNd},245{Double.longBitsToDouble(0xFFFFFFFFFFFFFFFFL), NaNd},246{Double.longBitsToDouble(0x7FFDeadBeef00000L), NaNd},247{Double.longBitsToDouble(0xFFFDeadBeef00000L), NaNd},248{Double.longBitsToDouble(0x7FFCafeBabe00000L), NaNd},249{Double.longBitsToDouble(0xFFFCafeBabe00000L), NaNd},250{Double.POSITIVE_INFINITY, Double.POSITIVE_INFINITY}251};252253for(int i = 0; i < specialTestCases.length; i++) {254failures += testSinhCaseWithUlpDiff(specialTestCases[i][0],255specialTestCases[i][1],2560.0);257}258259// For powers of 2 less than 2^(-27), the second and260// subsequent terms of the Taylor series expansion will get261// rounded away since |n-n^3| > 53, the binary precision of a262// double significand.263264for(int i = DoubleConsts.MIN_SUB_EXPONENT; i < -27; i++) {265double d = Math.scalb(2.0, i);266267// Result and expected are the same.268failures += testSinhCaseWithUlpDiff(d, d, 2.5);269}270271// For values of x larger than 22, the e^(-x) term is272// insignificant to the floating-point result. Util exp(x)273// overflows around 709.8, sinh(x) ~= exp(x)/2; will will test274// 10000 values in this range.275276long trans22 = Double.doubleToLongBits(22.0);277// (approximately) largest value such that exp shouldn't278// overflow279long transExpOvfl = Double.doubleToLongBits(Math.nextDown(709.7827128933841));280281for(long i = trans22;282i < transExpOvfl;283i +=(transExpOvfl-trans22)/10000) {284285double d = Double.longBitsToDouble(i);286287// Allow 3.5 ulps of error to deal with error in exp.288failures += testSinhCaseWithUlpDiff(d, StrictMath.exp(d)*0.5, 3.5);289}290291// (approximately) largest value such that sinh shouldn't292// overflow.293long transSinhOvfl = Double.doubleToLongBits(710.4758600739439);294295// Make sure sinh(x) doesn't overflow as soon as exp(x)296// overflows.297298/*299* For large values of x, sinh(x) ~= 0.5*(e^x). Therefore,300*301* sinh(x) ~= e^(ln 0.5) * e^x = e^(x + ln 0.5)302*303* So, we can calculate the approximate expected result as304* exp(x + -0.693147186). However, this sum suffers from305* roundoff, limiting the accuracy of the approximation. The306* accuracy can be improved by recovering the rounded-off307* information. Since x is larger than ln(0.5), the trailing308* bits of ln(0.5) get rounded away when the two values are309* added. However, high-order bits of ln(0.5) that310* contribute to the sum can be found:311*312* offset = log(0.5);313* effective_offset = (x + offset) - x; // exact subtraction314* rounded_away_offset = offset - effective_offset; // exact subtraction315*316* Therefore, the product317*318* exp(x + offset)*exp(rounded_away_offset)319*320* will be a better approximation to the exact value of321*322* e^(x + offset)323*324* than exp(x+offset) alone. (The expected result cannot be325* computed as exp(x)*exp(offset) since exp(x) by itself would326* overflow to infinity.)327*/328double offset = StrictMath.log(0.5);329for(long i = transExpOvfl+1; i < transSinhOvfl;330i += (transSinhOvfl-transExpOvfl)/1000 ) {331double input = Double.longBitsToDouble(i);332333double expected =334StrictMath.exp(input + offset) *335StrictMath.exp( offset - ((input + offset) - input) );336337failures += testSinhCaseWithUlpDiff(input, expected, 4.0);338}339340// sinh(x) overflows for values greater than 710; in341// particular, it overflows for all 2^i, i > 10.342for(int i = 10; i <= Double.MAX_EXPONENT; i++) {343double d = Math.scalb(2.0, i);344345// Result and expected are the same.346failures += testSinhCaseWithUlpDiff(d,347Double.POSITIVE_INFINITY, 0.0);348}349350return failures;351}352353public static int testSinhCaseWithTolerance(double input,354double expected,355double tolerance) {356int failures = 0;357failures += Tests.testTolerance("Math.sinh(double)",358input, Math.sinh(input),359expected, tolerance);360failures += Tests.testTolerance("Math.sinh(double)",361-input, Math.sinh(-input),362-expected, tolerance);363364failures += Tests.testTolerance("StrictMath.sinh(double)",365input, StrictMath.sinh(input),366expected, tolerance);367failures += Tests.testTolerance("StrictMath.sinh(double)",368-input, StrictMath.sinh(-input),369-expected, tolerance);370return failures;371}372373public static int testSinhCaseWithUlpDiff(double input,374double expected,375double ulps) {376int failures = 0;377failures += Tests.testUlpDiff("Math.sinh(double)",378input, Math.sinh(input),379expected, ulps);380failures += Tests.testUlpDiff("Math.sinh(double)",381-input, Math.sinh(-input),382-expected, ulps);383384failures += Tests.testUlpDiff("StrictMath.sinh(double)",385input, StrictMath.sinh(input),386expected, ulps);387failures += Tests.testUlpDiff("StrictMath.sinh(double)",388-input, StrictMath.sinh(-input),389-expected, ulps);390return failures;391}392393394/**395* Test accuracy of {Math, StrictMath}.cosh. The specified396* accuracy is 2.5 ulps.397*398* The defintion of cosh(x) is399*400* (e^x + e^(-x))/2401*402* The series expansion of cosh(x) =403*404* 1 + x^2/2! + x^4/4! + x^6/6! +...405*406* Therefore,407*408* 1. For large values of x cosh(x) ~= exp(|x|)/2409*410* 2. For small values of x, cosh(x) ~= 1.411*412* Additionally, cosh is an even function; cosh(-x) = cosh(x).413*414*/415static int testCosh() {416int failures = 0;417/*418* Array elements below generated using a quad cosh419* implementation. Rounded to a double, the quad result420* *should* be correctly rounded, unless we are quite unlucky.421* Assuming the quad value is a correctly rounded double, the422* allowed error is 3.0 ulps instead of 2.5 since the quad423* value rounded to double can have its own 1/2 ulp error.424*/425double [][] testCases = {426// x cosh(x)427{0.0625, 1.001953760865667607841550709632597376},428{0.1250, 1.007822677825710859846949685520422223},429{0.1875, 1.017629683800690526835115759894757615},430{0.2500, 1.031413099879573176159295417520378622},431{0.3125, 1.049226785060219076999158096606305793},432{0.3750, 1.071140346704586767299498015567016002},433{0.4375, 1.097239412531012567673453832328262160},434{0.5000, 1.127625965206380785226225161402672030},435{0.5625, 1.162418740845610783505338363214045218},436{0.6250, 1.201753692975606324229229064105075301},437{0.6875, 1.245784523776616395403056980542275175},438{0.7500, 1.294683284676844687841708185390181730},439{0.8125, 1.348641048647144208352285714214372703},440{0.8750, 1.407868656822803158638471458026344506},441{0.9375, 1.472597542369862933336886403008640891},442{1.0000, 1.543080634815243778477905620757061497},443{1.0625, 1.619593348374367728682469968448090763},444{1.1250, 1.702434658138190487400868008124755757},445{1.1875, 1.791928268324866464246665745956119612},446{1.2500, 1.888423877161015738227715728160051696},447{1.3125, 1.992298543335143985091891077551921106},448{1.3750, 2.103958159362661802010972984204389619},449{1.4375, 2.223839037619709260803023946704272699},450{1.5000, 2.352409615243247325767667965441644201},451{1.5625, 2.490172284559350293104864895029231913},452{1.6250, 2.637665356192137582275019088061812951},453{1.6875, 2.795465162524235691253423614360562624},454{1.7500, 2.964188309728087781773608481754531801},455{1.8125, 3.144494087167972176411236052303565201},456{1.8750, 3.337087043587520514308832278928116525},457{1.9375, 3.542719740149244276729383650503145346},458{2.0000, 3.762195691083631459562213477773746099},459{2.0625, 3.996372503438463642260225717607554880},460{2.1250, 4.246165228196992140600291052990934410},461{2.1875, 4.512549935859540340856119781585096760},462{2.2500, 4.796567530460195028666793366876218854},463{2.3125, 5.099327816921939817643745917141739051},464{2.3750, 5.422013837643509250646323138888569746},465{2.4375, 5.765886495263270945949271410819116399},466{2.5000, 6.132289479663686116619852312817562517},467{2.5625, 6.522654518468725462969589397439224177},468{2.6250, 6.938506971550673190999796241172117288},469{2.6875, 7.381471791406976069645686221095397137},470{2.7500, 7.853279872697439591457564035857305647},471{2.8125, 8.355774815752725814638234943192709129},472{2.8750, 8.890920130482709321824793617157134961},473{2.9375, 9.460806908834119747071078865866737196},474{3.0000, 10.067661995777765841953936035115890343},475{3.0625, 10.713856690753651225304006562698007312},476{3.1250, 11.401916013575067700373788969458446177},477{3.1875, 12.134528570998387744547733730974713055},478{3.2500, 12.914557062512392049483503752322408761},479{3.3125, 13.745049466398732213877084541992751273},480{3.3750, 14.629250949773302934853381428660210721},481{3.4375, 15.570616549147269180921654324879141947},482{3.5000, 16.572824671057316125696517821376119469},483{3.5625, 17.639791465519127930722105721028711044},484{3.6250, 18.775686128468677200079039891415789429},485{3.6875, 19.984947192985946987799359614758598457},486{3.7500, 21.272299872959396081877161903352144126},487{3.8125, 22.642774526961913363958587775566619798},488{3.8750, 24.101726314486257781049388094955970560},489{3.9375, 25.654856121347151067170940701379544221},490{4.0000, 27.308232836016486629201989612067059978},491{4.0625, 29.068317063936918520135334110824828950},492{4.1250, 30.941986372478026192360480044849306606},493{4.1875, 32.936562165180269851350626768308756303},494{4.2500, 35.059838290298428678502583470475012235},495{4.3125, 37.320111495433027109832850313172338419},496{4.3750, 39.726213847251883288518263854094284091},497{4.4375, 42.287547242982546165696077854963452084},498{4.5000, 45.014120148530027928305799939930642658},499{4.5625, 47.916586706774825161786212701923307169},500{4.6250, 51.006288368867753140854830589583165950},501{4.6875, 54.295298211196782516984520211780624960},502{4.7500, 57.796468111195389383795669320243166117},503{4.8125, 61.523478966332915041549750463563672435},504{4.8750, 65.490894152518731617237739112888213645},505{4.9375, 69.714216430810089539924900313140922323},506{5.0000, 74.209948524787844444106108044487704798},507{5.0625, 78.995657605307475581204965926043112946},508{5.1250, 84.090043934600961683400343038519519678},509{5.1875, 89.513013937957834087706670952561002466},510{5.2500, 95.285757988514588780586084642381131013},511{5.3125, 101.430833209098212357990123684449846912},512{5.3750, 107.972251614673824873137995865940755392},513{5.4375, 114.935573939814969189535554289886848550},514{5.5000, 122.348009517829425991091207107262038316},515{5.5625, 130.238522601820409078244923165746295574},516{5.6250, 138.637945543134998069351279801575968875},517{5.6875, 147.579099269447055276899288971207106581},518{5.7500, 157.096921533245353905868840194264636395},519{5.8125, 167.228603431860671946045256541679445836},520{5.8750, 178.013734732486824390148614309727161925},521{5.9375, 189.494458570056311567917444025807275896},522{6.0000, 201.715636122455894483405112855409538488},523{6.0625, 214.725021906554080628430756558271312513},524{6.1250, 228.573450380013557089736092321068279231},525{6.1875, 243.315034578039208138752165587134488645},526{6.2500, 259.007377561239126824465367865430519592},527{6.3125, 275.711797500835732516530131577254654076},528{6.3750, 293.493567280752348242602902925987643443},529{6.4375, 312.422169552825597994104814531010579387},530{6.5000, 332.571568241777409133204438572983297292},531{6.5625, 354.020497560858198165985214519757890505},532{6.6250, 376.852769667496146326030849450983914197},533{6.6875, 401.157602161123700280816957271992998156},534{6.7500, 427.029966702886171977469256622451185850},535{6.8125, 454.570960119471524953536004647195906721},536{6.8750, 483.888199441157626584508920036981010995},537{6.9375, 515.096242417696720610477570797503766179},538{7.0000, 548.317035155212076889964120712102928484},539{7.0625, 583.680388623257719787307547662358502345},540{7.1250, 621.324485894002926216918634755431456031},541{7.1875, 661.396422095589629755266517362992812037},542{7.2500, 704.052779189542208784574955807004218856},543{7.3125, 749.460237818184878095966335081928645934},544{7.3750, 797.796228612873763671070863694973560629},545{7.4375, 849.249625508044731271830060572510241864},546{7.5000, 904.021483770216677368692292389446994987},547{7.5625, 962.325825625814651122171697031114091993},548{7.6250, 1024.390476557670599008492465853663578558},549{7.6875, 1090.457955538048482588540574008226583335},550{7.7500, 1160.786422676798661020094043586456606003},551{7.8125, 1235.650687987597295222707689125107720568},552{7.8750, 1315.343285214046776004329388551335841550},553{7.9375, 1400.175614911635999247504386054087931958},554{8.0000, 1490.479161252178088627715460421007179728},555{8.0625, 1586.606787305415349050508956232945539108},556{8.1250, 1688.934113859132470361718199038326340668},557{8.1875, 1797.860987165547537276364148450577336075},558{8.2500, 1913.813041349231764486365114317586148767},559{8.3125, 2037.243361581700856522236313401822532385},560{8.3750, 2168.634254521568851112005905503069409349},561{8.4375, 2308.499132938297821208734949028296170563},562{8.5000, 2457.384521883751693037774022640629666294},563{8.5625, 2615.872194250713123494312356053193077854},564{8.6250, 2784.581444063104750127653362960649823247},565{8.6875, 2964.171506380845754878370650565756538203},566{8.7500, 3155.344133275174556354775488913749659006},567{8.8125, 3358.846335940117183452010789979584950102},568{8.8750, 3575.473303654961482727206202358956274888},569{8.9375, 3806.071511003646460448021740303914939059},570{9.0000, 4051.542025492594047194773093534725371440},571{9.0625, 4312.844028491571841588188869958240355518},572{9.1250, 4590.998563255739769060078863130940205710},573{9.1875, 4887.092524674358252509551443117048351290},574{9.2500, 5202.282906336187674588222835339193136030},575{9.3125, 5537.801321507079474415176386655744387251},576{9.3750, 5894.958815685577062811620236195525504885},577{9.4375, 6275.150989541692149890530417987358096221},578{9.5000, 6679.863452256851081801173722051940058824},579{9.5625, 7110.677626574055535297758456126491707647},580{9.6250, 7569.276928241617224537226019600213961572},581{9.6875, 8057.453343996777301036241026375049070162},582{9.7500, 8577.114433792824387959788368429252257664},583{9.8125, 9130.290784631065880205118262838330689429},584{9.8750, 9719.143945123662919857326995631317996715},585{9.9375, 10345.974871791805753327922796701684092861},586{10.0000, 11013.232920103323139721376090437880844591},587};588589for(int i = 0; i < testCases.length; i++) {590double [] testCase = testCases[i];591failures += testCoshCaseWithUlpDiff(testCase[0],592testCase[1],5933.0);594}595596597double [][] specialTestCases = {598{0.0, 1.0},599{NaNd, NaNd},600{Double.longBitsToDouble(0x7FF0000000000001L), NaNd},601{Double.longBitsToDouble(0xFFF0000000000001L), NaNd},602{Double.longBitsToDouble(0x7FF8555555555555L), NaNd},603{Double.longBitsToDouble(0xFFF8555555555555L), NaNd},604{Double.longBitsToDouble(0x7FFFFFFFFFFFFFFFL), NaNd},605{Double.longBitsToDouble(0xFFFFFFFFFFFFFFFFL), NaNd},606{Double.longBitsToDouble(0x7FFDeadBeef00000L), NaNd},607{Double.longBitsToDouble(0xFFFDeadBeef00000L), NaNd},608{Double.longBitsToDouble(0x7FFCafeBabe00000L), NaNd},609{Double.longBitsToDouble(0xFFFCafeBabe00000L), NaNd},610{Double.POSITIVE_INFINITY, Double.POSITIVE_INFINITY}611};612613for(int i = 0; i < specialTestCases.length; i++ ) {614failures += testCoshCaseWithUlpDiff(specialTestCases[i][0],615specialTestCases[i][1],6160.0);617}618619// For powers of 2 less than 2^(-27), the second and620// subsequent terms of the Taylor series expansion will get621// rounded.622623for(int i = DoubleConsts.MIN_SUB_EXPONENT; i < -27; i++) {624double d = Math.scalb(2.0, i);625626// Result and expected are the same.627failures += testCoshCaseWithUlpDiff(d, 1.0, 2.5);628}629630// For values of x larger than 22, the e^(-x) term is631// insignificant to the floating-point result. Util exp(x)632// overflows around 709.8, cosh(x) ~= exp(x)/2; will will test633// 10000 values in this range.634635long trans22 = Double.doubleToLongBits(22.0);636// (approximately) largest value such that exp shouldn't637// overflow638long transExpOvfl = Double.doubleToLongBits(Math.nextDown(709.7827128933841));639640for(long i = trans22;641i < transExpOvfl;642i +=(transExpOvfl-trans22)/10000) {643644double d = Double.longBitsToDouble(i);645646// Allow 3.5 ulps of error to deal with error in exp.647failures += testCoshCaseWithUlpDiff(d, StrictMath.exp(d)*0.5, 3.5);648}649650// (approximately) largest value such that cosh shouldn't651// overflow.652long transCoshOvfl = Double.doubleToLongBits(710.4758600739439);653654// Make sure sinh(x) doesn't overflow as soon as exp(x)655// overflows.656657/*658* For large values of x, cosh(x) ~= 0.5*(e^x). Therefore,659*660* cosh(x) ~= e^(ln 0.5) * e^x = e^(x + ln 0.5)661*662* So, we can calculate the approximate expected result as663* exp(x + -0.693147186). However, this sum suffers from664* roundoff, limiting the accuracy of the approximation. The665* accuracy can be improved by recovering the rounded-off666* information. Since x is larger than ln(0.5), the trailing667* bits of ln(0.5) get rounded away when the two values are668* added. However, high-order bits of ln(0.5) that669* contribute to the sum can be found:670*671* offset = log(0.5);672* effective_offset = (x + offset) - x; // exact subtraction673* rounded_away_offset = offset - effective_offset; // exact subtraction674*675* Therefore, the product676*677* exp(x + offset)*exp(rounded_away_offset)678*679* will be a better approximation to the exact value of680*681* e^(x + offset)682*683* than exp(x+offset) alone. (The expected result cannot be684* computed as exp(x)*exp(offset) since exp(x) by itself would685* overflow to infinity.)686*/687double offset = StrictMath.log(0.5);688for(long i = transExpOvfl+1; i < transCoshOvfl;689i += (transCoshOvfl-transExpOvfl)/1000 ) {690double input = Double.longBitsToDouble(i);691692double expected =693StrictMath.exp(input + offset) *694StrictMath.exp( offset - ((input + offset) - input) );695696failures += testCoshCaseWithUlpDiff(input, expected, 4.0);697}698699// cosh(x) overflows for values greater than 710; in700// particular, it overflows for all 2^i, i > 10.701for(int i = 10; i <= Double.MAX_EXPONENT; i++) {702double d = Math.scalb(2.0, i);703704// Result and expected are the same.705failures += testCoshCaseWithUlpDiff(d,706Double.POSITIVE_INFINITY, 0.0);707}708return failures;709}710711public static int testCoshCaseWithTolerance(double input,712double expected,713double tolerance) {714int failures = 0;715failures += Tests.testTolerance("Math.cosh(double)",716input, Math.cosh(input),717expected, tolerance);718failures += Tests.testTolerance("Math.cosh(double)",719-input, Math.cosh(-input),720expected, tolerance);721722failures += Tests.testTolerance("StrictMath.cosh(double)",723input, StrictMath.cosh(input),724expected, tolerance);725failures += Tests.testTolerance("StrictMath.cosh(double)",726-input, StrictMath.cosh(-input),727expected, tolerance);728return failures;729}730731public static int testCoshCaseWithUlpDiff(double input,732double expected,733double ulps) {734int failures = 0;735failures += Tests.testUlpDiff("Math.cosh(double)",736input, Math.cosh(input),737expected, ulps);738failures += Tests.testUlpDiff("Math.cosh(double)",739-input, Math.cosh(-input),740expected, ulps);741742failures += Tests.testUlpDiff("StrictMath.cosh(double)",743input, StrictMath.cosh(input),744expected, ulps);745failures += Tests.testUlpDiff("StrictMath.cosh(double)",746-input, StrictMath.cosh(-input),747expected, ulps);748return failures;749}750751752/**753* Test accuracy of {Math, StrictMath}.tanh. The specified754* accuracy is 2.5 ulps.755*756* The defintion of tanh(x) is757*758* (e^x - e^(-x))/(e^x + e^(-x))759*760* The series expansion of tanh(x) =761*762* x - x^3/3 + 2x^5/15 - 17x^7/315 + ...763*764* Therefore,765*766* 1. For large values of x tanh(x) ~= signum(x)767*768* 2. For small values of x, tanh(x) ~= x.769*770* Additionally, tanh is an odd function; tanh(-x) = -tanh(x).771*772*/773static int testTanh() {774int failures = 0;775/*776* Array elements below generated using a quad sinh777* implementation. Rounded to a double, the quad result778* *should* be correctly rounded, unless we are quite unlucky.779* Assuming the quad value is a correctly rounded double, the780* allowed error is 3.0 ulps instead of 2.5 since the quad781* value rounded to double can have its own 1/2 ulp error.782*/783double [][] testCases = {784// x tanh(x)785{0.0625, 0.06241874674751251449014289119421133},786{0.1250, 0.12435300177159620805464727580589271},787{0.1875, 0.18533319990813951753211997502482787},788{0.2500, 0.24491866240370912927780113149101697},789{0.3125, 0.30270972933210848724239738970991712},790{0.3750, 0.35835739835078594631936023155315807},791{0.4375, 0.41157005567402245143207555859415687},792{0.5000, 0.46211715726000975850231848364367256},793{0.5625, 0.50982997373525658248931213507053130},794{0.6250, 0.55459972234938229399903909532308371},795{0.6875, 0.59637355547924233984437303950726939},796{0.7500, 0.63514895238728731921443435731249638},797{0.8125, 0.67096707420687367394810954721913358},798{0.8750, 0.70390560393662106058763026963135371},799{0.9375, 0.73407151960434149263991588052503660},800{1.0000, 0.76159415595576488811945828260479366},801{1.0625, 0.78661881210869761781941794647736081},802{1.1250, 0.80930107020178101206077047354332696},803{1.1875, 0.82980190998595952708572559629034476},804{1.2500, 0.84828363995751289761338764670750445},805{1.3125, 0.86490661772074179125443141102709751},806{1.3750, 0.87982669965198475596055310881018259},807{1.4375, 0.89319334040035153149249598745889365},808{1.5000, 0.90514825364486643824230369645649557},809{1.5625, 0.91582454416876231820084311814416443},810{1.6250, 0.92534622531174107960457166792300374},811{1.6875, 0.93382804322259173763570528576138652},812{1.7500, 0.94137553849728736226942088377163687},813{1.8125, 0.94808528560440629971240651310180052},814{1.8750, 0.95404526017994877009219222661968285},815{1.9375, 0.95933529331468249183399461756952555},816{2.0000, 0.96402758007581688394641372410092317},817{2.0625, 0.96818721657637057702714316097855370},818{2.1250, 0.97187274591350905151254495374870401},819{2.1875, 0.97513669829362836159665586901156483},820{2.2500, 0.97802611473881363992272924300618321},821{2.3125, 0.98058304703705186541999427134482061},822{2.3750, 0.98284502917257603002353801620158861},823{2.4375, 0.98484551746427837912703608465407824},824{2.5000, 0.98661429815143028888127603923734964},825{2.5625, 0.98817786228751240824802592958012269},826{2.6250, 0.98955974861288320579361709496051109},827{2.6875, 0.99078085564125158320311117560719312},828{2.7500, 0.99185972456820774534967078914285035},829{2.8125, 0.99281279483715982021711715899682324},830{2.8750, 0.99365463431502962099607366282699651},831{2.9375, 0.99439814606575805343721743822723671},832{3.0000, 0.99505475368673045133188018525548849},833{3.0625, 0.99563456710930963835715538507891736},834{3.1250, 0.99614653067334504917102591131792951},835{3.1875, 0.99659855517712942451966113109487039},836{3.2500, 0.99699763548652601693227592643957226},837{3.3125, 0.99734995516557367804571991063376923},838{3.3750, 0.99766097946988897037219469409451602},839{3.4375, 0.99793553792649036103161966894686844},840{3.5000, 0.99817789761119870928427335245061171},841{3.5625, 0.99839182812874152902001617480606320},842{3.6250, 0.99858065920179882368897879066418294},843{3.6875, 0.99874733168378115962760304582965538},844{3.7500, 0.99889444272615280096784208280487888},845{3.8125, 0.99902428575443546808677966295308778},846{3.8750, 0.99913888583735077016137617231569011},847{3.9375, 0.99924003097049627100651907919688313},848{4.0000, 0.99932929973906704379224334434172499},849{4.0625, 0.99940808577297384603818654530731215},850{4.1250, 0.99947761936180856115470576756499454},851{4.1875, 0.99953898655601372055527046497863955},852{4.2500, 0.99959314604388958696521068958989891},853{4.3125, 0.99964094406130644525586201091350343},854{4.3750, 0.99968312756179494813069349082306235},855{4.4375, 0.99972035584870534179601447812936151},856{4.5000, 0.99975321084802753654050617379050162},857{4.5625, 0.99978220617994689112771768489030236},858{4.6250, 0.99980779516900105210240981251048167},859{4.6875, 0.99983037791655283849546303868853396},860{4.7500, 0.99985030754497877753787358852000255},861{4.8125, 0.99986789571029070417475400133989992},862{4.8750, 0.99988341746867772271011794614780441},863{4.9375, 0.99989711557251558205051185882773206},864{5.0000, 0.99990920426259513121099044753447306},865{5.0625, 0.99991987261554158551063867262784721},866{5.1250, 0.99992928749851651137225712249720606},867{5.1875, 0.99993759617721206697530526661105307},868{5.2500, 0.99994492861777083305830639416802036},869{5.3125, 0.99995139951851344080105352145538345},870{5.3750, 0.99995711010315817210152906092289064},871{5.4375, 0.99996214970350792531554669737676253},872{5.5000, 0.99996659715630380963848952941756868},873{5.5625, 0.99997052203605101013786592945475432},874{5.6250, 0.99997398574306704793434088941484766},875{5.6875, 0.99997704246374583929961850444364696},876{5.7500, 0.99997974001803825215761760428815437},877{5.8125, 0.99998212060739040166557477723121777},878{5.8750, 0.99998422147482750993344503195672517},879{5.9375, 0.99998607548749972326220227464612338},880{6.0000, 0.99998771165079557056434885235523206},881{6.0625, 0.99998915556205996764518917496149338},882{6.1250, 0.99999042981101021976277974520745310},883{6.1875, 0.99999155433311068015449574811497719},884{6.2500, 0.99999254672143162687722782398104276},885{6.3125, 0.99999342250186907900400800240980139},886{6.3750, 0.99999419537602957780612639767025158},887{6.4375, 0.99999487743557848265406225515388994},888{6.5000, 0.99999547935140419285107893831698753},889{6.5625, 0.99999601054055694588617385671796346},890{6.6250, 0.99999647931357331502887600387959900},891{6.6875, 0.99999689300449080997594368612277442},892{6.7500, 0.99999725808558628431084200832778748},893{6.8125, 0.99999758026863294516387464046135924},894{6.8750, 0.99999786459425991170635407313276785},895{6.9375, 0.99999811551081218572759991597586905},896{7.0000, 0.99999833694394467173571641595066708},897{7.0625, 0.99999853235803894918375164252059190},898{7.1250, 0.99999870481040359014665019356422927},899{7.1875, 0.99999885699910593255108365463415411},900{7.2500, 0.99999899130518359709674536482047025},901{7.3125, 0.99999910982989611769943303422227663},902{7.3750, 0.99999921442759946591163427422888252},903{7.4375, 0.99999930673475777603853435094943258},904{7.5000, 0.99999938819554614875054970643513124},905{7.5625, 0.99999946008444508183970109263856958},906{7.6250, 0.99999952352618001331402589096040117},907{7.6875, 0.99999957951331792817413683491979752},908{7.7500, 0.99999962892179632633374697389145081},909{7.8125, 0.99999967252462750190604116210421169},910{7.8750, 0.99999971100399253750324718031574484},911{7.9375, 0.99999974496191422474977283863588658},912{8.0000, 0.99999977492967588981001883295636840},913{8.0625, 0.99999980137613348259726597081723424},914{8.1250, 0.99999982471505097353529823063673263},915{8.1875, 0.99999984531157382142423402736529911},916{8.2500, 0.99999986348794179107425910499030547},917{8.3125, 0.99999987952853049895833839645847571},918{8.3750, 0.99999989368430056302584289932834041},919{8.4375, 0.99999990617672396471542088609051728},920{8.5000, 0.99999991720124905211338798152800748},921{8.5625, 0.99999992693035839516545287745322387},922{8.6250, 0.99999993551626733394129009365703767},923{8.6875, 0.99999994309330543951799157347876934},924{8.7500, 0.99999994978001814614368429416607424},925{8.8125, 0.99999995568102143535399207289008504},926{8.8750, 0.99999996088863858914831986187674522},927{8.9375, 0.99999996548434461974481685677429908},928{9.0000, 0.99999996954004097447930211118358244},929{9.0625, 0.99999997311918045901919121395899372},930{9.1250, 0.99999997627775997868467948564005257},931{9.1875, 0.99999997906519662964368381583648379},932{9.2500, 0.99999998152510084671976114264303159},933{9.3125, 0.99999998369595870397054673668361266},934{9.3750, 0.99999998561173404286033236040150950},935{9.4375, 0.99999998730239984852716512979473289},936{9.5000, 0.99999998879440718770812040917618843},937{9.5625, 0.99999999011109904501789298212541698},938{9.6250, 0.99999999127307553219220251303121960},939{9.6875, 0.99999999229851618412119275358396363},940{9.7500, 0.99999999320346438410630581726217930},941{9.8125, 0.99999999400207836827291739324060736},942{9.8750, 0.99999999470685273619047001387577653},943{9.9375, 0.99999999532881393331131526966058758},944{10.0000, 0.99999999587769276361959283713827574},945};946947for(int i = 0; i < testCases.length; i++) {948double [] testCase = testCases[i];949failures += testTanhCaseWithUlpDiff(testCase[0],950testCase[1],9513.0);952}953954955double [][] specialTestCases = {956{0.0, 0.0},957{NaNd, NaNd},958{Double.longBitsToDouble(0x7FF0000000000001L), NaNd},959{Double.longBitsToDouble(0xFFF0000000000001L), NaNd},960{Double.longBitsToDouble(0x7FF8555555555555L), NaNd},961{Double.longBitsToDouble(0xFFF8555555555555L), NaNd},962{Double.longBitsToDouble(0x7FFFFFFFFFFFFFFFL), NaNd},963{Double.longBitsToDouble(0xFFFFFFFFFFFFFFFFL), NaNd},964{Double.longBitsToDouble(0x7FFDeadBeef00000L), NaNd},965{Double.longBitsToDouble(0xFFFDeadBeef00000L), NaNd},966{Double.longBitsToDouble(0x7FFCafeBabe00000L), NaNd},967{Double.longBitsToDouble(0xFFFCafeBabe00000L), NaNd},968{Double.POSITIVE_INFINITY, 1.0}969};970971for(int i = 0; i < specialTestCases.length; i++) {972failures += testTanhCaseWithUlpDiff(specialTestCases[i][0],973specialTestCases[i][1],9740.0);975}976977// For powers of 2 less than 2^(-27), the second and978// subsequent terms of the Taylor series expansion will get979// rounded away since |n-n^3| > 53, the binary precision of a980// double significand.981982for(int i = DoubleConsts.MIN_SUB_EXPONENT; i < -27; i++) {983double d = Math.scalb(2.0, i);984985// Result and expected are the same.986failures += testTanhCaseWithUlpDiff(d, d, 2.5);987}988989// For values of x larger than 22, tanh(x) is 1.0 in double990// floating-point arithmetic.991992for(int i = 22; i < 32; i++) {993failures += testTanhCaseWithUlpDiff(i, 1.0, 2.5);994}995996for(int i = 5; i <= Double.MAX_EXPONENT; i++) {997double d = Math.scalb(2.0, i);998999failures += testTanhCaseWithUlpDiff(d, 1.0, 2.5);1000}10011002return failures;1003}10041005public static int testTanhCaseWithTolerance(double input,1006double expected,1007double tolerance) {1008int failures = 0;1009failures += Tests.testTolerance("Math.tanh(double",1010input, Math.tanh(input),1011expected, tolerance);1012failures += Tests.testTolerance("Math.tanh(double",1013-input, Math.tanh(-input),1014-expected, tolerance);10151016failures += Tests.testTolerance("StrictMath.tanh(double",1017input, StrictMath.tanh(input),1018expected, tolerance);1019failures += Tests.testTolerance("StrictMath.tanh(double",1020-input, StrictMath.tanh(-input),1021-expected, tolerance);1022return failures;1023}10241025public static int testTanhCaseWithUlpDiff(double input,1026double expected,1027double ulps) {1028int failures = 0;10291030failures += Tests.testUlpDiffWithAbsBound("Math.tanh(double)",1031input, Math.tanh(input),1032expected, ulps, 1.0);1033failures += Tests.testUlpDiffWithAbsBound("Math.tanh(double)",1034-input, Math.tanh(-input),1035-expected, ulps, 1.0);10361037failures += Tests.testUlpDiffWithAbsBound("StrictMath.tanh(double)",1038input, StrictMath.tanh(input),1039expected, ulps, 1.0);1040failures += Tests.testUlpDiffWithAbsBound("StrictMath.tanh(double)",1041-input, StrictMath.tanh(-input),1042-expected, ulps, 1.0);1043return failures;1044}104510461047public static void main(String argv[]) {1048int failures = 0;10491050failures += testSinh();1051failures += testCosh();1052failures += testTanh();10531054if (failures > 0) {1055System.err.println("Testing the hyperbolic functions incurred "1056+ failures + " failures.");1057throw new RuntimeException();1058}1059}10601061}106210631064