inline qreal qFastSin(qreal x)
{
int si = int(x * (0.5 * QT_SINE_TABLE_SIZE / M_PI)); // Would be more accurate with qRound, but slower.
qreal d = x - si * (2.0 * M_PI / QT_SINE_TABLE_SIZE);
int ci = si + QT_SINE_TABLE_SIZE / 4;
si &= QT_SINE_TABLE_SIZE - 1;
ci &= QT_SINE_TABLE_SIZE - 1;
return qt_sine_table[si] + (qt_sine_table[ci] - 0.5 * qt_sine_table[si] * d) * d;
}
这个qt_sine_table是一张长度为QT_SINE_TABLE_SIZE(256)的实数表。
const qreal qt_sine_table[QT_SINE_TABLE_SIZE] = {
qreal(0.0),
qreal(0.024541228522912288),
qreal(0.049067674327418015),
qreal(0.073564563599667426),
qreal(0.098017140329560604),
qreal(0.1224106751992162),
qreal(0.14673047445536175),
qreal(0.17096188876030122),
qreal(0.19509032201612825),
qreal(0.2191012401568698),
qreal(0.24298017990326387),
qreal(0.26671275747489837),
qreal(0.29028467725446233),
qreal(0.31368174039889152),
qreal(0.33688985339222005),
qreal(0.35989503653498811),
qreal(0.38268343236508978),
qreal(0.40524131400498986),
qreal(0.42755509343028208),
qreal(0.44961132965460654),
qreal(0.47139673682599764),
qreal(0.49289819222978404),
qreal(0.51410274419322166),
qreal(0.53499761988709715),
qreal(0.55557023301960218),
qreal(0.57580819141784534),
qreal(0.59569930449243336),
qreal(0.61523159058062682),
qreal(0.63439328416364549),
qreal(0.65317284295377676),
qreal(0.67155895484701833),
qreal(0.68954054473706683),
qreal(0.70710678118654746),
qreal(0.72424708295146689),
qreal(0.74095112535495911),
qreal(0.75720884650648446),
qreal(0.77301045336273699),
qreal(0.78834642762660623),
qreal(0.80320753148064483),
qreal(0.81758481315158371),
qreal(0.83146961230254524),
qreal(0.84485356524970701),
qreal(0.85772861000027212),
qreal(0.87008699110871135),
qreal(0.88192126434835494),
qreal(0.89322430119551532),
qreal(0.90398929312344334),
qreal(0.91420975570353069),
qreal(0.92387953251128674),
qreal(0.93299279883473885),
qreal(0.94154406518302081),
qreal(0.94952818059303667),
qreal(0.95694033573220894),
qreal(0.96377606579543984),
qreal(0.97003125319454397),
qreal(0.97570213003852857),
qreal(0.98078528040323043),
qreal(0.98527764238894122),
qreal(0.98917650996478101),
qreal(0.99247953459870997),
qreal(0.99518472667219682),
qreal(0.99729045667869021),
qreal(0.99879545620517241),
qreal(0.99969881869620425),
qreal(1.0),
qreal(0.99969881869620425),
qreal(0.99879545620517241),
qreal(0.99729045667869021),
qreal(0.99518472667219693),
qreal(0.99247953459870997),
qreal(0.98917650996478101),
qreal(0.98527764238894122),
qreal(0.98078528040323043),
qreal(0.97570213003852857),
qreal(0.97003125319454397),
qreal(0.96377606579543984),
qreal(0.95694033573220894),
qreal(0.94952818059303667),
qreal(0.94154406518302081),
qreal(0.93299279883473885),
qreal(0.92387953251128674),
qreal(0.91420975570353069),
qreal(0.90398929312344345),
qreal(0.89322430119551521),
qreal(0.88192126434835505),
qreal(0.87008699110871146),
qreal(0.85772861000027212),
qreal(0.84485356524970723),
qreal(0.83146961230254546),
qreal(0.81758481315158371),
qreal(0.80320753148064494),
qreal(0.78834642762660634),
qreal(0.7730104533627371),
qreal(0.75720884650648468),
qreal(0.74095112535495899),
qreal(0.72424708295146689),
qreal(0.70710678118654757),
qreal(0.68954054473706705),
qreal(0.67155895484701855),
qreal(0.65317284295377664),
qreal(0.63439328416364549),
qreal(0.61523159058062693),
qreal(0.59569930449243347),
qreal(0.57580819141784545),
qreal(0.55557023301960218),
qreal(0.53499761988709715),
qreal(0.51410274419322177),
qreal(0.49289819222978415),
qreal(0.47139673682599786),
qreal(0.44961132965460687),
qreal(0.42755509343028203),
qreal(0.40524131400498992),
qreal(0.38268343236508989),
qreal(0.35989503653498833),
qreal(0.33688985339222033),
qreal(0.31368174039889141),
qreal(0.29028467725446239),
qreal(0.26671275747489848),
qreal(0.24298017990326407),
qreal(0.21910124015687005),
qreal(0.19509032201612861),
qreal(0.17096188876030122),
qreal(0.1467304744553618),
qreal(0.12241067519921635),
qreal(0.098017140329560826),
qreal(0.073564563599667732),
qreal(0.049067674327417966),
qreal(0.024541228522912326),
qreal(0.0),
qreal(-0.02454122852291208),
qreal(-0.049067674327417724),
qreal(-0.073564563599667496),
qreal(-0.09801714032956059),
qreal(-0.1224106751992161),
qreal(-0.14673047445536158),
qreal(-0.17096188876030097),
qreal(-0.19509032201612836),
qreal(-0.2191012401568698),
qreal(-0.24298017990326382),
qreal(-0.26671275747489825),
qreal(-0.29028467725446211),
qreal(-0.31368174039889118),
qreal(-0.33688985339222011),
qreal(-0.35989503653498811),
qreal(-0.38268343236508967),
qreal(-0.40524131400498969),
qreal(-0.42755509343028181),
qreal(-0.44961132965460665),
qreal(-0.47139673682599764),
qreal(-0.49289819222978393),
qreal(-0.51410274419322155),
qreal(-0.53499761988709693),
qreal(-0.55557023301960196),
qreal(-0.57580819141784534),
qreal(-0.59569930449243325),
qreal(-0.61523159058062671),
qreal(-0.63439328416364527),
qreal(-0.65317284295377653),
qreal(-0.67155895484701844),
qreal(-0.68954054473706683),
qreal(-0.70710678118654746),
qreal(-0.72424708295146678),
qreal(-0.74095112535495888),
qreal(-0.75720884650648423),
qreal(-0.77301045336273666),
qreal(-0.78834642762660589),
qreal(-0.80320753148064505),
qreal(-0.81758481315158382),
qreal(-0.83146961230254524),
qreal(-0.84485356524970701),
qreal(-0.85772861000027201),
qreal(-0.87008699110871135),
qreal(-0.88192126434835494),
qreal(-0.89322430119551521),
qreal(-0.90398929312344312),
qreal(-0.91420975570353047),
qreal(-0.92387953251128652),
qreal(-0.93299279883473896),
qreal(-0.94154406518302081),
qreal(-0.94952818059303667),
qreal(-0.95694033573220882),
qreal(-0.96377606579543984),
qreal(-0.97003125319454397),
qreal(-0.97570213003852846),
qreal(-0.98078528040323032),
qreal(-0.98527764238894111),
qreal(-0.9891765099647809),
qreal(-0.99247953459871008),
qreal(-0.99518472667219693),
qreal(-0.99729045667869021),
qreal(-0.99879545620517241),
qreal(-0.99969881869620425),
qreal(-1.0),
qreal(-0.99969881869620425),
qreal(-0.99879545620517241),
qreal(-0.99729045667869021),
qreal(-0.99518472667219693),
qreal(-0.99247953459871008),
qreal(-0.9891765099647809),
qreal(-0.98527764238894122),
qreal(-0.98078528040323043),
qreal(-0.97570213003852857),
qreal(-0.97003125319454397),
qreal(-0.96377606579543995),
qreal(-0.95694033573220894),
qreal(-0.94952818059303679),
qreal(-0.94154406518302092),
qreal(-0.93299279883473907),
qreal(-0.92387953251128663),
qreal(-0.91420975570353058),
qreal(-0.90398929312344334),
qreal(-0.89322430119551532),
qreal(-0.88192126434835505),
qreal(-0.87008699110871146),
qreal(-0.85772861000027223),
qreal(-0.84485356524970723),
qreal(-0.83146961230254546),
qreal(-0.81758481315158404),
qreal(-0.80320753148064528),
qreal(-0.78834642762660612),
qreal(-0.77301045336273688),
qreal(-0.75720884650648457),
qreal(-0.74095112535495911),
qreal(-0.724247082951467),
qreal(-0.70710678118654768),
qreal(-0.68954054473706716),
qreal(-0.67155895484701866),
qreal(-0.65317284295377709),
qreal(-0.63439328416364593),
qreal(-0.61523159058062737),
qreal(-0.59569930449243325),
qreal(-0.57580819141784523),
qreal(-0.55557023301960218),
qreal(-0.53499761988709726),
qreal(-0.51410274419322188),
qreal(-0.49289819222978426),
qreal(-0.47139673682599792),
qreal(-0.44961132965460698),
qreal(-0.42755509343028253),
qreal(-0.40524131400499042),
qreal(-0.38268343236509039),
qreal(-0.359895036534988),
qreal(-0.33688985339222),
qreal(-0.31368174039889152),
qreal(-0.2902846772544625),
qreal(-0.26671275747489859),
qreal(-0.24298017990326418),
qreal(-0.21910124015687016),
qreal(-0.19509032201612872),
qreal(-0.17096188876030177),
qreal(-0.14673047445536239),
qreal(-0.12241067519921603),
qreal(-0.098017140329560506),
qreal(-0.073564563599667412),
qreal(-0.049067674327418091),
qreal(-0.024541228522912448)
};这段代码看意思应该是近似计算会很快，但是没看懂具体是为什么可以这么算，文档也没有关于这个函数的介绍……求指点一波……
怪我咯2017-04-17 13:11:46
This is a typical table lookup method.
To put it simply, the continuous range of 0 ~ 2 PI is simplified into 256 discrete values. I found a schematic picture (note, Just a hint):
Calculate the values of these discrete points in advance, which are the values in the table you listed.
Then, decompose x into two parts. The Fix part (si) directly looks up the table to get the value. The other floating point part is more troublesome and requires interpolation calculation.
For specific steps, please refer to: https://en.wikipedia.org/wiki/Lookup_table#Computing_sines
ci is actually si + PI / 2, qt_sine_table[ci] is just cos(si).
si &= QT_SINE_TABLE_SIZE - 1;
ci &= QT_SINE_TABLE_SIZE - 1;These two sentences are mainly for correction, converting values outside the range of 0 ~ 256 into this range, using the periodicity of sin.
This algorithm is very mature, you can refer to this explanation on SO: http://stackoverflow.com/questions/3688649/create-sine-lookup-table-in-c.
For a more accurate method, I think we should use Taylor series expansion to solve the problem.
