Sigmoid function

1 are also common in statistics as cumulative distribution

functions (which go from 0 to 1), such as the integrals
of the logistic distribution, the normal distribution, and
Students t probability density functions.

0.5

1 Denition
A sigmoid function is a bounded dierentiable real func-
0 tion that is dened for all real input values and has a non-
6 4 2 0 2 4 6 negative derivative at each point.[1]

The logistic curve

2 Properties
1.00
In general, a sigmoid function is real-valued, monotonic,
0.75 and dierentiable having a non-negative rst derivative
0.50 which is bell shaped. A sigmoid function is constrained
0.25 by a pair of horizontal asymptotes as x .
erf (x)

0.00

0.25

0.50
3 Examples
0.75

1.00

3 2 1 0 1 2 3
x

Plot of the error function

A sigmoid function is a mathematical function having a

characteristic S"-shaped curve or sigmoid curve. Of-
ten, sigmoid function refers to the special case of the
logistic function shown in the rst gure and dened by
the formula Some sigmoid functions compared. In the drawing all functions
are normalized in such a way that their slope at the origin is 1.

1
S(x) = . Logistic function
1 + ex
Other examples of similar shapes include the Gompertz
curve (used in modeling systems that saturate at large
values of x) and the ogee curve (used in the spillway of 1
f (x) =
some dams). Sigmoid functions have have domain of all 1 + ex
real numbers, with return value monotonically increasing
most often from 0 to 1 or alternatively from 1 to 1, de- hyperbolic tangent
pending on convention.
A wide variety of sigmoid functions have been used as
the activation function of articial neurons, including the ex ex
logistic and hyperbolic tangent functions. Sigmoid curves f (x) = tanh x =
ex + ex

1
2 5 REFERENCES

arctangent function Cumulative distribution function

Generalized logistic curve

Gompertz function
f (x) = arctan x
Heaviside step function
Gudermannian function Hyperbolic function

Logistic distribution
x Logistic function
1
f (x) = gd(x) = dt
0 cosh t Logistic regression

Error function Logit

Modied hyperbolic tangent

x Softplus function
2
et dt
2
f (x) = erf(x) =
0 Smoothstep function (Graphics)

Softmax function
Generalised logistic function
Weibull distribution

Netoid function
f (x) = (1 + ex ) , >0

Smoothstep function 5 References

[1] Han, Jun; Morag, Claudio (1995). The inuence of the
sigmoid function parameters on the speed of backprop-
( )1 (
1( )N x )N agation learning. In Mira, Jos; Sandoval, Francisco.
1 u2 du 1 u2 du |x| From
1 Natural
f (x) = 0 0 N 1 to Articial Neural Computation. pp. 195
sgn(x) |x| 201.
1
[2] Gibbs, M.N. (Nov 2000). Variational Gaussian process
Specic algebraic functions classiers. IEEE Transactions on Neural Networks. 11
(6): 14581464. doi:10.1109/72.883477.

Mitchell, Tom M. (1997). Machine Learning.

x
f (x) =
1 + x2
The integral of any continuous, non-negative, bump-
shaped function will be sigmoidal, thus the cumulative
distribution functions for many common probability dis-
tributions are sigmoidal. One such example is the error
function, which is related to the cumulative distribution
function (CDF) of a normal distribution.
Many natural processes, such as those of complex system
learning curves, exhibit a progression from small begin-
nings that accelerates and approaches a climax over time.
When a specic mathematical model is lacking, a sigmoid
function is often used.[2]
WCBMcGrawHill. ISBN 0-07-042807-7.. In
particular see Chapter 4: Articial Neural Net-
works (in particular pp. 9697) where Mitchell
uses the word logistic function and the sigmoid
function synonymously this function he also calls
the squashing function and the sigmoid (aka lo-
gistic) function is used to compress the outputs of
the neurons in multi-layer neural nets.
Humphrys, Mark. Continuous output, the sigmoid
function. Properties of the sigmoid, including how
it can shift along axes and how its domain may be
transformed.

4 See also
Activation function
 3

6 Text and image sources, contributors, and licenses

6.1 Text
Sigmoid function Source: https://en.wikipedia.org/wiki/Sigmoid_function?oldid=781040006 Contributors: The Anome, Michael Hardy,
Chris-martin, Kku, Chinju, Glenn, AnthonyQBachler, Pmcray, Tea2min, Giftlite, BenFrantzDale, Chinasaur, Jorge Stol, Mboverload, Ee-
quor, MarkSweep, Cacycle, Bender235, ZeroOne, Wolfman, Linas, Jacobolus, Doubleddoubleu, GregorB, Waldir, Wdanwatts, Melesse,
Strait, HappyCamper, Kri, Roboto de Ajvol, YurikBot, Cleared as led, Daniel Mietchen, Light current, Closedmouth, Cmglee, TravisTX,
SmackBot, Jfmiller28, CapitalSasha, KennethJ, ZackV, Nbarth, Michael.Pohoreski, Georg-Johann, Wvbailey, Dr. Sunglasses, Sbmehta,
Freewol, Tktktk, Jim.belk, JHunterJ, Knights who say ni, Hyperwiz, A. Pichler, Trialsanderrors, Pfhenshaw, Schroding79, Thijs!bot, Mkch,
Wasell, Paul Haymon, Pebkac, Loluengo, Daniel5Ko, Akeron, Barraki, Red Act, Technopat, Hagman, SieBot, Thelostchild, ShadowPhox,
Mike2vil, Dolphin51, Addbot, MrOllie, New Image Uploader 929, Cesiumfrog, Ale66, Luckas-bot, AnomieBOT, Csigabi, Materialsci-
entist, LilHelpa, Xqbot, Isheden, Sawomir Biay, John85, Tom.Reding, Vjost, Inferior Olive, Helwr, EmausBot, Slawekb, ZroBot, Mr-
matiko, Glosser.ca, ClueBot NG, KlappCK, Marcocapelle, Manish Singh321, ChrisGualtieri, Elen so, Cerabot~enwiki, Defpo, Monkbot,
Lolo often, ZxzStar, Deacon Vorbis and Anonymous: 73

6.2 Images
File:Commons-logo.svg Source: https://upload.wikimedia.org/wikipedia/en/4/4a/Commons-logo.svg License: PD Contributors: ? Origi-
nal artist: ?
File:Error_Function.svg Source: https://upload.wikimedia.org/wikipedia/commons/2/2f/Error_Function.svg License: Public domain
Contributors: self-made, Inkscape Original artist: Inductiveload
File:Gjl-t(x).svg Source: https://upload.wikimedia.org/wikipedia/commons/6/6f/Gjl-t%28x%29.svg License: CC BY-SA 3.0 Con-
tributors: This vector image includes elements that have been taken or adapted from this: <a href='//commons.wikimedia.
org/wiki/File:Error_function.svg' class='image'><img alt='Error function.svg' src='https://upload.wikimedia.org/wikipedia/commons/
thumb/8/81/Error_function.svg/40px-Error_function.svg.png' width='40' height='20' srcset='https://upload.wikimedia.org/wikipedia/
commons/thumb/8/81/Error_function.svg/60px-Error_function.svg.png 1.5x, https://upload.wikimedia.org/wikipedia/commons/thumb/
8/81/Error_function.svg/80px-Error_function.svg.png 2x' data-le-width='480' data-le-height='240' /></a> Error function.svg (by
Geek3). Original artist: Georg-Johann
File:Logistic-curve.svg Source: https://upload.wikimedia.org/wikipedia/commons/8/88/Logistic-curve.svg License: Public domain Con-
tributors: Created from scratch with gnuplot Original artist: Qef (talk)
File:Question_book-new.svg Source: https://upload.wikimedia.org/wikipedia/en/9/99/Question_book-new.svg License: Cc-by-sa-3.0
Contributors:
Created from scratch in Adobe Illustrator. Based on Image:Question book.png created by User:Equazcion Original artist:
Tkgd2007

6.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Logistic function

For the recurrence relation, see Logistic map. 1 Mathematical properties

A logistic function or logistic curve is a common S
The standard logistic function is the logistic function with
1 parameters (k = 1, x0 = 0, L = 1) which yields

1
f (x) = .
1 + ex
0.5
In practice, due to the nature of the exponential function
ex , it is often sucient to compute the standard logistic
function for x over a small range of real numbers such as
a range contained in [6, +6].
0
6 4 2 0 2 4 6
1.1 Derivative
Standard logistic sigmoid function i.e. L = 1, k = 1, x0 = 0
The standard logistic function has an easily calculated
shape (sigmoid curve), with equation: derivative:
1 ex
f (x) = 1+ex = 1+ex
d ex (1+ex )ex ex
L dx f (x) = (1+ex )2
f (x) =
1 + ek(xx0 ) ex
d
dx f (x) = (1+ex )2 = f (x)(1 f (x))
where The logistic function also has the property that:

e = the natural logarithm base (also known as Eulers

number),
x0 = the x-value of the sigmoids midpoint,
L = the curves maximum value, and
k = the steepness of the curve.[1]
1 f (x) = f (x).
Thus, x 7 f (x) 1/2 is an odd function.
The derivative of the logistic function has the property
that:

d d
For values of x in the range of real numbers from f (x) = f (x).
dx dx
to +, the S-curve shown on the right is obtained (with
the graph of f approaching L as x approaches + and
approaching zero as x approaches ). 1.2 Logistic dierential equation
The function was named in 18441845 by Pierre
The standard logistic function is the solution of the simple
Franois Verhulst, who studied it in relation to population
rst-order non-linear ordinary dierential equation
growth.[2] The initial stage of growth is approximately
exponential; then, as saturation begins, the growth slows,
and at maturity, growth stops. d
f (x) = f (x)(1 f (x))
The logistic function nds applications in a range of dx
elds, including articial neural networks, biology
(especially ecology), biomathematics, chemistry, with boundary condition f(0) = 1/2. This equation is the
demography, economics, geoscience, mathematical continuous version of the logistic map.
psychology, probability, sociology, political science, The qualitative behavior is easily understood in terms of
linguistics, and statistics. the phase line: the derivative is null when function is unit

1
2 2 APPLICATIONS

and the derivative is positive for f between 0 and 1, and

negative for f above 1 or less than 0 (though negative pop- ( )
ulations do not generally accord with a physical model). ex ex ex 1 e2x
tanh(x) = x = x
This yields an unstable equilibrium at 0, and a stable equi- e + ex e (1 + e2x )
librium at 1, and thus for any function value greater than
e2x e2x + 1 1
zero and less than unit, it grows to unit. = f (2x) 2x
= f (2x) = 2f (2x) 1.
1+e 1 + e2x
The above equation can be rewritten in the following
steps: The hyperbolic tangent relationship leads to another form
for the logistic functions derivative:

d
f (x) = f (x)(1 f (x)) d 1 ( )
dx f (x) = sech2 x2 ,
dx 4
dy
= y(1 y) which ties the logistic function into the logistic distribu-
dx
tion.
dy
= y y2
dx
dy
y = y 2
1.3 Rotational symmetry about (0, )
dx
Which is a special case of the Bernoulli dierential equa- The sum of the logistic function and its reection about
tion and has the following solution: the vertical axis, f (x) is

ex 1 1 (1 + ex ) + (1 + ex ) 2 + ex + ex
f (x) = + = =
ex + C 1 + ex 1 + e(x) (1 + ex )(1 + ex ) 1 + ex + ex + exx
Choosing the constant of integration C = 1 gives the The logistic function is thus rotationally symmetrical
other well-known form of the denition of the logistic about the point (0, 1/2).[4]
curve

ex 1
2 Applications
f (x) = =
x
e +1 1 + ex
2.1 In ecology: modeling population
More quantitatively, as can be seen from the analytical so-
growth
lution, the logistic curve shows early exponential growth
for negative argument, which slows to linear growth of
A typical application of the logistic equation is a common
slope 1/4 for an argument near zero, then approaches one
model of population growth (see also population dynam-
with an exponentially decaying gap.
ics), originally due to Pierre-Franois Verhulst in 1838,
The logistic function is the inverse of the natural logit where the rate of reproduction is proportional to both
function and so can be used to convert the logarithm of the existing population and the amount of available re-
odds into a probability. In mathematical notation the lo- sources, all else being equal. The Verhulst equation was
gistic function is sometimes written as expit [3] in the same published after Verhulst had read Thomas Malthus' An
form as logit. The conversion from the log-likelihood ra- Essay on the Principle of Population. Verhulst derived his
tio of two alternatives also takes the form of a logistic logistic equation to describe the self-limiting growth of
curve. a biological population. The equation was rediscovered
The logistic sigmoid function is related to the hyperbolic in 1911 by A. G. McKendrick for the growth of bacteria
tangent, A.p. by in broth and experimentally tested using a technique for
nonlinear parameter estimation.[5] The equation is also
sometimes called the Verhulst-Pearl equation following
(x) its rediscovery in 1920 by Raymond Pearl (18791940)
2 f (x) = 1 + tanh and Lowell Reed (18881966) of the Johns Hopkins Uni-
2
or versity.[6] Another scientist, Alfred J. Lotka derived the
equation again in 1925, calling it the law of population
growth.
tanh(x) = 2f (2x) 1 Letting P represent population size (N is often used in
ecology instead) and t represent time, this model is for-
The latter relationship follows from malized by the dierential equation:
2.2 In statistics and machine learning 3

lim P (t) = K
t

Which is to say that K is the limiting value of P: the high-

est value that the population can reach given innite time
(or come close to reaching in nite time). It is impor-
tant to stress that the carrying capacity is asymptotically
reached independently of the initial value P(0) > 0, and
also in the case that P(0) > K.
In ecology, species are sometimes referred to as r-
strategist or K-strategist depending upon the selective
processes that have shaped their life history strategies.
Choosing the variable dimensions so that n measures the
population in units of carrying capacity, and measures
time in units of 1/r, gives the dimensionless dierential
equation

dn
= n(1 n)
d

2.1.1 Time-varying carrying capacity

Since the environmental conditions inuence the carrying

capacity, as a consequence it can be time-varying: K(t) >
0, leading to the following mathematical model:

( )
Pierre-Franois Verhulst (18041849) dP P
= rP 1
dt K(t)
A particularly important case is that of carrying capacity
that varies periodically with period T:
( )
dP P
= rP 1
dt K
K(t + T ) = K(t)
where the constant r denes the growth rate and K is the It can be shown that in such a case, independently from
carrying capacity. the initial value P(0) > 0, P(t) will tend to a unique peri-
In the equation, the early, unimpeded growth rate is mod- odic solution P*(t), whose period is T.
eled by the rst term +rP. The value of the rate r repre- A typical value of T is one year: In such case K(t) may
sents the proportional increase of the population P in one reect periodical variations of weather conditions.
unit of time. Later, as the population grows, the modulus
of the second term (which multiplied out is rP 2 /K) be- Another interesting generalization is to consider that the
comes almost as large as the rst, as some members of the carrying capacity K(t) is a function of the population at
population P interfere with each other by competing for an earlier time, capturing a delay in the way population
some critical resource, such as food or living space. This modies [7] its environment. This leads to a logistic delay
antagonistic eect is called the bottleneck, and is mod- equation, which has a very rich behavior, with bistabil-
eled by the value of the parameter K. The competition ity in some parameter range, as well as a monotonic decay
diminishes the combined growth rate, until the value of to zero, smooth exponential growth, punctuated unlim-
P ceases to grow (this is called maturity of the popula- ited growth (i.e., multiple S-shapes), punctuated growth
tion). The solution to the equation (with P0 being the or alternation to a stationary level, oscillatory approach
initial population) is to a stationary level, sustainable oscillations, nite-time
singularities as well as nite-time death.

KP0 ert 2.2 In statistics and machine learning

P (t) =
K + P0 (ert 1)
Logistic functions are used in several roles in statistics.
where For example, they are the cumulative distribution func-
4 2 APPLICATIONS

tion of the logistic family of distributions, and they are, The logistic function is itself the derivative of another
a bit simplied, used to model the chance a chess player proposed activation function, the softplus.
has to beat his opponent in the Elo rating system. More
specic examples now follow.
2.3 In medicine: modeling of growth of tu-
mors
2.2.1 Logistic regression
See also: Gompertz curve Growth of tumors
Main article: Logistic regression

Another application of logistic curve is in medicine,

Logistic functions are used in logistic regression to model
how the probability p of an event may be aected by one
or more explanatory variables: an example would be to
have the model
p = f (a + bx)
where x is the explanatory variable and a and b are model
parameters to be tted and f is the standard logistic func-
tion.
Logistic regression and other log-linear models are also
commonly used in machine learning. A generalisation of
the logistic function to multiple inputs is the softmax ac- X = F (X)X,
tivation function, used in multinomial logistic regression.
Another application of the logistic function is in the
Rasch model, used in item response theory. In partic-
ular, the Rasch model forms a basis for maximum likeli-
hood estimation of the locations of objects or persons on
a continuum, based on collections of categorical data, for
example the abilities of persons on a continuum based on X = r 1 K X c(t)X,
responses that have been categorized as correct and in-
correct.
2.2.2 Neural networks
where the logistic dierential equation is used to model
the growth of tumors. This application can be considered
an extension of the above-mentioned use in the frame-
work of ecology (see also the Generalized logistic curve,
allowing for more parameters). Denoting with X(t) the
size of the tumor at time t, its dynamics are governed by:
( )
X
X = r 1 X
K
which is of the type:
F (X) 0
where F(X) is the proliferation rate of the tumor.
If a chemotherapy is started with a log-kill eect, the
equation may be revised to be
( )
X
where c(t) is the therapy-induced death rate. In the ide-
alized case of very long therapy, c(t) can be modeled as a
periodic function (of period T) or (in case of continuous
infusion therapy) as a constant function, and one has that

Logistic functions are often used in neural networks to

introduce nonlinearity in the model and/or to clamp sig- 1 T
c(t) dt > r lim x(t) = 0
nals to within a specied range. A popular neural net el- T 0 t+
ement computes a linear combination of its input signals,
i.e. if the average therapy-induced death rate is greater
and applies a bounded logistic function to the result; this
than the baseline proliferation rate then there is the erad-
model can be seen as a smoothed variant of the classical
ication of the disease. Of course, this is an oversimplied
threshold neuron.
model of both the growth and the therapy (e.g. it does not
A common choice for the activation or squashing func- take into account the phenomenon of clonal resistance).
tions, used to clip for large magnitudes to keep the re-
sponse of the neural network bounded[8] is
2.4 In chemistry: reaction models

1 The concentration of reactants and products in

g(h) =
1 + e2h
which is a logistic function. These relationships result in
simplied implementations of articial neural networks
with articial neurons. Practitioners caution that sig-
moidal functions which are antisymmetric about the ori-
gin (e.g. the hyperbolic tangent) lead to faster conver-
gence when training networks with backpropagation.[9]
autocatalytic reactions follow the logistic function.
2.5 In physics: Fermi distribution
The logistic function determines the statistical distribu-
tion of fermions over the energy states of a system in ther-
mal equilibrium. In particular, it is the distribution of the
 5

probabilities that each possible energy level is occupied frenzy, the rapid build out as synergy and the completion
by a fermion, according to FermiDirac statistics. as maturity.[15]

2.6 In linguistics: language change 3 See also

In linguistics, the logistic function can be used to model
language change:[10] an innovation that is at rst marginal Diusion of innovations
begins to spread more quickly with time, and then more Generalised logistic curve
slowly as it becomes more universally adopted.
Gompertz curve

2.7 In economics and sociology: diusion Heaviside step function

of innovations Hubbert curve
The logistic function can be used to illustrate the progress Logistic distribution
of the diusion of an innovation through its life cycle.
Logistic map
In The Laws of Imitation (1890), Gabriel Tarde describes
the rise and spread of new ideas through imitative chains. Logistic regression
In particular, Tarde identies three main stages through
which innovations spread: the rst one corresponds to the Logistic smooth-transmission model
dicult beginnings, during which the idea has to strug-
Logit
gle within a hostile environment full of opposing habits
and beliefs; the second one corresponds to the properly Log-likelihood ratio
exponential take-o of the idea, with f (x) = 2x ; -
nally, the third stage is logarithmic, with f (x) = log(x) Malthusian growth model
, and corresponds to the time when the impulse of the
idea gradually slows down while, simultaneously new op- Population dynamics
ponent ideas appear. The ensuing situation halts or stabi- r/K selection theory
lizes the progress of the innovation, which approaches an
asymptote. Shifted Gompertz distribution
In the history of economy, when new products are intro- Tipping point (sociology)
duced there is an intense amount of research and devel-
opment which leads to dramatic improvements in quality Rectier (neural networks)
and reductions in cost. This leads to a period of rapid
industry growth. Some of the more famous examples
are: railroads, incandescent light bulbs, electrication, 4 Notes
cars and air travel. Eventually, dramatic improvement
and cost reduction opportunities are exhausted, the prod-
[1] Verhulst, Pierre-Franois (1838). Notice sur la loi que la
uct or process are in widespread use with few remaining population poursuit dans son accroissement (PDF). Cor-
potential new customers, and markets become saturated. respondance mathmatique et physique. 10: 113121. Re-
Logistic analysis was used in papers by several re- trieved 3 December 2014.
searchers at the International Institute of Applied Sys- [2] Verhulst, Pierre-Franois (1845). Recherches mathma-
tems Analysis (IIASA). These papers deal with the dif- tiques sur la loi d'accroissement de la population [Math-
fusion of various innovations, infrastructures and energy ematical Researches into the Law of Population Growth
source substitutions and the role of work in the economy Increase]. Nouveaux Mmoires de l'Acadmie Royale des
as well as with the long economic cycle. Long economic Sciences et Belles-Lettres de Bruxelles. 18: 142. Re-
cycles were investigated by Robert Ayres (1989).[11] Ce- trieved 2013-02-18.
sare Marchetti published on long economic cycles and
[3] expit documentation for Rs clusterPower package
on diusion of innovations.[12][13] Arnulf Grblers book
(1990) gives a detailed account of the diusion of in- [4] Raul Rojas. Neural Networks - A Systematic Introduction
frastructures including canals, railroads, highways and (PDF). Retrieved 15 October 2016.
airlines, showing that their diusion followed logistic
shaped curves.[14] [5] A. G. McKendricka; M. Kesava Paia1 (January
1912). XLV.The Rate of Multiplication of Micro-
Carlota Perez used a logistic curve to illustrate the long organisms: A Mathematical Study. Proceedings
(Kondratiev) business cycle with the following labels: be- of the Royal Society of Edinburgh. 31: 649653.
ginning of a technological era as irruption, the ascent as doi:10.1017/S0370164600025426.
6 6 EXTERNAL LINKS

[6] Raymond Pearl and Lowell Reed (June 1920). On the Weisstein, Eric W. Sigmoid Function.
Rate of Growth of the Population of the United States MathWorld.
(PDF). Proc. of the National Academy of Sciences. 6 (6).
p. 275.
Online experiments with JSXGraph
[7] Yukalov, V. I.; Yukalova, E. P.; Sornette, D. (2009).
Punctuated evolution due to delayed carrying capac- Esses are everywhere.
ity. Physica D: Nonlinear Phenomena. 238 (17): 1752.
Seeing the s-curve is everything.
doi:10.1016/j.physd.2009.05.011.

[8] Gershenfeld 1999, p.150 Restricted Logarithmic Growth with Injection

[9] LeCun, Y.; Bottou, L.; Orr, G.; Muller, K. (1998). Orr,
G.; Muller, K., eds. Ecient BackProp (PDF). Neural Net-
works: Tricks of the trade. Springer. ISBN 3-540-65311-
2.

[10] Bod, Hay, Jennedy (eds.) 2003, pp. 147156

[11] Ayres, Robert (1989). Technological Transformations

and Long Waves (PDF).

[12] Marchetti, Cesare (1996). Pervasive Long Waves: Is So-

ciety Cyclotymic (PDF).

[13] Marchetti, Cesare (1988). Kondratiev Revisited-After

One Cycle (PDF).

[14] Grbler, Arnulf (1990). The Rise and Fall of Infrastruc-

tures: Dynamics of Evolution and Technological Change
in Transport (PDF). Heidelberg and New York: Physica-
Verlag.

[15] Perez, Carlota (2002). Technological Revolutions and Fi-

nancial Capital: The Dynamics of Bubbles and Golden
Ages. UK: Edward Elgar Publishing Limited. ISBN 1-
84376-331-1.

5 References
Jannedy, Stefanie; Bod, Rens; Hay, Jennifer
(2003). Probabilistic Linguistics. Cambridge, Mas-
sachusetts: MIT Press. ISBN 0-262-52338-8.
Gershenfeld, Neil A. (1999). The Nature of Math-
ematical Modeling. Cambridge, UK: Cambridge
University Press. ISBN 978-0-521-57095-4.
Kingsland, Sharon E. (1995). Modeling na-
ture: episodes in the history of population ecology.
Chicago: University of Chicago Press. ISBN 0-226-
43728-0.
Weisstein, Eric W. Logistic Equation.
MathWorld.

6 External links
L.J. Linacre, Why logistic ogive and not autocat-
alytic curve?, accessed 2009-09-12.
http://luna.cas.usf.edu/~{}mbrannic/files/
regression/Logistic.html
 7

7 Text and image sources, contributors, and licenses

7.1 Text
Logistic function Source: https://en.wikipedia.org/wiki/Logistic_function?oldid=781292713 Contributors: The Anome, Mrwojo,
JohnOwens, Michael Hardy, Chris-martin, Dominus, Kku, Karada, Ahoerstemeier, Charles Matthews, Populus, Fredrik, Swestrup, Hen-
rygb, Rursus, TittoAssini, Cyrius, Cutler, David Gerard, SimonMayer, Ancheta Wis, Giftlite, Cfp, Shanemcd, BenFrantzDale, Arne-
johs, Jorge Stol, Eequor, Just Another Dan, Alvestrand, MarkSweep, PhDP, Rubik-wuerfel, Qef, Bender235, Wolfman, Nabla, Mister-
Sheik, Edward Z. Yang, Bobo192, O18, Sbarthelme, Jclaer, Eric Kvaalen, Arthena, Plumbago, Menelik3, Dzhim, Woohookitty, 2004-
12-29T22:45Z, Linas, TigerShark, LOL, Tabletop, Mandarax, Qwertyus, Rjwilmsi, Voretus, Mathbot, Eyvin, Fisenko, Tardis, Scythe33,
Kri, Spencerk, Chobot, YurikBot, Wavelength, Bjf, Holon, Vatassery, DaveWF, Intershark, Ketil3, Cmglee, Jinxs, A bit iy, Smack-
Bot, Fons~enwiki, Numsgil, Tom Lougheed, Mcld, MonteChristof, Chris the speller, Fplay, Nbarth, Robth, Colonies Chris, Gramscis
cousin, RyanEberhart, Solarapex, Cybercobra, Richard001, Soobrickay, Metamagician3000, Runia~enwiki, Bcasterline, J Park, Load-
master, Knights who say ni, Debeo Morium, Cryptic C62, Ioannes Pragensis, Neelix, MaxEnt, Doctormatt, Farrwill, Rracecarr, Xan-
tharius, Fyedernoggersnodden, Zeev Grin, Mack2, Richard n, JAnDbot, Bequw, Ldecola, A3nm, Mbhiii, J.delanoy, Andreamubi, Pd-
cook, VolkovBot, TXiKiBoT, Heheman3000, Oanjao, UnitedStatesian, Phmoreno, AJRobbins, AlleborgoBot, Wikiscottcha, Lwolberg,
Jcmendez, Arbor to SJ, Epitron, Rhythmiccycle, Junling, Melcombe, Headlessplatter, ClueBot, DFRussia, The Thing That Should Not Be,
Phipperdee, UKoch, DragonBot, Calimo, Mleconte, Aprock, Humanengr, Mitch Ames, MystBot, Gbedia, Addbot, Fgnievinski, MrOllie,
Download, SpBot, Uscitizenjason, New Image Uploader 929, Echo R314, Lightbot, Cesiumfrog, Albertodonofrio, Legobot, Luckas-bot,
Yobot, MihalOrela, AnomieBOT, Jim1138, Mihal Orela, Materialscientist, Citation bot, ArthurBot, Pargeter1, Tamfatkh, Shadowjams,
A. di M., Foobarhoge, Citation bot 1, Pinethicket, Jonesey95, Vectornaut, Jonkerz, LilyKitty, Duoduoduo, Trinary M01, AlexBartlett4,
Thelema418, RjwilmsiBot, Helwr, EmausBot, Dewritech, RA0808, Dcirovic, ZroBot, AManWithNoPlan, Dsornette, Neduard, ClueBot
NG, Frietjes, Masssly, Helpful Pixie Bot, BG19bot, Ogoorcs, Soerfm, Tundebabzy, ChrisGualtieri, Dexbot, Limit-theorem, Mark viking,
Mre env, Mathmensch, ModalPeak, RingOfTheLords, Ajkirk, Oiyarbepsy, Who Cares X, Lond1984, Latex-yow, Deacon Vorbis, Akishor
and Anonymous: 132

7.2 Images
File:Logistic-curve.svg Source: https://upload.wikimedia.org/wikipedia/commons/8/88/Logistic-curve.svg License: Public domain Con-
tributors: Created from scratch with gnuplot Original artist: Qef (talk)
File:Pierre_Francois_Verhulst.jpg Source: https://upload.wikimedia.org/wikipedia/commons/0/04/Pierre_Francois_Verhulst.jpg Li-
cense: Public domain Contributors: ? Original artist: ?

7.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Hyperbolic function

Sech redirects here. For other uses, see Sech (disam- Hyperbolic functions occur in the solutions of many lin-
biguation). ear dierential equations, for example the equation den-
Sinh redirects here. For the garment, see sinh (cloth- ing a catenary, of some cubic equations, in calculations
ing). of angles and distances in hyperbolic geometry and of
Hyperbolic curve redirects here. For the geometric Laplaces equation in Cartesian coordinates. Laplaces
curve, see Hyperbola. equations are important in many areas of physics, includ-
ing electromagnetic theory, heat transfer, uid dynamics,
and special relativity.
Y In complex analysis, the hyperbolic functions arise as
the imaginary parts of sine and cosine. When consid-
cosh a ered dened by a complex variable, the hyperbolic func-
tions are rational functions of exponentials, and are hence
x y = 1 holomorphic.
sinh a
1 a/2 Hyperbolic functions were introduced in the 1760s in-
dependently by Vincenzo Riccati and Johann Heinrich
1 X Lambert.[9] Riccati used Sc. and Cc. ([co]sinus circulare)
to refer to circular functions and Sh. and Ch. ([co]sinus
hyperbolico) to refer to hyperbolic functions. Lambert
adopted the names but altered the abbreviations to what
they are today.[10] The abbreviations sh and ch are still
used in some other languages, like French and Russian.

A ray through the unit hyperbola x2 y2 = 1 in the point

(cosh a, sinh a) , where a is twice the area between the ray, the hy- 1 Standard analytic expressions
perbola, and the x -axis. For points on the hyperbola below the x
-axis, the area is considered negative (see animated version with
comparison with the trigonometric (circular) functions).

In mathematics, hyperbolic functions are analogs of the

ordinary trigonometric, or circular functions.
The basic hyperbolic functions are the hyperbolic sine
sinh (/snt/ or /an/),[1] and the hyperbolic cosine
cosh (/k/),[2] from which are derived the hyperbolic
tangent tanh (/tnt/ or /n/),[3] hyperbolic cose-
cant csch or cosech (/kok/[2] or /kost/), hy-
perbolic secant sech (/k/ or /st/),[4] and hyper-
bolic cotangent coth (/ko/ or /k/),[5][6] corre-
sponding to the derived trigonometric functions.
The inverse hyperbolic functions are the area hyper-
bolic sine arsinh (also called asinh or sometimes
arcsinh)[7][8] and so on.
Just as the points (cos t, sin t) form a circle with a unit
radius, the points (cosh t, sinh t) form the right half of
the equilateral hyperbola. The hyperbolic functions take sinh, cosh and tanh
a real argument called a hyperbolic angle. The size of a
hyperbolic angle is twice the area of its hyperbolic sector. The hyperbolic functions are:
The hyperbolic functions may be dened in terms of the
legs of a right triangle covering this sector. Hyperbolic sine:

1
2 1 STANDARD ANALYTIC EXPRESSIONS

csch, sech and coth

(b) sinh(x) is half the dierence of ex and ex

Hyperbolic tangent:

sinh x ex ex
tanh x = = x =
cosh x e + ex
e2x 1 1 e2x
= =
2x
e +1 1 + e2x

Hyperbolic cotangent: x = 0

cosh x ex + ex
(a) cosh(x) is the average of ex and ex coth x = = x =
sinh x e ex
e2x + 1 1 + e2x
= =
e2x 1 1 e2x
ex ex e2x 1 1 e2x
sinh x = = = Hyperbolic secant:
2 2ex 2ex

Hyperbolic cosine:
1 2
sech x = = x =
cosh x e + ex

ex + ex e2x + 1 1 + e2x 2ex 2ex

cosh x = = = = =
2 2e x 2ex e2x +1 1 + e2x
 3

Hyperbolic cosecant: x = 0 2 Special meanings

2.1 Hyperbolic cosine

1 2
csch x = = x = It can be shown that the area under the curve of cosh (x)
sinh x e ex over a nite interval is always equal to the arc length cor-
responding to that interval:[11]
2ex 2ex
= =
e2x 1 1 e2x
( )2
b b
Hyperbolic functions can be introduced via imaginary area = d
cosh (x) dx = 1+ cosh (x) dx = length arc
circular angles: a a dx

Hyperbolic sine: 2.2 Hyperbolic tangent

The hyperbolic tangent is the solution to the dierential
equation f = 1 f 2 with f(0)=0 and the nonlinear
sinh x = i sin(ix) boundary value problem:[12] [13]

Hyperbolic cosine: 1
f = f3 f; f (0) = f () = 0
2

cosh x = cos(ix) 3 Useful relations

Odd and even functions:
Hyperbolic tangent:

sinh(x) = sinh x
cosh(x) = cosh x
tanh x = i tan(ix)
Hence:
Hyperbolic cotangent:

tanh(x) = tanh x
coth(x) = coth x
coth x = i cot(ix) sech(x) = sech x
csch(x) = csch x
Hyperbolic secant:
It can be seen that cosh x and sech x are even functions;
the others are odd functions.

sech x = sec(ix) 1
arsech x = arcosh
x
Hyperbolic cosecant: 1
arcsch x = arsinh
x
1
arcoth x = artanh
x
csch x = i csc(ix) Hyperbolic sine and cosine satisfy:

where i is the imaginary unit with the property that i2 =

1. cosh x + sinh x = ex
The complex forms in the denitions above derive from cosh x sinh x = ex
Eulers formula. cosh2 x sinh2 x = 1
4 5 DERIVATIVES

(x) cosh(x)1
the last which is similar to the Pythagorean trigonometric tanh 2 = sinh(x) = coth(x) csch(x)
identity. [15]

One also has

sech2 x = 1 tanh2 x
4 Inverse functions as logarithms
csch2 x = coth2 x 1
Main article: Inverse hyperbolic function
for the other functions. ( )
arsinh(x) = ln x + x2 + 1
( )
3.1 Sums of arguments arcosh(x) = ln x + x2 1 ; x 1
( )
sinh(x + y) = sinh(x) cosh(y) + cosh(x) sinh(y) 1 1+x
artanh(x) = ln ; |x| < 1
cosh(x + y) = cosh(x) cosh(y) + sinh(x) sinh(y) 2 1x
( )
tanh x + tanh y 1 x+1
tanh(x + y) = arcoth(x) = ln ; |x| > 1
1 + tanh x tanh y 2 x1
( ) ( )
particularly 1 1 1 + 1 x2
arsech(x) = ln + 1 = ln ;0 < x 1
x x2 x
( ) ( )
cosh(2x) = sinh2 x + cosh2 x = 2 sinh2 x + 1 = 2 cosh2 x 1 1 1 1 + x2 + 1
arcsch(x) = ln + + 1 = ln ; x = 0
sinh(2x) = 2 sinh x cosh x x x2 x
Also:

5 Derivatives
x+y xy
sinh x + sinh y = 2 sinh cosh
2 2
x+y xy d
cosh x + cosh y = 2 cosh cosh sinh x = cosh x
2 2 dx
d
cosh x = sinh x
3.2 Subtraction formulas dx
d
sinh(x y) = sinh(x) cosh(y) cosh(x) sinh(y) tanh x = 1 tanh2 x = sech2 x = 1/ cosh2 x
dx
cosh(x y) = cosh(x) cosh(y) sinh(x) sinh(y)
Also: d
coth x = 1 coth2 x = csch2 x = 1/ sinh2 x
dx

x+y xy d
sinh x sinh y = 2 cosh sinh sech x = tanh x sech x
2 2 dx
x+y xy
cosh x cosh y = 2 sinh sinh d
2 2 csch x = coth x csch x
dx
Source.[14]
d 1
arsinh x =
dx x2 + 1
3.3 Half argument formulas
d 1
arcosh x =
(x) sinh(x) cosh(x) 1 dx x 1
2
sinh = = sgn(x)
2 2(cosh(x) + 1) 2 d 1
artanh x =
where sgn is the sign function. dx 1 x2
d 1
arcoth x =
(x) cosh(x) + 1 dx 1 x2
cosh =
2 2 d 1
arsech x =
(x) sinh(x) cosh(x) 1 dx1
ex x 1 x2
tanh = = sgn(x) = x
2 cosh(x) + 1 cosh(x) + 1 e +d 1 1
arcsch x =
If x 0, then dx |x| 1 + x2
 5

6 Second derivatives

x3 x5 x7 x2n+1
Sinh and cosh are both equal to their second derivative, sinh x = x + + + + =
3! 5! 7! (2n + 1)!
that is: n=0

The function sinh x has a Taylor series expression with

only odd exponents for x. Thus it is an odd function, that
d2 is, sinh x = sinh(x), and sinh 0 = 0.
sinh x = sinh x
dx2

d2 x2 x4 x6
x2n
cosh x = cosh x . cosh x = 1 + + + + =
dx2 2! 4! 6! (2n)!
n=0
All functions with this property are linear combinations
of sinh and cosh, in particular the exponential functions The function cosh x has a Taylor series expression with
ex and ex , and the zero function f (x) = 0 . only even exponents for x. Thus it is an even function,
that is, symmetric with respect to the y-axis. The sum of
the sinh and cosh series is the innite series expression of
the exponential function.
7 Standard integrals
For a full list, see list of integrals of hyperbolic functions.
x3 2x5 17x7 22n (22n 1)B2n x2n1
tanh x = x + + = ,
3 15 315 n=1
(2n)!
sinh(ax) dx = a1 cosh(ax) + C x x3 2x5
22n B2n x2n1
coth x = x1 + + + = x1 + ,0 <
3 45 945 (2n)!
cosh(ax) dx = a1 sinh(ax) + C n=1

x2 5x4 61x6 E2n x2n
sech x = 1 + + = , |x| <
tanh(ax) dx = a1 ln(cosh(ax)) + C 2 24 720 n=0
(2n)! 2

x 7x3 31x5 2(1 22n1 )B
coth(ax) dx = a1 ln(sinh(ax)) + C csch x = x1 + + = x1 +
6 360 15120 (2n)!
n=1

sech(ax) dx = a1 arctan(sinh(ax)) + C where:

( ( ax ))
csch(ax) dx = a1 ln tanh + C = a1 ln |csch(ax) Bn is the
coth(ax)| + C number
nth Bernoulli
2
The following integrals can be En is the nth Euler number
proved
using hyperbolic substitution:
1 (u)
du = arsinh +C

a2 + u2 a 9 Comparison with circular func-
1 (u)
du = arcosh +C tions
u 2 a2 a
(u)
1 1
du = a artanh + C; u2 < a2 The hyperbolic functions represent an expansion of
a2 u2 a
(u) trigonometry beyond the circular functions. Both
1 1 2 2 types depend on an argument, either circular angle or
du = a arcoth + C; u > a
a2 u2 a hyperbolic angle.
(u)
1
du = a1 arsech +C Since the area of a circular sector with radius r and angle
u a2 u2 a 2
u u is r 2u , it will be equal to u when r = square root of 2.
1  
du = a1 arcsch   + C In the diagram such a circle is tangent to the hyperbola xy
u a2 + u2 a = 1 at (1,1). The yellow sector depicts an area and angle
where C is the constant of integration. magnitude. Similarly, the yellow and red sectors together
depict an area and hyperbolic angle magnitude.
The legs of the two right triangles with hypotenuse on the
8 Taylor series expressions ray dening the angles are of length 2 times the circular
and hyperbolic functions.
It is possible to express the above functions as Taylor se- Mellon Haskell of University of California, Berkeley de-
ries: scribed the basis of hyperbolic functions in areas of
6 11 RELATIONSHIP TO THE EXPONENTIAL FUNCTION

y
y=1/x
sinh(2x) = 2 sinh x cosh x
y=x
cosh(2x) = cosh2 x + sinh2 x = 2 cosh2 x 1 = 2 sinh2 x + 1
2 tanh x
tanh(2x) =
1 + tanh2 x
2 tanh x
x2+y2=2 u) sinh(2x) =
h(
1 tanh2 x

2
os

si
c

y=ax: a<1

nh
2

1 + tanh2 x
)
2
(u

(u
si
cosh(2x) =
os

)
c

n(
1 tanh2 x
2

u)
u
x and the half-argument formulas[17]

sinh x2 = 2 (cosh x 1) Note: This is
1

equivalent to its circular counterpart multiplied

by 1.

Circle and hyperbola tangent at (1,1) display geometry of circu- cosh x2 = 12 (cosh x + 1) Note: This corre-
lar functions in terms of circular sector area u and hyperbolic sponds to its circular counterpart.
functions depending on hyperbolic sector area u.
tanh x2 = cosh x1 sinh x cosh x1
cosh x+1 = cosh x+1 = sinh x =
coth x csch x.

hyperbolic sectors in an 1895 article in Bulletin of the coth x2 = coth x + csch x.

American Mathematical Society (see External links). He
refers to the hyperbolic angle as an invariant measure with The derivative of sinh x is cosh x and the derivative of
respect to the squeeze mapping just as circular angle is cosh x is sinh x; this is similar to trigonometric functions,
invariant under rotation. albeit the sign is dierent (i.e., the derivative of cos x is
sin x).
The Gudermannian function gives a direct relationship
between the circular functions and the hyperbolic ones
that does not involve complex numbers.
The graph of the function a cosh(x/a) is the catenary, the
10 Identities curve formed by a uniform exible chain hanging freely
between two xed points under uniform gravity.

The hyperbolic functions satisfy many identities, all of

them similar in form to the trigonometric identities. In
fact, Osborns rule[16] states that one can convert any 11 Relationship to the exponential
trigonometric identity into a hyperbolic identity by ex- function
panding it completely in terms of integral powers of sines
and cosines, changing sine to sinh and cosine to cosh, and
switching the sign of every term which contains a prod- From the denitions of the hyperbolic sine and cosine, we
uct of 2, 6, 10, 14, ... sinhs. This yields for example the can derive the following identities:
addition theorems

ex = cosh x + sinh x

and
sinh(x + y) = sinh(x) cosh(y) + cosh(x) sinh(y)
cosh(x + y) = cosh(x) cosh(y) + sinh(x) sinh(y)
tanh(x) + tanh(y) ex = cosh x sinh x
tanh(x + y) =
1 + tanh(x) tanh(y)
These expressions are analogous to the expressions for
sine and cosine, based on Eulers formula, as sums of
the double argument formulas complex exponentials.
 7

12 Hyperbolic functions for com- [3] Collins Concise Dictionary, p. 1520

plex numbers [4] Collins Concise Dictionary, p. 1340

[5] Collins Concise Dictionary, p. 329

Since the exponential function can be dened for any
complex argument, we can extend the denitions of the
hyperbolic functions also to complex arguments. The
functions sinh z and cosh z are then holomorphic.
Relationships to ordinary trigonometric functions are
given by Eulers formula for complex numbers:
[6] tanh
[7] Abramowitz, Milton; Stegun, Irene A., eds. (1972),
Handbook of Mathematical Functions with Formulas,
Graphs, and Mathematical Tables, New York: Dover Pub-
lications, ISBN 978-0-486-61272-0

[8] Some examples of using arcsinh found in Google Books.

eix = cos x + i sin x [9] Robert E. Bradley, Lawrence A. D'Antonio, Charles Ed-
ix ward Sandifer. Euler at 300: an appreciation. Mathemat-
e = cos x i sin x
ical Association of America, 2007. Page 100.
so:
[10] Georg F. Becker. Hyperbolic functions. Read Books,
1931. Page xlviii.

1 ( ix ) [11] N.P., Bali (2005). Golden Integral Calculus. Firewall Me-

cosh(ix) = e + eix = cos x dia. p. 472. ISBN 81-7008-169-6.
2
1 ( ix )
sinh(ix) = e eix = i sin x [12] Eric W. Weisstein. Hyperbolic Tangent. MathWorld.
2 Retrieved 2008-10-20.
cosh(x + iy) = cosh(x) cos(y) + i sinh(x) sin(y)
[13] Derivation of tanh solution to 1/2 f=f^3f.... Math
sinh(x + iy) = sinh(x) cos(y) + i cosh(x) sin(y)
StackExchange. Retrieved 18 March 2016.
tanh(ix) = i tan x
[14] Martin, George E. (1986). The foundations of geometry
cosh x = cos(ix)
and the non-euclidean plane (1., corr. Springer ed.). New
sinh x = i sin(ix) York: Springer-Verlag. p. 416. ISBN 3-540-90694-0.
tanh x = i tan(ix)
[15] "math.stackexchange.com/q/1565753/88985".
Thus, hyperbolic functions are periodic with respect to StackExchange (mathematics). Retrieved 24 January
2016.
the imaginary component, with period 2i ( i for hy-
perbolic tangent and cotangent). [16] Osborn, G. (July 1902). Mnemonic for hyperbolic for-
mulae. The Mathematical Gazette. 2 (34): 189. JSTOR
3602492.
13 See also [17] Peterson, John Charles (2003). Technical mathematics
with calculus (3rd ed.). Cengage Learning. p. 1155.
e (mathematical constant) ISBN 0-7668-6189-9., Chapter 26, page 1155

Equal incircles theorem, based on sinh

Inverse hyperbolic functions 15 External links

List of integrals of hyperbolic functions Mellen W. Haskell (1895) On the introduction of the
Poinsots spirals notion of hyperbolic functions Bulletin of the Amer-
ican Mathematical Society 1(6):1559.
Sigmoid function
Hazewinkel, Michiel, ed. (2001), Hyperbolic
Trigonometric functions functions, Encyclopedia of Mathematics, Springer,
ISBN 978-1-55608-010-4
Modied hyperbolic tangent
Hyperbolic functions on PlanetMath
Hyperbolic functions entry at MathWorld
14 References
GonioLab: Visualization of the unit circle, trigono-
[1] (1999) Collins Concise Dictionary, 4th edition, Harper- metric and hyperbolic functions (Java Web Start)
Collins, Glasgow, ISBN 0 00 472257 4, p. 1386
Web-based calculator of hyperbolic functions
[2] Collins Concise Dictionary, p. 328
8 16 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES

16 Text and image sources, contributors, and licenses

16.1 Text
Hyperbolic function Source: https://en.wikipedia.org/wiki/Hyperbolic_function?oldid=776492647 Contributors: AxelBoldt, The Anome,
Josh Grosse, Michael Hardy, Gabbe, TakuyaMurata, Eric119, SebastianHelm, Cyp, Suisui, Smack, Hashar, Charles Matthews, Saltine,
Robbot, Ktims, Fredrik, Georg Muntingh, Sverdrup, Wereon, Chris-gore, Lzur, Jleedev, Tea2min, Giftlite, Gene Ward Smith, Ben-
FrantzDale, Herbee, Wwoods, Macrakis, Pne, LucasVB, MarkSweep, Icairns, Agro r, Urhixidur, Tsemii, Shen, Moxfyre, Roberdin, Sam
Derbyshire, Bender235, Gauge, Rgdboer, Kwamikagami, Robotje, Anthony Appleyard, PAR, Wiccan Quagga, Geraldshields11, Woj-
ciechSwiderski~enwiki, Oleg Alexandrov, Linas, Georgia guy, LOL, Prophile, Guardian of Light, KarlJorgensen, Qwertyus, Canderson7,
Sjakkalle, MarSch, Brighterorange, Bhadani, Mathbot, Compotatoj, CiaPan, Chobot, Krishnavedala, DVdm, Bgwhite, YurikBot, Wave-
length, Me and, Deskana, Gorie, Googl, Tob~enwiki, Ninly, Cojoco, Gesslein, Mebden, Zvika, AndrewWTaylor, SmackBot, Maksim-
e~enwiki, KnowledgeOfSelf, KennethJ, Skizzik, Bh3u4m, Bluebot, Silly rabbit, Metacomet, Nbarth, Noerdosnum, Can't sleep, clown
will eat me, V1adis1av, SKvalen, Ctifumdope, Seraph85, SashatoBot, Lambiam, Tim bates, Sohale, Dr Greg, Courcelles, Heliomance,
Gco, JRSpriggs, Patchouli, Xanthoxyl, Doctormatt, A876, Gogo Dodo, Lee, Sharonlees, Calvinballing, Nonagonal Spider, Escarbot,
Ben pcc, Pikalax, JAnDbot, Deective, Husond, TV4Fun, Montparnasse, Albmont, Email4mobile, Bcherkas, Error792, Hdt83, Vino-
grad19, Allispaul, HowiAuckland, Salih, Matt24, Trumpet marietta 45750, Policron, Fylwind, KylieTastic, DavidCBryant, Moverington,
Deor, VolkovBot, LokiClock, Jobu0101, Rei-bot, Aaron Rotenberg, Jackfork, Societelibre, Andy Dingley, Snevets, Stephen J. Brooks,
TheBendster, Carlodn6, Dcqec111, SieBot, Derepi, GeiwTeol, Dragnmn, COBot, AlanUS, Azo bob, Swetrix, Justin W Smith, JP.Martin-
Flatin, JoeHillen, Ocer781, DragonBot, Watchduck, Mleconte, Egmontaz, Kiensvay, WikHead, Kal-El-Bot, MystBot, Eettttt, Addbot,
Fgnievinski, Chzz, Jasper Deng, West.andrew.g, Barak Sh, Andy Fisher aKa Shine, HalFonts, Legobot, Luckas-bot, Yobot, Amirobot,
Jdh8, Nallimbot, TestEditBot, AnomieBOT, Archon 2488, Daniele Pugliesi, Materialscientist, Kritikool99, Txebixev, Drilnoth, Grou-
choBot, Rocky2009, Michiganfan9000, FrescoBot, Cottonmouse, Footyfanatic3000, Deathbeast, Fanix1128, Sergiacid, Wagnerjp25, Vt
Zajac, Jeandavid54, Double sharp, Duoduoduo, Searchpow, N4m3, Nascar1996, Ripchip Bot, Eleolar, K4vin, EmausBot, John of Read-
ing, Orphan Wiki, WikitanvirBot, Richnfg, Alle, ZroBot, Access Denied, Suslindisambiguator, Quondum, D.Lazard, Michael.YX.Wu,
Coasterlover1994, Scientic29, Culero101, Vizziee, TYelliot, Rocketrod1960, Faramir1138, ClueBot NG, Matthiaspaul, Razor-ockham,
MathKeduor7, IgnorantArmies, Helpful Pixie Bot, BG19bot, Vagobot, Brad7777, Dlituiev, Reciprocist, BattyBot, Steve911411, Doy-
oon1995, ChrisGualtieri, MadGuy7023, Elen so, Blah314, Cerabot~enwiki, Pokajanje, Qantas94Heavy, WIKIWIZDOM, Ashis Kumar
Sahoo, Armeria wiki, Jrgsampaio, Shubham10,000, Johanvts, Rabin.jcet13, Thelad23, WillemienH, Loraof, Degenerate prodigy, Kaspar-
Bot, Yuvalr, Myawr, InternetArchiveBot, Ksks56123, GreenC bot, NeonSavage, Bender the Bot and Anonymous: 334

16.2 Images
File:Circular_and_hyperbolic_angle.svg Source: https://upload.wikimedia.org/wikipedia/commons/c/c6/Circular_and_hyperbolic_
angle.svg License: CC0 Contributors: Own work Original artist: Original: Maschen

File:Commons-logo.svg Source: https://upload.wikimedia.org/wikipedia/en/4/4a/Commons-logo.svg License: PD Contributors: ? Origi-

nal artist: ?
File:Complex_Cosh.jpg Source: https://upload.wikimedia.org/wikipedia/commons/4/4e/Complex_Cosh.jpg License: Public domain
Contributors: Eigenes Werk (own work) made with mathematica 5.0 Original artist: Jan Homann
File:Complex_Coth.jpg Source: https://upload.wikimedia.org/wikipedia/commons/4/44/Complex_Coth.jpg License: Public domain
Contributors: Eigenes Werk (own work) made with mathematica 5.0 Original artist: Jan Homann
File:Complex_Csch.jpg Source: https://upload.wikimedia.org/wikipedia/commons/4/49/Complex_Csch.jpg License: Public domain
Contributors: Eigenes Werk (own work), made with mathematica 5.0 Original artist: Jan Homann
File:Complex_Sech.jpg Source: https://upload.wikimedia.org/wikipedia/commons/a/a4/Complex_Sech.jpg License: Public domain
Contributors: Eigenes Werk (own work), made with mathematica 5.0 Original artist: Jan Homann
File:Complex_Sinh.jpg Source: https://upload.wikimedia.org/wikipedia/commons/9/9a/Complex_Sinh.jpg License: Public domain Con-
tributors: Eigenes Werk (own work), made with mathematica 5.0 Original artist: Jan Homann
File:Complex_Tanh.jpg Source: https://upload.wikimedia.org/wikipedia/commons/6/63/Complex_Tanh.jpg License: Public domain
Contributors: Eigenes Werk (own work), made with mathematica 5.0 Original artist: Jan Homann
File:Csch_sech_coth.svg Source: https://upload.wikimedia.org/wikipedia/commons/6/61/Csch_sech_coth.svg License: Public domain
Contributors: Own work Original artist: Fylwind at English Wikipedia
File:Hyperbolic_and_exponential;_cosh.svg Source: https://upload.wikimedia.org/wikipedia/commons/e/ef/Hyperbolic_and_
exponential%3B_cosh.svg License: CC BY-SA 3.0 Contributors: Own work Original artist: Krishnavedala
File:Hyperbolic_and_exponential;_sinh.svg Source: https://upload.wikimedia.org/wikipedia/commons/7/7a/Hyperbolic_and_
exponential%3B_sinh.svg License: CC BY-SA 3.0 Contributors: Own work Original artist: Krishnavedala
File:Hyperbolic_functions-2.svg Source: https://upload.wikimedia.org/wikipedia/commons/b/bc/Hyperbolic_functions-2.svg License:
Public domain Contributors:
Hyperbolic_functions.svg Original artist: Hyperbolic_functions.svg: Original uploader was Marco Polo at en.wikipedia
File:Sinh_cosh_tanh.svg Source: https://upload.wikimedia.org/wikipedia/commons/7/76/Sinh_cosh_tanh.svg License: Public domain
Contributors: Own work Original artist: Fylwind at English Wikipedia

16.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Inverse trigonometric functions

In mathematics, the inverse trigonometric functions 2.1 Principal values

(occasionally called cyclometric functions[1] ) are the
inverse functions of the trigonometric functions (with
suitably restricted domains). Specically, they are the Since none of the six trigonometric functions are one-to-
inverses of the sine, cosine, tangent, cotangent, secant, one, they are restricted in order to have inverse functions.
and cosecant functions, and are used to obtain an an- Therefore the ranges of the inverse functions are proper
gle from any of the angles trigonometric ratios. Inverse subsets of the domains of the original functions
trigonometric functions are widely used in engineering, For example, using function in the sense of multivalued
navigation, physics, and geometry. functions, just as the square root function y = x could be
dened from y2 = x, the function y = arcsin(x) is dened
so that sin(y) = x. There are multiple numbers y such
1 Notation that sin(y) = x; for example, sin(0) = 0, but also sin() =
0, sin(2) = 0, etc. When only one value is desired, the
function may be restricted to its principal branch. With
There are several notations used for the inverse trigono-
this restriction, for each x in the domain the expression
metric functions.
arcsin(x) will evaluate only to a single value, called its
The most common convention is to name inverse trigono- principal value. These properties apply to all the inverse
metric functions using an arc- prex, e.g., arcsin(x), ar- trigonometric functions.
ccos(x), arctan(x), etc. This convention is used through-
The principal inverses are listed in the following table.
out the article. When measuring in radians, an angle of
radians will correspond to an arc whose length is r, (Note: Some authors dene the range of arcsecant to be (
where r is the radius of the circle. Thus, in the unit cir- 0 y < /2 or y < 3/2 ), because the tangent function
cle, the arc whose cosine is x is the same as the an- is nonnegative on this domain. This makes some com-
gle whose cosine is x, because the length of the arc of putations more consistent. For example using this range,
the circle in radii is the same as the measurement of the tan(arcsec(x)) = x2 1, whereas with the range ( 0 y <
angle in radians.[2] Similarly, in computer programming /2 or /2 < y ), we would have to write tan(arcsec(x))
languages the inverse trigonometric functions are usually = x2 1, since tangent is nonnegative on 0 y < /2
called asin, acos, atan. but nonpositive on /2 < y . For a similar reason, the
1 1 1 same authors dene the range of arccosecant to be <
The notations sin (x), cos (x), tan (x), etc., as intro-
[3][4] y /2 or 0 < y /2.)
duced by John Herschel in 1813, are often used as
well, but this convention logically conicts with the com- If x is allowed to be a complex number, then the range of
mon semantics for expressions like sin2 (x), which refer y applies only to its real part.
to numeric power rather than function composition, and
therefore may result in confusion between multiplicative
inverse and compositional inverse. The confusion is
somewhat ameliorated by the fact that each of the recip-
rocal trigonometric functions has its own namefor ex-
ample, (cos(x))1 = sec(x). Nevertheless, certain authors
advise against using it for its ambiguity.[5] 2.2 Relationships between trigonometric
Another convention used by a few authors[6] is to use functions and inverse trigonometric
a majuscule (capital/upper-case) rst letter along with a functions
1 superscript, e.g., Sin1 (x), Cos1 (x), Tan1 (x), etc.,
which avoids confusing them with the multiplicative in-
Trigonometric functions of inverse trigonometric func-
verse, which should be represented by sin1 (x), cos1 (x),
tions are tabulated below. A quick way to derive them
etc.
is by considering the geometry of a right-angled trian-
gle, with one side of length 1, and another side of length
x (any real number between 0 and 1), then applying the
2 Basic properties Pythagorean theorem and denitions of the trigonomet-
ric ratios. Purely algebraic derivations are longer.

1
2 2 BASIC PROPERTIES

Principal values of the arcsec(x) and arccsc(x) functions graphed

on the cartesian plane.

arccos(x) = arcsin(x)
2

arccot(x) = arctan(x)
2

arccsc(x) = arcsec(x)
2
Negative arguments:

arcsin(x) = arcsin(x)
arccos(x) = arccos(x)
arctan(x) = arctan(x)
arccot(x) = arccot(x)
arcsec(x) = arcsec(x)
arccsc(x) = arccsc(x)
The usual principal values of the arcsin(x) (red) and arccos(x)
(blue) functions graphed on the cartesian plane. Reciprocal arguments:

( )
1
arccos = arcsec(x)
x
( )
1
arcsin = arccsc(x)
x
( )
1
arctan = arctan(x) = arccot(x) , if x > 0
x 2
( )
1
arctan = arctan(x) = arccot(x) , if x < 0
x 2
( )
1
arccot = arccot(x) = arctan(x) , if x > 0
x 2
The usual principal values of the arctan(x) and arccot(x) func- ( )
tions graphed on the cartesian plane. 1 3
arccot = arccot(x) = + arctan(x) , if x < 0
x 2
( )
1
arcsec = arccos(x)
2.3 Relationships among the inverse x
( )
trigonometric functions arccsc
1
= arcsin(x)
x
Complementary angles: If you only have a fragment of a sine table:
3.2 Expression as denite integrals 3

( ) d 1
arcsin(z) = ; z = 1, +1
arccos(x) = arcsin 1 x2 , if 0 x 1 dz 1 z2
1 ( ) d 1
arccos(x) = arccos 2x2 1 , if 0 x 1 arccos(z) = ; z = 1, +1
2 dz 1 z2
1 ( ) d 1
arcsin(x) = arccos 1 2x2 , if 0 x 1 arctan(z) = ; z = i, +i
2 ( ) dz 1 + z2
x d 1
arctan(x) = arcsin arccot(z) = ; z = i, +i
x2 + 1 dz 1 + z2
d 1
Whenever the square root of a complex number is used arcsec(z) = ; z = 1, 0, +1
dz z2 1 1
here, we choose the root with the positive real part (or z2
positive imaginary part if the square was negative real). d 1
( ) arccsc(z) = ; z = 1, 0, +1
From the half-angle formula, tan 2 = 1+cossin dz z2 1 1
, we get: z2

Only for real values of x:

( )
x
arcsin(x) = 2 arctan d 1
1 + 1 x2 arcsec(x) = ; |x| > 1
( ) dx |x| x2 1
1 x2 d 1
arccos(x) = 2 arctan , if 1 < x +1 arccsc(x) = ; |x| > 1
1+x dx |x| x2 1
( ) For a sample derivation: if = arcsin x, we get:
x
arctan(x) = 2 arctan
1 + 1 + x2
d arcsin(x) d d 1 1 1
= = = = =
dx d sin() cos()d cos()
1 sin2 () 1
2.4 Arctangent addition formula
( )
u+v 3.2 Expression as denite integrals
arctan(u)+arctan(v) = arctan (mod ) , uv = 1 .
1 uv
Integrating the derivative and xing the value at one point
This is derived from the tangent addition formula gives an expression for the inverse trigonometric function
as a denite integral:

tan() + tan() x
tan( + ) = , 1
1 tan() tan()
arcsin(x) = dz , |x| 1
0 1 z2
1
by letting 1
arccos(x) = dz , |x| 1
1 z2
xx
1
arctan(x) = 2+1
dz ,
z
= arctan(u) , = arctan(v) .
0
1
arccot(x) = dz ,
x z2 + 1
x 1
1 1
3 In calculus arcsec(x) = dz = + dz , x1
z z 21 z z 21
1 x
x
1 1
arccsc(x) = dz = dz , x1
3.1 Derivatives of inverse trigonometric x z z2 1 z z 1
2

functions When x equals 1, the integrals with limited domains are

improper integrals, but still well-dened.
Main article: Dierentiation of trigonometric
functions
3.3 Innite series
Like the sine and cosine functions, the inverse trigono-
The derivatives for complex values of z are as follows: metric functions can be calculated using power series, as
4 3 IN CALCULUS

follows. For arcsine, the series can be derived by expand- numbers running from 1 to 1. The partial denominators
1
ing its derivative, 1z 2
, as a binomial series, and in- are the odd natural numbers, and the partial numerators
tegrating term by term (using the integral denition as (after the rst) are just (nz)2 , with each perfect square ap-
above). The series for arctangent can similarly be derived pearing once. The rst was developed by Leonhard Euler;
by expanding its derivative 1+z 1
2 in a geometric series
the second by Carl Friedrich Gauss utilizing the Gaussian
and applying the integral denition above (see Leibniz hypergeometric series.
series).

3.4 Indenite integrals of inverse trigono-

( ) 3 ( ) ( )
metric (2n) 2n+1
1 z 1 3 z5 1 3 5 z7 (2n functions
1)!! z 2n+1
n z
arcsin(z) = z+ + + + = = ; |z| 1
2 3 24 5 246 7 n
(2n)!! 2n + 1 n=0 4 (2n + 1)
For real and complex values of z:
n=0

z3 z5 z7 (1)n z 2n+1
arctan(z) = z + + = ; |z| 1 z = i, i
3 5 7 2n + 1
n=0
arcsin(z) dz = z arcsin(z) + 1 z 2 + C
Series for the other inverse trigonometric functions can
be given in terms of these according to the relationships arccos(z) dz = z arccos(z) 1 z 2 + C
given above. For example, arccos x = /2 arcsin x
, arccsc x = arcsin(1/x) , and so on. Another series is 1 ( )
arctan(z) dz = z arctan(z) ln 1 + z 2 + C
given by: 2

1 ( )
( )2 x2n [7]
arccot(z) dz = z arccot(z) + ln 1 + z 2 + C
2 arcsin x2 = n=1 2
n2 (2n
n) [ ( )]
z2 1
arcsec(z) dz = z arcsec(z) ln z 1 + +C
Leonhard Euler found a more ecient series for the arc- z2
tangent, which is: [ ( )]
z2 1
arccsc(z) dz = z arccsc(z) + ln z 1 + +C
z2
n
z 2kz 2
arctan(z) = . For real x 1:
1 + z 2 n=0 (2k + 1)(1 + z 2 )
k=1

(Notice that the term in the sum for n = 0 is the empty ( )

product which is 1.) arcsec(x) dx = x arcsec(x) ln x + x2 1 + C
Alternatively, this can be expressed: ( )
arccsc(x) dx = x arccsc(x) + ln x + x2 1 + C

22n (n!)2 z 2n+1 For all real x not between 1 and 1:
arctan z =
n=0
(2n + 1)! (1 + z 2 )n+1
( )
 
arcsec(x) dx = x arcsec(x) sgn(x) ln x + x2 1 + C
3.3.1 Variant: Continued fractions for arctangent
( )
 
Two alternatives to the power series for arctangent are arccsc(x) dx = x arccsc(x) + sgn(x) ln x + x2 1 + C
these generalized continued fractions:
The absolute value is necessary to compensate for both
negative and positive values of the arcsecant and arccose-
z cant functions. z function is also necessary due
The signum
arctan(z) = =
(1z)2 the 2derivatives of the two func-
to the absolute values in(1z)
1+ 1 +
tions, which create two dierent
(3z)2 (2z)2 solutions for positive and
3 1z 2 + negative values 3
of +
x. These can be further simplied us-
(5z)2 (3z)2
5 3z 2 + ing2 the logarithmic denitions
5+ of the inverse hyperbolic
(7z)
functions: (4z)2
7 5z +
2 7+
. .
9 7z 2 + . . 9 + ..

The second of these is valid in the cut complex plane.
arcsec(x) dx = x arcsec(x) arcosh(|x|) + C
There are two cuts, from i to the point at innity, go-

ing down the imaginary axis, and from i to the point at
innity, going up the same axis. It works best for real arccsc(x) dx = x arccsc(x) + arcosh(|x|) + C
4.1 Logarithmic forms 5

The absolute value in the argument of the arcosh function

creates a negative half of its graph, making it identical to
the signum logarithmic function shown above. arccot(z) = arctan(z) z = i, i
2
All of these antiderivatives can be derived using which has the same cut as arctan;
integration by parts and the simple derivative forms
shown above. ( )
1
arcsec(z) = arccos z = 1, 0, +1
z
3.4.1 Example
where the part of the real axis between 1 and +1 inclu-
Using u dv = uv v du , set sive is the cut between the principal sheet of arcsec and
other sheets;

u = arcsin(x) dv = dx ( )
1
du =
dx
v=x arccsc(z) = arcsin z = 1, 0, +1
z
1 x2
which has the same cut as arcsec.
Then

4.1 Logarithmic forms

x
arcsin(x) dx = x arcsin(x) dx
1 x2 These functions may also be expressed using complex
logarithms. This extends in a natural fashion their domain
which by a simple substitution yields the nal result:
to the complex plane.

( )
arcsin(x) dx = x arcsin(x) + 1 x2 + C
arcsin(z) = i ln iz + 1 z 2

( ) ( )
4 Extension to complex plane arccos(z) = i ln z + z 2 1 = + i ln iz + 1 z 2 =
2 2
Since the inverse trigonometric functions are analytic
functions, they can be extended from the real line to the arctan(z) = 21 i [ln (1 iz) ln (1 + iz)]
complex plane. This results in functions with multiple
sheets and branch points. One possible way of dening [ ( ) ( )]
the extensions is: i i
arccot(z) = 1
2i ln 1 ln 1 +
z z
z ( ) ( )
dx 1 1 1 i
arctan(z) = z = i, +i arcsec(z) = i ln 1+ = i ln 1 2 + + =
0 1 + x2 z2 z z z 2 2
where the part of the imaginary axis which does not lie ( )
strictly between i and +i is the cut between the principal 1 i
sheet and other sheets; arccsc(z) = i ln 1 2 +
z z

( ) Elementary proofs of these relations proceed via expan-

z sion to exponential forms of the trigonometric functions.
arcsin(z) = arctan z = 1, +1
1 z2
where (the square-root function has its cut along the neg- 4.1.1 Example proof
ative real axis and) the part of the real axis which does
not lie strictly between 1 and +1 is the cut between the sin() = z
principal sheet of arcsin and other sheets; = arcsin(z)
Using the exponential denition of sine, one obtains

arccos(z) = arcsin(z) z = 1, +1
2
ei ei
which has the same cut as arcsin; z=
2i
6 5 APPLICATIONS

Let 5.1.1 Application: nding the angle of a right tri-

angle

= ei
Solving for

1

z=
2i
1
2iz =

1
2iz = 0

2 2iz 1 = 0

= iz 1 z 2 = ei
( )
i = ln iz 1 z 2
( )
= i ln iz 1 z 2
(the positive branch is chosen)
A right triangle.
( )
= arcsin(z) = i ln iz + 1 z 2 Inverse trigonometric functions are useful when trying to
determine the remaining two angles of a right triangle
when the lengths of the sides of the triangle are known.
5 Applications Recalling the right-triangle denitions of sine, for exam-
ple, it follows that

5.1 General solutions

( )
opposite
Each of the trigonometric functions is periodic in the real = arcsin .
hypotenuse
part of its argument, running through all its values twice in
each interval of 2. Sine and cosecant begin their period Often, the hypotenuse is unknown and would need to
at 2k /2 (where k is an integer), nish it at 2k + be calculated before using arcsine or arccosine using the
/2, and then reverse themselves over 2k + /2 to 2k + Pythagorean Theorem: a2 + b2 = h2 where h is the
3/2. Cosine and secant begin their period at 2k, nish length of the hypotenuse. Arctangent comes in handy
it at 2k + , and then reverse themselves over 2k + to in this situation, as the length of the hypotenuse is not
2k + 2. Tangent begins its period at 2k /2, nishes needed.
it at 2k + /2, and then repeats it (forward) over 2k +
/2 to 2k + 3/2. Cotangent begins its period at 2k, ( )
nishes it at 2k + , and then repeats it (forward) over opposite
= arctan .
2k + to 2k + 2. adjacent
This periodicity is reected in the general inverses where For example, suppose a roof drops 8 feet as it runs out
k is some integer: 20 feet. The roof makes an angle with the horizontal,
where may be computed as follows:

sin(y) = x y = arcsin(x)+2k or y = arcsin(x)+2k ( ) ( ) ( )

opposite rise 8
sin(y) = x y = (1) arcsin(x) + k
k = arctan = arctan = arctan 21.8 .
adjacent run 20
cos(y) = x y = arccos(x)+2k or y = 2arccos(x)+2k
cos(y) = x y = arccos(x) + 2k 5.2 In computer science and engineering
tan(y) = x y = arctan(x) + k
cot(y) = x y = arccot(x) + k 5.2.1 Two-argument variant of arctangent
sec(y) = x y = arcsec(x)+2k or y = 2arcsec(x)+2k
Main article: atan2
csc(y) = x y = arccsc(x)+2k or y = arccsc(x)+2k
 7

The two-argument atan2 function computes the arctan- computer implementation (due to the limited number of
gent of y / x given y and x, but with a range of (, ]. In digits).[8] Similarly, arcsine is inaccurate for angles near
other words, atan2(y, x) is the angle between the positive /2 and /2.
x-axis of a plane and the point (x, y) on it, with posi-
tive sign for counter-clockwise angles (upper half-plane,
y > 0), and negative sign for clockwise angles (lower half-
plane, y < 0). It was rst introduced in many computer
6 See also
programming languages, but it is now also common in
other elds of science and engineering. Inverse exsecant
In terms of the standard arctan function, that is with
Inverse versine
range of (/2, /2), it can be expressed as follows:
Inverse hyperbolic function


arctan( xy ) x>0 List of integrals of inverse trigonometric functions



arctan( xy ) + y0, x<0


arctan( y ) y<0, x<0 List of trigonometric identities
atan2(y, x) = x

y>0, x=0

2 Trigonometric function


2

y<0, x=0


undened y=0, x=0

It also equals the principal value of the argument of the 7 References

complex number x + iy.
[1] For example Drrie, Heinrich (1965). Triumph der Math-
This function may also be dened using the tangent half-
ematik. Trans. David Antin. Dover. p. 69. ISBN 0-486-
angle formulae as follows:
61348-8.

( ) [2] Beach, Frederick Converse; Rines, George Edwin, eds.

y (1912). Inverse trigonometric functions. The Ameri-
atan2(y, x) = 2 arctan cana: a universal reference library. 21.
x2 + y 2 + x

provided that either x > 0 or y 0. However this fails if
given x 0 and y = 0 so the expression is unsuitable for
computational use.
The above argument order (y, x) seems to be the most
common, and in particular is used in ISO standards such
as the C programming language, but a few authors may
use the opposite convention (x, y) so some caution is war-
ranted. These variations are detailed at atan2.
5.2.2 Arctangent function with location parameter
In many applications the solution y of the equation x =
tan y is to come as close as possible to a given value
< < . The adequate solution is produced
by the parameter modied arctangent function
arctan(x)
y = arctan (x) := arctan(x) + rni . [7] Borwein, Jonathan; Bailey, David; Gingersohn, Roland
The function rni rounds to the nearest integer.
[3] Cajori, Florian (1919). A History of Mathematics (2 ed.).
New York, USA: The Macmillan Company. p. 272.
[4] Herschel, John Frederick William (1813). On a re-
markable Application of Cotess Theorem. Philosophical
Transactions. Royal Society, London. 103 (1): 8.
[5] Korn, Grandino Arthur; Korn, Theresa M. (2000) [1961].
21.2.4. Inverse Trigonometric Functions. Mathemati-
cal handbook for scientists and engineers: Denitions, the-
orems, and formulars for reference and review (3 ed.).
Mineola, New York, USA: Dover Publications, Inc. p.
811. ISBN 978-0-486-41147-7.
[6] Bhatti, Sanaullah; Nawab-ud-Din; Ahmed, Bashir;
Yousuf, S. M.; Taheem, Allah Bukhsh (1999). Dier-
entiation of Trigonometric, Logarithmic and Exponential
Functions. In Ellahi, Mohammad Maqbool; Dar, Kara-
mat Hussain; Hussain, Faheem. Calculus and Analytic Ge-
ometry (1 ed.). Lahore: Punjab Textbook Board. p. 140.
(2004). Experimentation in Mathematics: Computational
Paths to Discovery (1 ed.). Wellesley, MA:A K Peters. p.
51. ISBN 1-56881-136-5.

5.2.3 Numerical accuracy [8] Gade, Kenneth (2010). A non-singular horizontal po-
sition representation (PDF). The Journal of Naviga-
For angles near 0 and , arccosine is ill-conditioned and tion. Cambridge University Press. 63 (3): 395417.
will thus calculate the angle with reduced accuracy in a doi:10.1017/S0373463309990415.
8 8 EXTERNAL LINKS

8 External links
Weisstein, Eric W. Inverse Trigonometric Func-
tions. MathWorld.

Weisstein, Eric W. Inverse Tangent. MathWorld.

 9

9 Text and image sources, contributors, and licenses

9.1 Text
Inverse trigonometric functions Source: https://en.wikipedia.org/wiki/Inverse_trigonometric_functions?oldid=779619486 Contributors:
XJaM, Patrick, Michael Hardy, Stevenj, Dysprosia, Fibonacci, Robbot, Tea2min, Giftlite, Anville, SURIV, Daniel,levine, Pmanderson, Ab-
dull, Discospinster, Osrevad, Bender235, Zenohockey, Army1987, Jrme, Alansohn, Anthony Appleyard, Wtmitchell, StradivariusTV,
Armando, Gerbrant, Emallove, R.e.b., Kri, Glenn L, Salvatore Ingala, Chobot, Visor, DVdm, Algebraist, YurikBot, Wavelength, Scep-
tre, Hede2000, KSmrq, Grafen, Int 80h, NorsemanII, Bamse, RDBury, Maksim-e~enwiki, Thelukeeect, Eskimbot, Mhss, Mirokado,
JCSantos, PrimeHunter, Deathanatos, V1adis1av, BentSm, Saippuakauppias, Lambiam, Eridani, Ian Vaughan, ChaoticLlama, CapitalR,
JRSpriggs, Conrad.Irwin, HenningThielemann, Fommil, Rian.sanderson, Palmtree3000, Zalgo, Thijs!bot, Spikedmilk, Nonagonal Spider,
EdJohnston, BigJohnHenry, Luna Santin, Hannes Eder, Pichote, JAnDbot, Ricardo sandoval, Jetstreamer, JNW, Albmont, Gammy, Joer-
genB, Ac44ck, Gwern, Isamil, Mythealias, Pomte, Knorlin, TungstenWolfram, Hennessey, Patrick, Bobianite, BentonMiller, Sigmundur,
DavidCBryant, Alan U. Kennington, VolkovBot, Indubitably, LokiClock, Justtysen, VasilievVV, Riku92mr, Anonymous Dissident, Corvus
coronoides, Rdengler, Dmcq, Vertciel, Logan, CagedKiller360, Ck, Aly89, AlanUS, ClueBot, JoeHillen, Excirial, Bender2k14, Cenar-
ium, Leandropls, Kiensvay, Nikhilkrgvr, Aaron north, Dthomsen8, DaL33T, Addbot, Fgnievinski, Iceblock, Zarcadia, Sleepaholic, Jasper
Deng, Zorrobot, Luckas-bot, Yobot, Tohd8BohaithuGh1, Ptbotgourou, TaBOT-zerem, AnomieBOT, Archon 2488, JackieBot, Nickwee-
don, Geek1337~enwiki, Diego Queiroz, Txebixev, St.nerol, Hdullin, GrouchoBot, Uniwersalista, SassoBot, Prari, Nixphoeni, D'ohBot,
Kusluj, Emjaye, Number Googol, Serols, Double sharp, Trappist the monk, Adammerlinsmith, TjBot, Jowa fan, EmausBot, Mod-
Wilson, Velowiki, X-4-V-I, RA0808, Wham Bam Rock II, Dcirovic, ZroBot, Michael.YX.Wu, Isaac Euler, Tolly4bolly, Jay-Sebastos,
Colin.campbell.27, Maschen, ChuispastonBot, Rmashhadi, Anita5192, ClueBot NG, Matthiaspaul, Hdreuter, Helpful Pixie Bot, KLBot2,
Vagobot, Gar, Crh23, YatharthROCK, , StevinSimon, Tfr000, Modalanalytiker, Pratyya Ghosh, Ahmed Magdy Hosny,
Brirush, Yardimsever, Wamiq, Jerming, Mathmensch, Blackbombchu, Pqnlrn, DTL LAPOS, De Riban5, Monkbot, Cpt Wise, Arsenal
CR7, Cdserio99, Lond1984, Deacon Vorbis and Anonymous: 185

9.2 Images
File:Arcsecant_Arccosecant.svg Source: https://upload.wikimedia.org/wikipedia/commons/5/56/Arcsecant_Arccosecant.svg License:
CC BY-SA 3.0 Contributors: Own work Original artist: Geek3
File:Arcsine_Arccosine.svg Source: https://upload.wikimedia.org/wikipedia/commons/b/b4/Arcsine_Arccosine.svg License: CC BY-
SA 3.0 Contributors: Own work Original artist: Geek3
File:Arctangent_Arccotangent.svg Source: https://upload.wikimedia.org/wikipedia/commons/9/9a/Arctangent_Arccotangent.svg Li-
cense: CC BY-SA 3.0 Contributors: Own work Original artist: Geek3
File:Complex_ArcCot.jpg Source: https://upload.wikimedia.org/wikipedia/commons/6/60/Complex_ArcCot.jpg License: Public do-
main Contributors: Eigenes Werk (own work) made with mathematica Original artist: Jan Homann
File:Complex_ArcCsc.jpg Source: https://upload.wikimedia.org/wikipedia/commons/f/fd/Complex_ArcCsc.jpg License: Public do-
main Contributors: Eigenes Werk (own work) made with mathematica Original artist: Jan Homann
File:Complex_ArcSec.jpg Source: https://upload.wikimedia.org/wikipedia/commons/e/ec/Complex_ArcSec.jpg License: Public domain
Contributors: Eigenes Werk (own work) made with mathematica Original artist: Jan Homann
File:Complex_arccos.jpg Source: https://upload.wikimedia.org/wikipedia/commons/4/4d/Complex_arccos.jpg License: Public domain
Contributors: made with mathematica, own work Original artist: Jan Homann
File:Complex_arcsin.jpg Source: https://upload.wikimedia.org/wikipedia/commons/b/be/Complex_arcsin.jpg License: Public domain
Contributors: made with mathematica, own work Original artist: Jan Homann
File:Complex_arctan.jpg Source: https://upload.wikimedia.org/wikipedia/commons/f/f5/Complex_arctan.jpg License: Public domain
Contributors: made with mathematica, own work Original artist: Jan Homann
File:Sinus_und_Kosinus_am_Einheitskreis_1.svg Source: https://upload.wikimedia.org/wikipedia/commons/7/72/Sinus_und_
Kosinus_am_Einheitskreis_1.svg License: CC0 Contributors: Own work Original artist: Stephan Kulla (User:Stephan Kulla)
File:Trigonometric_functions_and_inverse.svg Source: https://upload.wikimedia.org/wikipedia/commons/0/0b/Trigonometric_
functions_and_inverse.svg License: CC0 Contributors: Own work Original artist: Maschen
File:Trigonometric_functions_and_inverse2.svg Source: https://upload.wikimedia.org/wikipedia/commons/d/d8/Trigonometric_
functions_and_inverse2.svg License: CC0 Contributors: Own work Original artist: Maschen
File:Trigonometric_functions_and_inverse3.svg Source: https://upload.wikimedia.org/wikipedia/commons/f/fa/Trigonometric_
functions_and_inverse3.svg License: CC0 Contributors: Own work Original artist: Maschen
File:Trigonometric_functions_and_inverse4.svg Source: https://upload.wikimedia.org/wikipedia/commons/2/24/Trigonometric_
functions_and_inverse4.svg License: CC0 Contributors: Own work Original artist: Maschen
File:Trigonometric_functions_and_inverse5.svg Source: https://upload.wikimedia.org/wikipedia/commons/1/17/Trigonometric_
functions_and_inverse5.svg License: CC0 Contributors: Own work Original artist: Maschen
File:Trigonometric_functions_and_inverse6.svg Source: https://upload.wikimedia.org/wikipedia/commons/6/61/Trigonometric_
functions_and_inverse6.svg License: CC0 Contributors: Own work Original artist: Maschen
File:Trigonometry_triangle.svg Source: https://upload.wikimedia.org/wikipedia/commons/7/7e/Trigonometry_triangle.svg License:
CC-BY-SA-3.0 Contributors: Transferred from en.wikipedia to Commons. Original artist: The original uploader was Tarquin at English
Wikipedia Later versions were uploaded by Limaner at en.wikipedia.

9.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Gudermannian function

Graph of the Gudermannian function Graph of the inverse Gudermannian function

The Gudermannian function, named after Christoph 1.3 Inverse

Gudermann (17981852), relates the circular func- x
tions and hyperbolic functions without explicitly using gd1 x = 1
dt /2 < x < /2
complex numbers. 0 cos t
   
It is dened for all x by[1][2][3]  1 + sin x  1  1 + sin x 

= ln   = ln  
cos x  2  1 sin x 
x 1  (
gd x = 0 cosh )
t dt. = ln |tan x + sec x| = ln tan 14 + 21 x 

= artanh(sin x) = arsinh(tan x)
= arcoth(csc x) = arcsch(cot x)
1 Properties
= sgn(x) arcosh(sec x) = sgn(x) arsech(cos x)
= i gd(ix)
1.1 Alternative denitions
(See inverse hyperbolic functions.)

gd x = arcsin (tanh x) = arctan(sinh x) = arccsc(coth x)

= sgn(x) arccos (sech x) = sgn(x) arcsec(cosh x) 1.4 Some identities
[ ( )]
= 2 arctan tanh 12 x sinh(gd1 x) = tan x; csch(gd1 x) = cot x;
= 2 arctan(e ) 2 .
x 1
cosh(gd1 x) = sec x; sech(gd1 x) = cos x;
tanh(gd1 x) = sin x; coth(gd1 x) = csc x.
Some related formula, such as arccot(csch x) , doesn't
quite work as denition. (See inverse trigonometric func-
tions.) 1.5 Derivatives
d d
gd x = sech x; gd1 x = sec x.
dx dx
1.2 Some identities

sin(gd x) = tanh x; csc(gd x) = coth x;

2 History
cos(gd x) = sech x; sec(gd x) = cosh x;
The function was introduced by Johann Heinrich Lam-
tan(gd x) = sinh x; cot(gd x) = csch x; bert in the 1760s at the same time as the hyperbolic func-
tan( 12 gd x) = tanh( 12 x). tions. He called it the transcendent angle, and it went

1
2 5 REFERENCES

by various names until 1862 when Arthur Cayley sug- 5 References

gested it be given its current name as a tribute to Gu-
dermanns work in the 1830s on the theory of special
functions.[4] Gudermann had published articles in Crelles
Journal that were collected in Theorie der potenzial- oder
cyklisch-hyperbolischen Functionen (1833), a book which
expounded sinh and cosh to a wide audience (under the
guises of Sin and Cos ).
The notation gd was introduced by Cayley[5] where he
starts by calling gd. u the inverse of the integral of the
secant function:
[1] Olver, F. W.J.; Lozier, D.W.; Boisvert, R.F.; Clark, C.W.,
eds. (2010), NIST Handbook of Mathematical Functions,
Cambridge University Press. Section 4.23(viii).
[2] CRC Handbook of Mathematical Sciences 5th ed. pp.
323325
[3] Weisstein, Eric W. Gudermannian. MathWorld.
[4] George F. Becker, C. E. Van Orstrand. Hyperbolic func-
tions. Read Books, 1931. Page xlix. Scanned copy avail-
able at archive.org

[5] Cayley, A. (1862). On the transcendent gd. u.

( ( ))
u= sec t dt = ln tan 14 + 12 Philosophical Magazine (4th ser.). 24: 1921.
0 doi:10.1080/14786446208643307 (inactive 2017-
01-16).
and then derives the denition of the transcendent:
[6] Osborne, P (2013), The Mercator projections, p74

( ( )) [7] John S. Robertson (1997). Gudermann and the Simple

gd u = i1 ln tan 14 + 12 ui Pendulum. The College Mathematics Journal. 28 (4):
271276. JSTOR 2687148. Review.
observing immediately that it is a real function of u.
[8] Good, Michael R. R.; Anderson, Paul R.; Evans, Charles
R. (2013). Time dependence of particle creation from
accelerating mirrors. Physical Review D. 88 (2): 025023.
3 Applications
arXiv:1303.6756 . Bibcode:2013PhRvD..88b5023G.
doi:10.1103/PhysRevD.88.025023.
The angle of parallelism function in hyperbolic ge-
ometry is dened by

1
2 gd x

On a Mercator projection a line of constant latitude

is parallel to the equator (on the projection) and is
displaced by an amount proportional to the inverse
Gudermannian of the latitude.

The Gudermannian (with a complex argument) may

be used in the denition of the transverse Mercator
projection.[6]

The Gudermannian appears in a non-periodic solu-

tion of the inverted pendulum.[7]

The Gudermannian also appears in a moving mirror

solution of the dynamical Casimir eect.[8]

4 See also
Hyperbolic secant distribution

Mercator projection

Tangent half-angle formula

Tractrix

Trigonometric identity
 3

6 Text and image sources, contributors, and licenses

6.1 Text
Gudermannian function Source: https://en.wikipedia.org/wiki/Gudermannian_function?oldid=776487256 Contributors: Michael Hardy,
AugPi, Smack, Dysprosia, Sverdrup, Enochlau, Giftlite, Fropu, Jason Quinn, Macrakis, MarkSweep, Burn, Linas, Marudubshinki, Ober-
sachse, NeoUrfahraner, George Burgess, Salix alba, Salvatore Ingala, WriterHound, KSmrq, Selfworm, Maksim-e~enwiki, InverseHy-
percube, Kaimbridge, Darth Panda, CRGreathouse, Xanthoxyl, Michaelas10, Thijs!bot, CharlotteWebb, User77764, Aaron Rotenberg,
SieBot, Ck, Rumping, Crowsnest, Kiensvay, Kal-El-Bot, Addbot, Fgnievinski, Quadrescence, Hombre1729, Geek1337~enwiki, Citation
bot, Txebixev, RibotBOT, Peter Mercator, Jonesey95, Trappist the monk, Akjz, ZroBot, Ebrambot, AManWithNoPlan, Matthiaspaul,
Titodutta, BG19bot, Boehm, Bender the Bot, Deacon Vorbis and Anonymous: 23

6.2 Images
File:Lock-green.svg Source: https://upload.wikimedia.org/wikipedia/commons/6/65/Lock-green.svg License: CC0 Contributors: en:File:
Free-to-read_lock_75.svg Original artist: User:Trappist the monk
File:Mplwp_gudermann_inv_piaxis.svg Source: https://upload.wikimedia.org/wikipedia/commons/f/fd/Mplwp_gudermann_inv_
piaxis.svg License: CC BY 3.0 Contributors: Own work Original artist: Geek3
File:Mplwp_gudermann_piaxis.svg Source: https://upload.wikimedia.org/wikipedia/commons/9/95/Mplwp_gudermann_piaxis.svg Li-
cense: CC BY 3.0 Contributors: Own work Original artist: Geek3

6.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Error function

no greater than x. This is true for any random variable

1.00
with distribution N (0, 12 ) ; but the application to error
0.75
variables is how the error function got its name.
0.50
The previous paragraph can be generalized to any vari-
0.25
ance: given a variable (such as an unbiased error vari-
erf (x)

( N
2
0.00
able) ) (0, ) , evaluating the error function at
0.25
erf x
1
2
describes the probability of falling in the
0.50
[3]
range [x, x]. This is used in statistics to predict behav-
0.75
ior of any sample with respect to the population mean.
1.00 This usage is similar to the Q-function, which in fact can
3 2 1 0
x
1 2 3
be written in terms of the error function.

Plot of the error function

2 Derived and related functions
In mathematics, the error function (also called the
Gauss error function) is a special function (non- 2.1 Complementary error function
elementary) of sigmoid shape that occurs in probability,
statistics, and partial dierential equations describing dif- The complementary error function, denoted erfc , is
fusion. It is dened as:[1][2] dened as

x
1 erfc(x) = 1 erf(x)
et dt
2
erf(x) =
x
x =
2
et dt
2
2
et dt.
2
= x
0
= ex erfcx(x),
2

In statistics, for nonnegative values of x, the error function

has the following interpretation: for a random variable X which also denes erfcx , the scaled complementary
that is normally distributed with mean 0 and variance 12 , error function[4] (which can be used instead of erfc
erf(x) describes the probability of X falling in the range to avoid arithmetic underow[4][5] ). Another form of
[x, x]. erfc(x) for non-negative x is known as Craigs formula:[6]

( )
1 The name 'error function' erfc(x | x 0) =
2 /2
exp 2
x2
d.
0 sin
The error function is used in measurement theory (using
probability and statistics), and its use in other branches of 2.2 Imaginary error function
mathematics is typically unrelated to the characterization
of measurement errors. The imaginary error function, denoted er, is dened
In statistics, it is common to have a variable Y and its as
unbiased estimator Y . The error is then dened as
= Y Y . This makes the error a normally dis-
tributed random variable with mean 0 (because the esti- erfi(x) = i erf(ix)
mator is unbiased) and some variance 2 ; this is written x
2 2
as N (0, 2 ) . For the case where 2 = 21 , i.e. an = et dt
unbiased error variable N (0, 21 ) , erf(x) describes 0
the probability of the error falling in the range [x, x]; 2 2

in other words, the probability that the absolute error is = ex D(x),

1
2 3 PROPERTIES

where D(x) is the Dawson function (which can be used

instead of er to avoid arithmetic overow[4] ).
Despite the name imaginary error function, erfi(x) is
real when x is real.
When the error function is evaluated for arbitrary
complex arguments z, the resulting complex error func-
tion is usually discussed in scaled form as the Faddeeva
function:

w(z) = ez erfc(iz) = erfcx(iz).

2

erf(z)

The property erf(z) = erf(z) means that the error

function is an odd function. This directly results from the
fact that the integrand et is an even function.
2

2.3 Cumulative distribution function For any complex number z:

The error function is related to the cumulative distribution

, the integral of the standard normal distribution, by[2] erf(z) = erf(z)
where z is the complex conjugate of z.
The integrand = exp(z2 ) and = erf(z) are shown in
the complex z-plane in gures 2 and 3. Level of Im()
1 1 ( ) 1 ( ) = 0 is shown with a thick green line. Negative integer
(x) = + erf x/ 2 = erfc x/ 2 . values of Im() are shown with thick red lines. Positive
2 2 2
integer values of Im(f) are shown with thick blue lines.
Intermediate levels of Im() = constant are shown with
thin green lines. Intermediate levels of Re() = constant
are shown with thin red lines for negative values and with
thin blue lines for positive values.
The error function at + is exactly 1 (see Gaussian inte-
3 Properties gral). At the real axis, erf(z) approaches unity at z +
and 1 at z . At the imaginary axis, it tends to i.

Plots in the complex plane

3.1 Taylor series
The error function is an entire function; it has no singu-
larities (except that at innity) and its Taylor expansion
always converges.
The dening integral cannot be evaluated in closed form
in terms of elementary functions, but by expanding the
2
integrand ez into its Maclaurin series and integrating
term by term, one obtains the error functions Maclaurin
series as:

(
2 (1)n z 2n+1 2 z3 z5 z7 z9
erf(z) = = z + +
n=0 n!(2n + 1) 3 10 42 216

Integrand which holds for every complex number z. The denomi-

exp(z2 ) nator terms are sequence A007680 in the OEIS.
3.3 Brmann series 3

For iterative calculation of the above series, the following 3.3 Brmann series
alternative formulation may be useful:
An expansion,[8] which converges more rapidly for all real
values of x than a Taylor expansion, is obtained by using

( n ) [9]
2 (2k 1)z 2
2 zHans
nHeinrich
z 2 Brmann's theorem:
erf(z) = z =
n=0 k(2k + 1) n=0 2n + 1 k
k=1 k=1
erf(x)
because (2k1)z
2

k(2k+1) expresses the multiplier to turn the (

k term into the (k + 1)th term (considering z as the rst
th 2 1 7
1 (1 ex ) (1 ex )2
2 2
= sgn(x) 1 ex2
term). 12 480
The imaginary error function has a very similar Maclaurin 5 787
(1 ex )3 (1 ex
2

series, which is:

896 276480
( )
( 2 )


7 9 x kx2
z sgn(x) 1 e
2
2 z 2n+1 2 z3 z5 z= + ck e .
erfi(z) = = z+ + + + + 2
k=1
n=0 n!(2n + 1) 3 10 42 216
By keeping only the rst two coecients and choosing
which holds for every complex number z. c1 = 20031
and c2 = 8000
341
, the resulting approximation
shows its largest relative error at x = 1.3796 , where it
is less than 3.6127 103 :
3.2 Derivative and integral
( )
The derivative of the error function follows immediately erf(x) 2 sgn(x) 1 ex2
+
31 x2
e
341 2x2
e .
from its denition: 2 200 8000

3.4 Inverse functions

d 2
erf(z) = ez .
2

dz

From this, the derivative of the imaginary error function

is also immediate:

d 2 2
erfi(z) = ez .
dz

An antiderivative of the error function, obtainable by

integration by parts, is

ez
2

z erf(z) + . Inverse error function

An antiderivative of the imaginary error function, also ob- Given complex number z, there is not a unique complex
tainable by integration by parts, is number w satisfying erf(w) = z , so a true inverse func-
tion would be multivalued. However, for 1 < x < 1,
1
there
( is 1
a unique
) real number denoted erf (x) satisfying
2 erf erf (x) = x .
ez
z erfi(z) . The inverse error function is usually dened with do-

main (1,1), and it is restricted to this domain in many
Higher order derivatives are given by computer algebra systems. However, it can be extended
to the disk |z| < 1 of the complex plane, using the Maclau-
rin series
k1 k1 ( )
2(1) 2 d
Hk1 (z)ez = z
2 2
erf(k) (z) = e , k = 1, 2, . . .
dz k1 ( )2k+1
1 ck
[7]
erf (z) = z ,
where H are the physicists Hermite polynomials. 2k + 1 2
k=0
4 4 APPROXIMATION WITH ELEMENTARY FUNCTIONS

( )
RN (x) = O x12N ex as x .
2
where c0 = 1 and

k1 { Indeed,}the exact value of the remainder is
cm ck1m 7 127 4369 34807
ck = = 1, 1, , , , ,... .
m=0
(m + 1)(2m + 1) 6 90 2520 16200

So we have the series expansion (note that common fac- (1)N 12N (2N )! 2N t2
2
tors have been canceled from numerators and denomina- RN (x) := N! x
t e dt,
tors):
by induction, writing et =
2
which follows
( 2 )easily

( 4 1 et
)
1 3 7 2 5 127 3 7 (2t)
4369 34807 and5 integrating by parts.
1
erf (z) = 2 z + z + z + z + 9
z + z + .
11
12 480 40320 5806080 182476800
For large enough values of x, only the rst few terms of
(After cancellation the numerator/denominator fractions this asymptotic expansion are needed to obtain a good
are entries A092676/ A132467 in the OEIS; with- approximation of erfc(x) (while for not too large values
out cancellation the numerator terms are given in entry of x note that the above Taylor expansion at 0 provides a
A002067.) Note that the error functions value at is very fast convergence).
equal to 1.
( )
For |z| < 1, we have erf erf1 (z) = z . 3.6 Continued fraction expansion
The inverse complementary error function is dened
as A continued fraction expansion of the complementary er-
ror function is:[11]

erfc1 (1 z) = erf1 (z).

z 1 m
real number erfi1 (x) sat- erfc(z) = ez
2
For real x, there
( 1 is a unique
) am = .
isfying erfi erfi (x) = x . The inverse imaginary a1 2
z2 +
1 [10] a2
error function is dened as erfi (x) . 1+
a3
For any real x, Newtons method can be used to compute z2 +
1
erfi (x) , and for 1 x 1 , the following Maclau- 1 +
rin series converges:
3.7 Integral of error function with Gaus-

( )2k+1 sian density function
(1)k ck
erfi1 (z) = z ,
2k + 1 2 [ ] [ ]
k=0 b ac erf (ax + b) (x + c)2
erf = exp dx, a, b, c, d
where ck is dened as above. 1 + 2a2 d2 2d2 2d2

3.5 Asymptotic expansion 4 Approximation with elementary

A useful asymptotic expansion of the complementary er- functions
ror function (and therefore also of the error function) for
large real x is Abramowitz and Stegun give several approximations of
varying accuracy (equations 7.1.2528). This allows one
[
] to choose the fastest approximation suitable for a given
ex
2
1 3 5 (2n 1) ex In order
2
(2nof increasing
1)!!
erfc(x) = 1 + (1)n = application.
(1)n , accuracy, they are:
x (2x 2 )n x (2x 2 )n
n=1 n=0

where (2n 1)!! is the double factorial: the product of erf(x) 1

all odd numbers up to (2n 1). This series diverges for
1
(1+a1 x+a2 x2 +a3 x3 +a4 x4 )4 , x 0
every nite x, and its meaning as asymptotic expansion is 4
(maximum error: 510 )
that, for any N N one has
where a1 = 0.278393, a2 = 0.230389, a3 = 0.000972, a4

N 1 = 0.078108
ex
2
(2n 1)!!
erfc(x) = (1)n + RN (x)
x (2x2 )n
erf(x) 1 (a1 t + a2 t2 + a3 t3 )ex , t =
2
n=0

where the remainder, in Landau notation, is 1

1+px , x 0 (maximum error: 2.5105 )
 5

where p = 0.47047, a1 = 0.3480242, a2 = 0.0958798,

a3 = 0.7478556
2e 1 x2
erfc(x) e , x 0, > 1,

erf(x) 1
1
, x 0 where the parameter can be picked to minimize error
(1+a1 x+a2 x2 ++a6 x6 )16
on the desired interval of approximation.
(maximum error: 3107 )

where a1 = 0.0705230784, a2 = 0.0422820123, a3 =

0.0092705272, a4 = 0.0001520143, a5 = 0.0002765672,
5 Numerical approximations
a6 = 0.0000430638
Over the complete range of values, there is an approxima-
tion with a maximal error of 1.2 107 , as follows:[15]
erf(x) 1 (a1 t + a2 t2 + +
a5 t5 )ex ,
2
1
t = 1+px (maximum error:
{
1.5107 ) 1 forx 0
erf(x) =
1 forx < 0
where p = 0.3275911, a1 = 0.254829592, a2 =
0.284496736, a3 = 1.421413741, a4 = 1.453152027, with
a5 = 1.061405429
(
All of these approximations are valid for x 0. To = t exp x2 1.26551223 + 1.00002368t + 0.37409196t2 + 0.096
use these approximations for negative x, use the fact that
0.18628806t4 + 0.27886807t5 1.13520398t6 + 1.48851587t7
erf(x) is an odd function, so erf(x) = erf(x). )
0.82215223t8 + 0.17087277t9
Another approximation is given by
and
( )
4
+ ax2 1
erf(x) sgn(x) 1 exp x2 t= .
1 + ax2 1 + 0.5|x|
where Also, over the complete range of values, the following
simple approximation holds for x erfc(x) , with a maxi-
mal error of 6.3 104 :
8( 3)
a= 0.140012.
3(4 ) ( )
x erfc(x) = 1.3693x exp 0.8072(x + 0.6388)2
This is designed to be very accurate in a neighborhood
of 0 and a neighborhood of innity, and the error is less
than 0.00035 for all x. Using the alternate value a 0.147 6 Applications
reduces the maximum error to about 0.00012.[12]
This approximation can also be inverted to calculate the When the results of a series of measurements are de-
inverse error function: scribed by a normal distribution ( with standard
) deviation
a
and expected value 0, then erf 2 is the probabil-

v ity that the error of a single measurement lies between

u ( )2 ( for positive a. )
u ln(1 x ) ln(1 xa
2 ) and +a, ln(1 x2 ) This is useful, for example, in
1 t 2 2 2
erf (x) sgn(x) + + rate. of a digital communication
a 2 a determininga the bit error
2
system.
Exponential bounds and a pure exponential approxima- The error and complementary error functions occur, for
tion for the complementary error function are given by example, in solutions of the heat equation when boundary
[13]
conditions are given by the Heaviside step function.
The error function and its approximations can be used to
1 2x2 1 x2 estimate results that hold with high probability. Given
ex ,
2
erfc(x) e + e x>0 random variable X Norm[, ] and constant L < :
2 2
1 1 4 2
erfc(x) ex + e 3 x ,
2
x > 0. ( ) ( ( )2 )
6 2
1 1 L L
Pr[X L] = + erf A exp B
A single-term lower bound is[14] 2 2 2
6 7 RELATED FUNCTIONS

where A and B are certain numeric constants.

If L is suf- ( ) ( )
ciently far from the mean, i.e. L ln k , then: 1 2 sgn(x) 1 2
erf(x) = sgn(x)P ,x = ,x .
2 2
A sgn(x) is the sign function.
Pr[X L] A exp(B ln k) =
kB
so the probability goes to 0 as k . 7.1 Generalized error functions

n=5

7 Related functions 3
n=4
n=3

The error function is essentially identical to the standard 1

n=2
n=1
normal cumulative distribution function, denoted , also

En (x)
0
named norm(x) by software languages, as they dier only
by scaling and translation. Indeed, 1

x [ ( )] (3 )
1 t2 1 x 1 x
(x) = e 2 dt = 1 + erf = erfc
2 2 2 2 3 2 2 1 0 1 2 3

x
or rearranged for erf and erfc:
Graph of generalised error functions En(x):
grey curve: E1 (x) = (1 e x )/
( )
red curve: E2 (x) = erf(x)
erf(x) = 2 x 2 1
green curve: E3 (x)
( ) ( ( ))
blue curve: E4 (x)
erfc(x) = 2 x 2 = 2 1 x 2 .
gold curve: E5 (x).

Consequently, the error function is also closely related

Some authors discuss the more general functions:
to the Q-function, which is the tail probability of the
standard normal distribution. The Q-function can be ex-
x
pressed in terms of the error function as n! n! xnp+1
et dt =
n
En (x) = (1)p .
0 p=0 (np + 1)p!
( ) ( )
1 1 x 1 x Notable cases are:
Q(x) = erf = erfc .
2 2 2 2 2
E 0 (x) is a straight line through the origin: E0 (x) =
The inverse of is known as the normal quantile func- x

e
tion, or probit function and may be expressed in terms of
the inverse error function as E 2 (x) is the error function, erf(x).

After division by n!, all the En for odd n look similar (but
not identical) to each other. Similarly, the En for even n
probit(p) = 1 (p) = 2 erf1 (2p1) = 2 erfc1 (2p).
look similar (but not identical) to each other after a simple
The standard normal cdf is used more often in probability division by n!. All generalised error functions for n > 0
and statistics, and the error function is used more often in look similar on the positive x side of the graph.
other branches of mathematics. These generalised functions can equivalently be ex-
The error function is a special case of the Mittag-Leer pressed for x > 0 using the Gamma function and
function, and can also be expressed as a conuent hyper- incomplete Gamma function:
geometric function (Kummers function):
( ( ) ( ))
1 1 1 n
En (x) = (n) ,x , x > 0.
( ) n n
2x 1 3
erf(x) = F1 , , x2 . Therefore, we can dene the error function in terms of
1 2 2
the incomplete Gamma function:
It has a simple expression in terms of the Fresnel integral.
( )
In terms of the regularized Gamma function P and the 1 1 2
incomplete gamma function, erf(x) = 1 ,x .
2
 7

7.2 Iterated integrals of the complemen- Go: Provides math.Erf() and math.Erfc() for oat64
tary error function arguments.

The iterated integrals of the complementary error func- Google search: Googles search also acts as a cal-
tion are dened by culator and will evaluate erf(...)" and erfc(...)" for
real arguments.
Haskell: An erf package[18] exists that provides a
in erfc(z) = in1 erfc() d. typeclass for the error function and implementations
z
for the native (real) oating point types.
They have the power series
IDL: provides both erf and erfc for real and complex

arguments.
(z)j
in erfc(z) = ( ),
2nj j! 1+ nj Java: Apache commons-math[19] provides imple-
j=0 2
mentations of erf and erfc for real arguments.
from which follow the symmetry properties
Julia: Includes erf and erfc for real and complex ar-
guments. Also has er for calculating i erf(ix)
m
z 2q
i2m erfc(z) = i2m erfc(z)+
q=0
22(mq)1 (2q)!(m q)! Maple: Maple implements both erf and erfc for real
and complex arguments.
and
MathCAD provides both erf(x) and erfc(x) for real
arguments.
m
z 2q+1
i2m+1 erfc(z) = i2m+1 erfc(z)+ Mathematica:
. erf is implemented as Erf and Erfc in
22(mq)1 (2q + 1)!(m q)!
q=0 Mathematica for real and complex arguments, which
are also available in Wolfram Alpha.

8 Implementations
C: C99 provides the functions double erf(double x)
and double erfc(double x) in the header math.h. The
pairs of functions {er(),erfcf()} and {er(),erfcl()}
take and return values of type oat and long dou-
ble respectively. For complex double arguments, the
function names cerf and cerfc are reserved for fu-
ture use"; the missing implementation is provided
by the open-source project libcerf, which is based
on the Faddeeva package.
C++: C++11 provides erf() and erfc() in the header
cmath. Both functions are overloaded to accept ar-
guments of type oat, double, and long double. For
complex<double>, the Faddeeva package provides
a C++ complex<double> implementation.
D: A D package[16] exists providing ecient and ac-
curate implementations of complex error functions,
along with Dawson, Faddeeva, and Voigt functions.
Excel: Microsoft Excel provides the erf, and the erfc
functions, nonetheless both inverse functions are not
in the current library.[17]
Fortran: The Fortran 2008 standard provides the
ERF, ERFC and ERFC_SCALED functions to cal-
culate the error function and its complement for real
arguments. Fortran 77 implementations are avail-
able in SLATEC.
Matlab provides both erf and erfc for real argu-
ments, also via W. J. Codys algorithm.[20]
Maxima provides both erf and erfc for real and com-
plex arguments.
PARI/GP: provides erfc for real and complex argu-
ments, via tanh-sinh quadrature plus special cases.
Perl: erf (for real arguments, using Codys
algorithm[20] ) is implemented in the Perl module
Math::SpecFun
Python: Included since version 2.7 as math.erf()
and math.erfc() for real arguments. For previ-
ous versions or for complex arguments, SciPy
includes implementations of erf, erfc, er, and
related functions for complex arguments in
scipy.special.[21] A complex-argument erf is also in
the arbitrary-precision arithmetic mpmath library
as mpmath.erf()
R: The so-called 'error function'"[22] is not provided
directly, but is detailed as an example of the normal
cumulative distribution function (?pnorm), which is
based on W. J. Codys rational Chebyshev approxi-
mation algorithm.[20]
Ruby: Provides Math.erf() and Math.erfc() for real
arguments.
8 11 FURTHER READING
9 See also
9.1 Related functions
Gaussian integral, over the whole real line
Gaussian function, derivative
Dawson function, renormalized imaginary error
function
GoodwinStaton integral
9.2 In probability
Normal distribution
Normal cumulative distribution function, a scaled
and shifted form of error function
Probit, the inverse or quantile function of the normal
CDF
Q-function, the tail probability of the normal distri-
bution
10 References
[1] Andrews, Larry C.; Special functions of mathematics for
engineers
[2] Greene, William H.; Econometric Analysis (fth edition),
Prentice-Hall, 1993, p. 926, fn. 11
[3] Van Zeghbroeck, Bart; Principles of Semiconductor De-
vices, University of Colorado, 2011.
[4] Cody, W. J. (March 1993), Algorithm 715:
SPECFUNA portable FORTRAN package of
special function routines and test drivers (PDF),
ACM Trans. Math. Softw., 19 (1): 2232,
doi:10.1145/151271.151273
[5] Zaghloul, M. R. (March 1, 2007), On the calcula-
tion of the Voigt line prole: a single proper inte-
gral with a damped sine integrand, Monthly Notices of
the Royal Astronomical Society, 375 (3): 10431048,
doi:10.1111/j.1365-2966.2006.11377.x
[6] John W. Craig, A new, simple and exact result for calculat-
ing the probability of error for two-dimensional signal con-
stellations, Proc. 1991 IEEE Military Commun. Conf.,
vol. 2, pp. 571575.
[7] Wolfram MathWorld
[8] H. M. Schpf and P. H. Supancic, On Brmanns Theo-
rem and Its Application to Problems of Linear and Non-
linear Heat Transfer and Diusion, The Mathematica
Journal, 2014. doi:10.3888/tmj.1611.Schpf, Supancic
[9] E. W. Weisstein. Brmanns Theorem from Wolfram
MathWorldA Wolfram Web Resource./ E. W. Weis-
stein
[10] Bergsma, Wicher. On a new correlation coecient, its
orthogonal decomposition and associated tests of inde-
pendence (PDF).
[11] Cuyt, Annie A. M.; Petersen, Vigdis B.; Verdonk,
Brigitte; Waadeland, Haakon; Jones, William B. (2008).
Handbook of Continued Fractions for Special Functions.
Springer-Verlag. ISBN 978-1-4020-6948-2.
[12] Winitzki, Sergei (6 February 2008). A handy approxi-
mation for the error function and its inverse (PDF). Re-
trieved 2011-10-03.
[13] Chiani, M., Dardari, D., Simon, M.K. (2003). New
Exponential Bounds and Approximations for the Compu-
tation of Error Probability in Fading Channels. IEEE
Transactions on Wireless Communications, 4(2), 840
845, doi=10.1109/TWC.2003.814350.
[14] Chang, Seok-Ho; Cosman, Pamela C.; Milstein, Lau-
rence B. (November 2011). Cherno-Type Bounds
for the Gaussian Error Function. IEEE Transac-
tions on Communications. 59 (11): 29392944.
doi:10.1109/TCOMM.2011.072011.100049.
[15] Numerical Recipes in Fortran 77: The Art of Scien-
tic Computing (ISBN 0-521-43064-X), 1992, page 214,
Cambridge University Press.
[16] DlangScience/libcerf, A package for use with the D Pro-
gramming language.
[17] These results can however be obtained using the
NormSInv function as follows: erf_inverse(p) = -
NormSInv((1 - p)/2)/SQRT(2); erfc_inverse(p) = -
NormSInv(p/2)/SQRT(2). See .
[18] http://hackage.haskell.org/package/erf
[19] Commons Math: The Apache Commons Mathematics Li-
brary
[20] Cody, William J. (1969). Rational Chebyshev Approx-
imations for the Error Function (PDF). Math. Comp.
23 (107): 631637. doi:10.1090/S0025-5718-1969-
0247736-4.
[21] Error Function and Fresnel Integrals, SciPy v0.13.0 Ref-
erence Guide.
[22] R Development Core Team (25 February 2011), R: The
Normal Distribution
11 Further reading
Abramowitz, Milton; Stegun, Irene Ann, eds.
(1983) [June 1964]. Chapter 7. Handbook of
Mathematical Functions with Formulas, Graphs, and
Mathematical Tables. Applied Mathematics Se-
ries. 55 (Ninth reprint with additional corrections
of tenth original printing with corrections (Decem-
ber 1972); rst ed.). Washington D.C.; New York:
United States Department of Commerce, National
Bureau of Standards; Dover Publications. p. 297.
ISBN 978-0-486-61272-0. LCCN 64-60036. MR
0167642. LCCN 65-12253.
 9

Press, William H.; Teukolsky, Saul A.; Vetterling,

William T.; Flannery, Brian P. (2007), Section 6.2.
Incomplete Gamma Function and Error Function,
Numerical Recipes: The Art of Scientic Computing
(3rd ed.), New York: Cambridge University Press,
ISBN 978-0-521-88068-8
Temme, Nico M. (2010), Error Functions, Daw-
sons and Fresnel Integrals, in Olver, Frank W.
J.; Lozier, Daniel M.; Boisvert, Ronald F.; Clark,
Charles W., NIST Handbook of Mathematical Func-
tions, Cambridge University Press, ISBN 978-
0521192255, MR 2723248

12 External links
MathWorld Erf
A Table of Integrals of the Error Functions
10 13 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES

13 Text and image sources, contributors, and licenses

13.1 Text
Error function Source: https://en.wikipedia.org/wiki/Error_function?oldid=781069539 Contributors: AxelBoldt, Bryan Derksen, Tar-
quin, Michael Hardy, Chinju, Cyp, Stevenj, BenKovitz, Gutza, Wilke, Finlay McWalter, LaurentPerrinet, Chris Roy, Tualha, Safor-
rest, Wile E. Heresiarch, Tea2min, Giftlite, Lethe, Jorge Stol, Sietse, Nayuki, Vivero~enwiki, Alberto da Calvairate~enwiki, Nickptar,
Urhixidur, Fintor, Askewchan, Chepry, Rich Farmbrough, MuDavid, Bender235, Zaslav, Kwamikagami, Shanes, Rbj, Larryv, Danski14,
Sligocki, PAR, RJFJR, Oleg Alexandrov, Waabu, Linas, Gisling, Karam.Anthony.K, Qwertyus, Rjwilmsi, Strait, Salix alba, R.e.b., Erk-
can, Alejo2083, FlaBot, Ausinha, Scythe33, Chet Gray, Chobot, Meawoppl, Peterl, Dstrozzi, YurikBot, Nbrouard, Pnrj, Crasshopper,
Dbrs, Entropeneur, NormDor, Arthur Rubin, Pred, Zvika, SmackBot, BahramH, GoOdCoNtEnT, Anachronist, Bh3u4m, Oli Filth, Papa
November, Nbarth, Cybercobra, Cubbi, Mwtoews, PSeibert~enwiki, Sbmehta, Texas Dervish, Domitori, GeordieMcBain, CRGreathouse,
Doomed Rasher, Thijs!bot, , Headbomb, AllUltima, Lklundin, Xypron, Catslash, Albmont, QrczakMK, Baccyak4H, Beabroad, KCon-
Wiki, Tbleher, Sullivan.t.j, Falcor84, Dima373, Ricardogpn, Pedrito, Uncle Dick, Leaet, Liangent, Salih, Stan J Klimas, KohanX, The-
owoll, Epistemenical, DonAndre, Sheliak, X!, VolkovBot, PMajer, TXiKiBoT, Anonymous Dissident, Stafusa, Aaron Rotenberg, Jamelan,
Flyer22 Reborn, JackSchmidt, Svick, AlanUS, Melcombe, Headlessplatter, Crgato, ClueBot, UKoch, Njasloane, P. M. Sakkas, IMneme,
Pkeastman, Qwfp, DumZiBoT, AlexFekken, Addbot, Michele.allegra, Fgnievinski, Favonian, Jasper Deng, Luckas-bot, Yobot, Wateenel-
lende, AnomieBOT, Erel Segal, Joule36e5, Hegpetz, Geek1337~enwiki, Jacobakkerboom, Rubenstreb, ChrisKuklewicz, MathHisSci,
Ichbin-dcw, SkyMachine, Duoduoduo, Olawlor, Thompson2212, Jxwx, John of Reading, Gz33, Zueignung, ClueBot NG, BarrelProof, Cn-
tras, Dru of Id, Hikenstu, Helpful Pixie Bot, Bgde, BG19bot, ServiceAT, Nsda, Hypnotoad33, Brad7777, Jagan seshadri, ChrisGualtieri,
Qsq, Kondormari, Katterjohn, EmileContal, Kogge, Pscrape, Robdonne, Mchiani, Rakmath15, DoctorTerrella, AntoineDeRivarol, Ab-
solutelypuremilk, KasparBot, Boehm, Cdserio99, Dmt137, Bender the Bot, Deacon Vorbis and Anonymous: 191

13.2 Images
File:ComplexErf.jpg Source: https://upload.wikimedia.org/wikipedia/commons/0/00/ComplexErf.jpg License: CC BY 3.0 Contributors:
This mathematical image was created with Mathematica Original artist: Domitori
File:ComplexEx2.jpg Source: https://upload.wikimedia.org/wikipedia/commons/1/18/ComplexEx2.jpg License: CC BY-SA 3.0 Con-
tributors: This mathematical image was created with Mathematica Original artist: Domitori
File:Error_Function.svg Source: https://upload.wikimedia.org/wikipedia/commons/2/2f/Error_Function.svg License: Public domain
Contributors: self-made, Inkscape Original artist: Inductiveload
File:Error_Function_Generalised.svg Source: https://upload.wikimedia.org/wikipedia/commons/0/02/Error_Function_Generalised.
svg License: Public domain Contributors: self-made, Inkscape and Mathematica Original artist: Inductiveload
File:Mplwp_erf_inv.svg Source: https://upload.wikimedia.org/wikipedia/commons/e/ec/Mplwp_erf_inv.svg License: CC BY 3.0 Con-
tributors: Own work Original artist: Geek3
File:OEISicon_light.svg Source: https://upload.wikimedia.org/wikipedia/commons/d/d8/OEISicon_light.svg License: Public do-
main Contributors: Own work Original artist: <a href='//commons.wikimedia.org/wiki/File:Watchduck.svg' class='image'><img
alt='Watchduck.svg' src='https://upload.wikimedia.org/wikipedia/commons/thumb/d/d8/Watchduck.svg/40px-Watchduck.svg.png'
width='40' height='46' srcset='https://upload.wikimedia.org/wikipedia/commons/thumb/d/d8/Watchduck.svg/60px-Watchduck.svg.png
1.5x, https://upload.wikimedia.org/wikipedia/commons/thumb/d/d8/Watchduck.svg/80px-Watchduck.svg.png 2x' data-le-width='703'
data-le-height='806' /></a> Watchduck (a.k.a. Tilman Piesk)

13.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Generalised logistic function

A=0, K=1, B=3, Q==0.5, M=0, C=1

Eect of varying parameter B. A = 0, all other parameters are

1.

Eect of varying parameter A. All other parameters are 1.

The generalised logistic function or curve, also known

as Richards curve, originally developed for growth
modelling, is an extension of the logistic or sigmoid func-
tions, allowing for more exible S-shaped curves: Eect of varying parameter C. A = 0, all other parameters are
1.

K A K : the upper asymptote. If A = 0 then K is called

Y (t) = A +
(C + QeBt )1/ the carrying capacity;
where Y = weight, height, size etc., and t = time.
B : the growth rate;
It has ve parameters:
> 0 : aects near which asymptote maximum
A : the lower asymptote; growth occurs.

1
2 1 GENERALISED LOGISTIC DIFFERENTIAL EQUATION

Eect of varying parameter K. A = 0, all other parameters are Eect of varying parameter . A = 0, all other parameters are
1. 1.

this representation simplies the setting of both a starting

time and the value of Y at that time.
The logistic, with maximum growth rate at time M , is
the case where Q = = 1.

1 Generalised logistic dierential

equation
A particular case of the generalised logistic function is:

K
Y (t) =
(1 + Qe(tt0 ) )1/
which is the solution of the so-called Richards dierential
equation (RDE):
Eect of varying parameter Q. A = 0, all other parameters are
1.
( ( ) )
Y
Y (t) = 1 Y
Q : is related to the value Y (0) K
C : typically takes a value of 1. with initial condition

The equation can also be written:

Y (t0 ) = Y0
K A
Y (t) = A + where
(C + eB(tM ) )1/
where M can be thought of a starting time, t0 (at which ( )
KA
1/ )
Y (t0 ) = A + (C+1) K
Q = 1 +
Y0
Including both Q and M can be convenient:
provided that > 0 and > 0.
K A The classical logistic dierential equation is a particu-
Y (t) = A + lar case of the above equation, with =1, whereas the
(C + QeB(tM ) )1/
 3

Gompertz curve can be recovered in the limit 0+ 5 References

provided that:
Richards, F. J. (1959). A Flexible Growth
( ) Function for Empirical Use. Journal of
1 Experimental Botany. 10 (2): 290300.
=O
doi:10.1093/jxb/10.2.290.
In fact, for small it is Pella, J. S.; Tomlinson, P. K. (1969). A Gener-
alised Stock-Production Model. Bull. Inter-Am.
( ( Y )) ( ) Trop. Tuna Comm. 13: 421496.
1 exp ln K Y
Y (t) = Y r rY ln Lei, Y. C.; Zhang, S. Y. (2004). Features and
K
Partial Derivatives of BertalanyRichards Growth
The RDE suits to model many growth phenomena, in- Model in Forestry. Nonlinear Analysis: Modelling
cluding the growth of tumours. Concerning its applica- and Control. 9 (1): 6573.
tions in oncology, its main biological features are similar
to those of Logistic curve model.
6 External links
2 Gradient of generalized logistic YAN Kun(2011). Research on adaptive connection
function equation in discontinuous area of data curve( Gen-
eral tendency equation of natural saturation process
When estimating parameters from data, it is often nec- curve and creep process curve), DOI:10.3969/j.issn.
essary to compute the partial derivatives of the logistic 1004-2903.2011.01.018.
function with respect to parameters at a given data point
t (see [1] ). For the case where C = 1 ,

Y
= 1 (1 + QeB(tM ) )1/
A
Y
= (1 + QeB(tM ) )1/
K
Y (K A)(t M )QeB(tM )
= 1
B (1 + QeB(tM ) ) +1
Y (K A) ln(1 + QeB(tM ) )
= 1
2 (1 + QeB(tM ) )
Y (K A)eB(tM )
= 1
Q (1 + QeB(tM ) ) +1
Y (K A)QBeB(tM )
= 1
M (1 + QeB(tM ) ) +1

3 See also
Logistic function
Gompertz curve
Ludwig von Bertalany

4 Footnotes
[1] Fekedulegn, Desta; Mairitin P. Mac Siurtain; Jim J. Col-
bert (1999). Parameter Estimation of Nonlinear Growth
Models in Forestry (PDF). Silva Fennica. 33 (4): 327
336. Retrieved 2011-05-31.
4 7 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES

7 Text and image sources, contributors, and licenses

7.1 Text
Generalised logistic function Source: https://en.wikipedia.org/wiki/Generalised_logistic_function?oldid=770138149 Contributors:
JohnOwens, Michael Hardy, Wapcaplet, Charles Matthews, David Gerard, BenFrantzDale, Bender235, HenkvD, Rjwilmsi, TeaDrinker,
BradBeattie, Mikeblas, Can't sleep, clown will eat me, Fidlej, Baccyak4H, LokiClock, Masterken, Bmcm, Vvnataraj, DFRussia, Phip-
perdee, Genie05, Bbanerje, Fgnievinski, New Image Uploader 929, Rhaertel80, Yobot, Julia W, Some standardized rigour, Debiatan,
RjwilmsiBot, Tdc204, Reverend T. R. Malthus, Khazar2, Monkbot, Mikelonergan7, Armavica and Anonymous: 18

7.2 Images
File:GeneralizedLogisticA.svg Source: https://upload.wikimedia.org/wikipedia/en/9/98/GeneralizedLogisticA.svg License: CC-BY-
SA-3.0 Contributors: ? Original artist: ?
File:GeneralizedLogisticB.svg Source: https://upload.wikimedia.org/wikipedia/en/d/d0/GeneralizedLogisticB.svg License: CC-BY-SA-
3.0 Contributors: ? Original artist: ?
File:GeneralizedLogisticC.svg Source: https://upload.wikimedia.org/wikipedia/en/8/8b/GeneralizedLogisticC.svg License: CC-BY-
SA-3.0 Contributors: ? Original artist: ?
File:GeneralizedLogisticK.svg Source: https://upload.wikimedia.org/wikipedia/en/c/cc/GeneralizedLogisticK.svg License: CC-BY-SA-
3.0 Contributors: ? Original artist: ?
File:GeneralizedLogisticNu.svg Source: https://upload.wikimedia.org/wikipedia/en/d/d1/GeneralizedLogisticNu.svg License: CC-BY-
SA-3.0 Contributors: ? Original artist: ?
File:GeneralizedLogisticQ.svg Source: https://upload.wikimedia.org/wikipedia/en/c/cf/GeneralizedLogisticQ.svg License: CC-BY-SA-
3.0 Contributors: ? Original artist: ?
File:Generalized_logistic_function_A0_K1_B1.5_Q0.5_0.5_M0.5.png Source: https://upload.wikimedia.org/wikipedia/commons/
5/54/Generalized_logistic_function_A0_K1_B1.5_Q0.5_%CE%BD0.5_M0.5.png License: Public domain Contributors: Transferred from
en.wikipedia to Commons by Ronhjones. Original artist: Debiatan at en.wikipedia

7.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Smoothstep

1 polynomial in the general smoothstep is 2N+1. With N =

smoothstep(x)
0.9 smootherstep(x) 1, the slopes or rst derivatives of the smoothstep are equal
0.8 to zero at the left and right edge (x=0 and x=1) where
0.7
the curve is appended to the constant or saturated levels.
With higher integer N, the second and higher derivatives
0.6
are zero at the edges making the polynomial functions as
0.5
at as possible making the splice to the limit values of 0
0.4
or 1 more seamless.
0.3
In MSDN and OpenGL libraries, smoothstep imple-
0.2
ments the S1 (x) , the cubic Hermite interpolation after
0.1
clamping:
0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1


0 x0
A plot of the smoothstep(x) and smootherstep(x) functions, using
0 as the left edge and 1 as the right edge. smoothstep(x) = S1 (x) = 3x2 2x3 0x1


1 1x
Smoothstep is a family of sigmoid-like interpolation and
clamping functions commonly used in computer graph- Again assuming that the left edge is 0, the right edge is 1,
ics[1][2] and video game engines.[3] with the transition between edges taking place where 0
x 1.
The function depends on three parameters, the input x, [4]
the left edge and the right edge, with the left edge A C/C++ example implementation provided by AMD
being assumed smaller than the right edge. The function follows.
receives a real number x as an argument and returns 0 if x oat smoothstep(oat edge0, oat edge1, oat x) { //
is less than or equal to the left edge, 1 if x is greater than or Scale, bias and saturate x to 0..1 range x = clamp((x -
equal to the right edge, and smoothly interpolates, using edge0)/(edge1 - edge0), 0.0, 1.0); // Evaluate polyno-
a Hermite polynomial, between 0 and 1 otherwise. The mial return x*x*(3 - 2*x); } oat clamp(oat x, oat
slope of the smoothstep function is zero at both edges. lowerlimit, oat upperlimit) { if (x < lowerlimit) x =
This is convenient for creating a sequence of transitions lowerlimit; if (x > upperlimit) x = upperlimit; return x; }
using smoothstep to interpolate each segment as an al-
ternative to using more sophisticated or expensive inter-
polation techniques.
With no salient loss of generality, the left edge may be set 1 Variations
to 0 and the right edge to 1. Then the general form for
smoothstep is Ken Perlin suggests[5] an improved version of the smooth-
step function which has zero 1st and 2nd order derivatives
at x=0 and x=1:

0 ifx 0
(N +n)(2N +1)
SN (x) = xN +1 N N n (x)
n
if0 x 1


n=0 n
1 if1 x
0 x0
smootherstep(x) = S2 (x) = 6x5 15x4 + 10x3 0x1
S0 (x) is identical to the clamping function:

1 1x
C/C++ reference implementation:

0 ifx 0
S0 (x) = x if0 x 1 oat smootherstep(oat edge0, oat edge1, oat x)

{ // Scale, and clamp x to 0..1 range x = clamp((x -
1 if1 x
edge0)/(edge1 - edge0), 0.0, 1.0); // Evaluate polynomial
The characteristic S"-shaped sigmoid curve is obtained return x*x*x*(x*(x*6 - 15) + 10); } oat clamp(oat x,
with SN (x) only for integers N 1. The order of the oat lowerlimit, oat upperlimit) { if (x < lowerlimit) x

1
2 2 ORIGIN

= lowerlimit; if (x > upperlimit) x = upperlimit; return x;

}
f (0) = 0 0 + 0 + 0 + 0 + 0 + a0 = 0
f (1) = 1 a5 + a4 + a3 + a2 + a1 + a0 = 1

Applying the desired values for the rst derivative of the

2 Origin function at both endpoints we get:

2.1 3rd order equation

f (0) = 0 0 + 0 + 0 + 0 + a1 = 0
We start with a generic third order polynomial function f (1) = 0 5a5 + 4a4 + 3a3 + 2a2 + a1 = 0
and its rst derivative:
Applying the desired values for the second derivative of
the function at both endpoints we get:
f (x) = a3 x3 + a2 x2 + a1 x + a0
f (x) = 3a3 x2 + 2a2 x + a1
f (0) = 0 0+ 0 + 0 + 2a2 = 0
Applying the desired values for the function at both end- f (1) = 0 20a5 + 12a4 + 6a3 + 2a2 = 0
points we get:
Solving the system of 6 unknowns formed by the last 6
equations we obtain the values of the polynomial coe-
f (0) = 0 0 + 0 + 0 + a0 = 0 cients:
f (1) = 1 a3 + a2 + a1 + a0 = 1

Applying the desired values for the rst derivative of the a0 = 0, a1 = 0, a2 = 0, a3 = 10, a4 = 15, a5 = 6
function at both endpoints we get:
Introducing these coecients back into the rst equation
gives the fth order smootherstep function:
f (0) = 0 0 + 0 + a1 = 0
f (1) = 0 3a3 + 2a2 + a1 = 0
f (x) = 6x5 15x4 + 10x3
Solving the system of 4 unknowns formed by the last 4
equations we obtain the values of the polynomial coe-
cients: 2.3 7th order equation

Also called smootheststep, the 7th order equation was

a0 = 0, a1 = 0, a2 = 3, a3 = 2 derived by Kyle McDonald and rst posted to Twitter[6]
with a derivation on GitHub:[7]
Introducing these coecients back into the rst equation
gives the third order smoothstep function:
f (x) = 20x7 + 70x6 84x5 + 35x4

f (x) = 2x3 + 3x2

2.4 Generalization of higher-order equa-
tions
2.2 5th order equation
Smoothstep polynomials are generalized, with 0x1 as:
We start with a generic fth order polynomial function,
its rst derivative and its second derivative:
N ( )( )
N + n 2N + 1
SN (x) = x N +1
(x)n N Z0
f (x) = a5 x5 + a4 x4 + a3 x3 + a2 x2 + a1 x + a0 n=0
n N n
( )( )
f (x) = 5a5 x4 + 4a4 x3 + 3a3 x2 + 2a2 x + a1 N
n N +n 2N + 1 N +n+1
= (1) x
f (x) = 20a5 x3 + 12a4 x2 + 6a3 x + 2a2 n=0
n N n
N ( )( )
Applying the desired values for the function at both end- N 1 2N + 1 N +n+1
= x
points we get: n=0
n N n
 3

where N determines the order of the resulting polyno- [6] kcimc (25 March 2015).
mial function, which is 2N+1. The rst seven Smoothstep smootheststep(t)=20*t^7+70*t^6-84*t^5+35*t^4
polynomials, with 0x1, are expressed as: // when smoothersteps second derivative isn't enough
(Tweet) via Twitter.

[7] Kyle McDonald (27 March 2015). Derivation of 7th-

S0 (x) = x order smoothstep function with zeros in third derivative..
S1 (x) = 2x3 + 3x2 Github.com. Retrieved 20 December 2015.

S2 (x) = 6x5 15x4 + 10x3 [8] http://stackoverflow.com/questions/41195063/

S3 (x) = 20x + 70x 84x + 35x
7 6 5 4 general-smoothstep-equation/

S4 (x) = 70x9 315x8 + 540x7 420x6 + 126x5

4 7 +External
S5 (x) = 252x11 + 1386x10 3080x9 + 3465x8 1980x 462x6 links
S6 (x) = 924x13 6006x12 + 16380x11 24024x10 + 20020x9 9009x8 + 1716x7
Using smoothstep (in the RenderMan Shading Lan-
It can be shown that the Smoothstep polynomials SN (x) guage) by Prof. Malcolm Kesson.
that transition from 0 to 1 when x transitions from 0 to 1
can be simply mapped to odd-symmetry polynomials, Interpolation tricks by Jari Komppa

Swift Interpolation Playground demonstrates

( )1 smoothStep(), smootherStep() and smoothestStep() in
1 ( )N x ( )N
RN (x) = 1 u2 du 1 u2 du a Swift playground by Simon Gladman
0 0

where

SN (x) = 1
2 RN (2x 1) + 1
2

and RN (x) = RN (x) .

The argument of RN(x) is 1x1 and is appended to
the constant 1 on the left and +1 at the right.
An implementation of SN (x) in Javascript:[8]
function generalSmoothStep(N, x) //Generalized
smoothstep { x = clamp(x, 0, 1); //x must be equal
to or between 0 and 1 var result = 0; for (var n=0;
n<=N; n++) { result += (pascalTriangle(-N-1, n) *
pascalTriangle(2*N+1, N-n) * Math.pow(x, N+n+1)); }
return result; } function pascalTriangle(a, b) { //Returns
binomial coecient without explicit use of factorials
which can't be used with negative integers var result =
1; for(var i=0; i<b; i++) { result *= (a-i)/(i+1); } return
result; } function clamp(x, lowerlimit, upperlimit) { if
(x < lowerlimit) x = lowerlimit; if (x > upperlimit) x =
upperlimit; return x; }

3 References
[1] Smoothstep at Microsoft Developer Network

[2] GLSL Language Specication, Version 1.40

[3] Unity game engine SmoothStep documentation

[4] ATI R3x0 Pixel Shaders

[5] Texturing and Modeling, Third Edition: A Procedural

Approach
4 5 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES

5 Text and image sources, contributors, and licenses

5.1 Text
Smoothstep Source: https://en.wikipedia.org/wiki/Smoothstep?oldid=781117381 Contributors: AxelBoldt, Axlrosen, Brighterorange, Bg-
white, Malcolma, MoritzMoeller, Magioladitis, Addbot, Yobot, Tcpp, Otaviogood, Erik9bot, FrescoBot, Magwo001, Cnwilliams, Daniel-
Brauer, BG19bot, Smoothstep graph, MathieuRouvinez, Cycloverid, Narky Blert, Trung0246 and Anonymous: 25

5.2 Images
File:Smoothstep_and_Smootherstep.svg Source: https://upload.wikimedia.org/wikipedia/commons/5/57/Smoothstep_and_
Smootherstep.svg License: CC0 Contributors: Own work Original artist: Smoothstep graph

5.3 Content license

Creative Commons Attribution-Share Alike 3.0
 Algebraic function

This article is about algebraic functions in calculus, An algebraic function in m variables is similarly de-
mathematical analysis, and abstract algebra. For func- ned as a function y which solves a polynomial equation
tions in elementary algebra, see function (mathematics). in m + 1 variables:

In mathematics, an algebraic function is a function that

can be dened as the root of a polynomial equation. Quite p(y, x1 , x2 , . . . , xm ) = 0.
often algebraic functions can be expressed using a nite
number of terms, involving only the algebraic operations It is normally assumed that p should be an irreducible
addition, subtraction, multiplication, division, and raising polynomial. The existence of an algebraic function is
to a fractional power. then guaranteed by the implicit function theorem.
Examples of such functions are: Formally, an algebraic function in m variables over the
eld K is an element of the algebraic closure of the eld
f (x) = 1/x of rational functions K(x1 ,...,xm).

f (x) = x

f (x) = 1+x

3 1 Algebraic functions in one vari-
x3/7 7x1/3
able
Some algebraic functions, however, cannot be expressed
by such nite expressions (this is AbelRuni theorem). 1.1 Introduction and overview
This is the case, for example, of the Bring radical, which
is the function implicitly dened by The informal denition of an algebraic function provides
a number of clues about their properties. To gain an intu-
itive understanding, it may be helpful to regard algebraic
f (x)5 + f (x) + x = 0 functions as functions which can be formed by the usual
algebraic operations: addition, multiplication, division,
In more precise terms, an algebraic function of degree n and taking an nth root. This is something of an oversim-
in one variable x is a function y = f (x) that satises a plication; because of the fundamental theorem of Ga-
polynomial equation lois theory, algebraic functions need not be expressible
by radicals.
First, note that any polynomial function y = p(x) is an
an (x)y n + an1 (x)y n1 + + a0 (x) = 0
algebraic function, since it is simply the solution y to the
where the coecients ai(x) are polynomial functions of equation
x, with coecients belonging to a set S. Quite often, S =
Q , and one then talks about function algebraic over Q
", and the evaluation at a given rational value of such an y p(x) = 0.
algebraic function gives an algebraic number.
p(x)
A function which is not algebraic is called a More generally, any rational function y = q(x) is alge-
transcendental function, as it is for example the braic, being the solution to
case of exp(x), tan(x), ln(x), (x) . A composition of
transcendental functions can give an algebraic function:
f (x) = cos(arcsin(x)) = 1 x2 . q(x)y p(x) = 0.
As an equation of degree n has n roots, a polynomial
equation does not implicitly dene a single function, but n Moreover, the nth root of any polynomial y = p(x) is
n

functions, sometimes also called branches. Consider for an algebraic function, solving the equation
example the equation of the unit circle: y 2 + x2 = 1.
This determines y, except only up to an overall sign; ac-
cordingly, it has two branches: y = 1 x2 . y n p(x) = 0.

1
2 1 ALGEBRAIC FUNCTIONS IN ONE VARIABLE

Surprisingly, the inverse function of an algebraic function

is an algebraic function. For supposing that y is a solution
to y 3 xy + 1 = 0.

Using the cubic formula, we get

an (x)y n + + a0 (x) = 0,

108 + 12 81 12x3
3
2x
for each value of x, then x is also a solution of this equa- y = + .
tion for each value of y. Indeed, interchanging the roles
3
108 + 12 81 12x3 6
of x and y and gathering terms,
For x 3
3 , the square root is real and the cubic root
4
is thus well dened, providing the unique real root. On
3
the other hand, for x > 3 , the square root is not real,
bm (y)xm + bm1 (y)xm1 + + b0 (y) = 0. 4
and one has to choose, for the square root, either non-
Writing x as a function of y gives the inverse function, real square root. Thus the cubic root has to be chosen
also an algebraic function. among three non-real numbers. If the same choices are
done in the two terms of the formula, the three choices
However, not every function has an inverse. For example, for the cubic root provide the three branches shown, in
y = x2 fails the horizontal line test: it fails to be one-to- the accompanying image.

one. The inverse is the algebraic function x = y .
Another way to understand this, is that the set of branches It may be proven that there is no way to express this func-
of the polynomial equation dening our algebraic func- tion in terms of nth roots using real numbers only, even
tion is the graph of an algebraic curve. though the resulting function is real-valued on the domain
of the graph shown.
On a more signicant theoretical level, using complex
1.2 The role of complex numbers numbers allows one to use the powerful techniques of
complex analysis to discuss algebraic functions. In par-
From an algebraic perspective, complex numbers enter ticular, the argument principle can be used to show that
quite naturally into the study of algebraic functions. First any algebraic function is in fact an analytic function, at
of all, by the fundamental theorem of algebra, the com- least in the multiple-valued sense.
plex numbers are an algebraically closed eld. Hence any
Formally, let p(x, y) be a complex polynomial in the com-
polynomial relation p(y, x) = 0 is guaranteed to have at
plex variables x and y. Suppose that x0 C is such that
least one solution (and in general a number of solutions
the polynomial p(x0 ,y) of y has n distinct zeros. We shall
not exceeding the degree of p in x) for y at each point x,
show that the algebraic function is analytic in a neighbor-
provided we allow y to assume complex as well as real
hood of x0 . Choose a system of n non-overlapping discs
values. Thus, problems to do with the domain of an alge-
i containing each of these zeros. Then by the argument
braic function can safely be minimized.
principle

I
1 py (x0 , y)
dy = 1.
2i i p(x0 , y)

By continuity, this also holds for all x in a neighborhood

of x0 . In particular, p(x,y) has only one root in i, given
by the residue theorem:

I
1 py (x, y)
fi (x) = y dy
2i i p(x, y)
3
A graph of three branches of the algebraic function y, where y which is an analytic function.
xy + 1 = 0, over the domain 3/22/3 < x < 50.

Furthermore, even if one is ultimately interested in real 1.3 Monodromy

algebraic functions, there may be no means to express
the function in terms of addition, multiplication, division
and taking nth roots without resorting to complex num-
bers (see casus irreducibilis). For example, consider the
algebraic function determined by the equation
Note that the foregoing proof of analyticity derived an
expression for a system of n dierent function elements
fi(x), provided that x is not a critical point of p(x, y).
A critical point is a point where the number of distinct
 3

zeros is smaller than the degree of p, and this occurs only Rational function
where the highest degree term of p vanishes, and where
the discriminant vanishes. Hence there are only nitely Special functions
many such points c1 , ..., cm. Transcendental function
A close analysis of the properties of the function elements
fi near the critical points can be used to show that the
monodromy cover is ramied over the critical points (and 4 References
possibly the point at innity). Thus the holomorphic ex-
tension of the fi has at worst algebraic poles and ordinary Ahlfors, Lars (1979). Complex Analysis. McGraw
algebraic branchings over the critical points. Hill.
Note that, away from the critical points, we have
van der Waerden, B.L. (1931). Modern Algebra,
Volume II. Springer.
p(x, y) = an (x)(y f1 (x))(y f2 (x)) (y fn (x))

since the fi are by denition the distinct zeros of p. The 5 External links
monodromy group acts by permuting the factors, and
thus forms the monodromy representation of the Galois Denition of Algebraic function in the Encyclo-
group of p. (The monodromy action on the universal cov- pedia of Math
ering space is related but dierent notion in the theory of
Riemann surfaces.) Weisstein, Eric W. Algebraic Function.
MathWorld.

2 History Algebraic Function at PlanetMath.org.

Denition of Algebraic function in David J. Dar-
The ideas surrounding algebraic functions go back at least ling's Internet Encyclopedia of Science
as far as Ren Descartes. The rst discussion of algebraic
functions appears to have been in Edward Waring's 1794
An Essay on the Principles of Human Knowledge in which
he writes:

let a quantity denoting the ordinate, be an alge-

braic function of the abscissa x, by the common
methods of division and extraction of roots, re-
duce it into an innite series ascending or de-
scending according to the dimensions of x, and
then nd the integral of each of the resulting
terms.

3 See also
Algebraic expression

Analytic function

Complex function

Elementary function

Function (mathematics)

Generalized function

List of special functions and eponyms

List of types of functions

Polynomial
4 6 TEXT AND IMAGE SOURCES, CONTRIBUTORS, AND LICENSES

6 Text and image sources, contributors, and licenses

6.1 Text
Algebraic function Source: https://en.wikipedia.org/wiki/Algebraic_function?oldid=774389885 Contributors: Michael Hardy, Charles
Matthews, Haukurth, Giftlite, Gene Ward Smith, Wmahan, Almit39, GrAfFiT, Photonique, Ksnow, Waldir, MarSch, Margosbot~enwiki,
YurikBot, SmackBot, InverseHypercube, Bluebot, Bethling, Silly rabbit, Tsca.bot, BenWhitey, Krashlandon, Cronholm144, Mets501,
Valoem, CBM, Dugwiki, AntiVandalBot, VictorAnyakin, Dekimasu, Johnbibby, MetsBot, Pomte, AlleborgoBot, Kai Su?, EmxBot,
SieBot, VVVBot, JackSchmidt, ClueBot, PipepBot, JP.Martin-Flatin, DumZiBoT, Mattsem, Addbot, PV=nRT, Luckas-bot, AnomieBOT,
Ciphers, Wisems, Isheden, GrouchoBot, Sawomir Biaa, EmausBot, WikitanvirBot, D.Lazard, ClueBot NG, The1337gamer, Jochen
Burghardt, JCMPC, Snowright, Elkevn, Loraof, tale.cohomology and Anonymous: 47

6.2 Images
File:Commons-logo.svg Source: https://upload.wikimedia.org/wikipedia/en/4/4a/Commons-logo.svg License: PD Contributors: ? Origi-
nal artist: ?
File:Y\char"005E\relax{}3-xy+1=0.png Source: https://upload.wikimedia.org/wikipedia/commons/0/06/Y%5E3-xy%2B1%3D0.png
License: CC-BY-SA-3.0 Contributors: http://en.wikipedia.org/wiki/File:Y%5E3-xy%2B1%3D0.png Original artist: Silly rabbit

6.3 Content license

Creative Commons Attribution-Share Alike 3.0