Professional Documents
Culture Documents
10
8
6
4
2
0
-2
-4
-6
-8
-10
-10 -8 -6 -4 -2 0
8 10
The exponential function ex (in blue), and the sum of the rst
n+1 terms of its Taylor series at 0 (in red).
4 ANALYTIC FUNCTIONS
of order zero of is dened to be itself and (x a)0 and of the paradox, but the mathematical content was appar0! are both dened to be 1. When a = 0, the series is also ently unresolved until taken up by Democritus and then
called a Maclaurin series.
Archimedes. It was through Archimedess method of exhaustion that an innite number of progressive subdivisions could be performed to achieve a nite result.[1] Liu
Hui independently employed a similar method a few cen2 Examples
turies later.[2]
The Maclaurin series for any polynomial is the polyno- In the 14th century, the earliest examples of the use of
Taylor series and closely related methods were given by
mial itself.
Madhava of Sangamagrama.[3][4] Though no record of his
The Maclaurin series for (1 x)1 is the geometric series
work survives, writings of later Indian mathematicians
suggest that he found a number of special cases of the
Taylor series, including those for the trigonometric func1 + x + x2 + x3 +
tions of sine, cosine, tangent, and arctangent. The Kerala
school of astronomy and mathematics further expanded
1
so the Taylor series for x at a = 1 is
his works with various series expansions and rational approximations until the 16th century.
In the 17th century, James Gregory also worked in this
area and published several Maclaurin series. It was not
By integrating the above Maclaurin series, we nd the until 1715 however that a general method for constructMaclaurin series for log(1 x), where log denotes the ing these series for all functions for which they exist was
natural logarithm:
nally provided by Brook Taylor,[5] after whom the series
are now named.
1 (x 1) + (x 1)2 (x 1)3 + .
1
1
1
x x2 x3 x4
2
3
4
and the corresponding Taylor series for log(x) at a = 1 is
1
1
1
(x 1) (x 1)2 + (x 1)3 (x 1)4 + ,
2
3
4
4 Analytic functions
1+
-1/x
1
0
-1
x1 x2 x3 x4 x5
x2 x3 x4 x5
xn
+ + + + + = 1+x+ + + +
+ =
.
1! 2! 3! 4! 5!
2
6 24 120
n!
n=0
History
-2
-2
-1
an (x b)n .
sin(x)
f(x)
n=0
-2
-0.5
-1
-1.5
-3.5
-2
-2.5
-3
-4
-1.5
-1
-0.5
0.5
1.5
5.1 Generalization
The Taylor approximations for log(1+x) (black). For x > 1, the
approximations diverge.
ifx = 0
ifx = 0
There is, however, a generalization[7][8] of the Taylor series that does converge to the value of the function itself
for any bounded continuous function on (0,), using the
calculus of nite dierences. Specically, one has the
following theorem, due to Einar Hille, that for any t > 0,
lim+
h0
t nh f (a)
= f (a + t).
n! hn
n=0
Here n
h is the n-th nite dierence operator with step size h.
The series is precisely the Taylor series, except that divided dierences appear in place of dierentiation: the
series is formally similar to the Newton series. When the
function f is analytic at a, the terms in the series converge to the terms of the Taylor series, and in this sense
generalizes the usual Taylor series.
In general, for any innite sequence ai, the following
power series identity holds:
un n
uj
ai = eu
ai+j .
n!
j!
n=0
j=0
So in particular,
f (a + t) = lim+ et/h
h0
j=0
f (a + jh)
(t/h)j
.
j!
line, a consequence of Borels lemma (see also Nonanalytic smooth function). As a result, the radius of The law of large numbers implies that the identity holds.
30
20
10
z
0
-10
-20
-30
-40
15
3
2
10
30
20
10
-6
1
-4
0
-2
x
-5
-10
-1
2
-2
6 -3
-10
-20
-15
-30
-40
3
2
-6
1
-4
0
-2
-1
2
-2
log(1 + x) =
6 -3
(1)n+1
n=1
1
=
xn
1 x n=0
30
20
10
20
10
z
-10
-20
-10
-20
-30
-30
-40
-40
3
2
-6
1
-4
0
-2
-1
2
30
xn
n
-2
6 -3
1
=
nxn1
(1 x)2
n=1
x
=
nxn
(1 x)2
n=0
2
=
(n 1)nxn2
(1 x)3
n=2
2x2
=
(n 1)nxn
(1 x)3
n=0
n
x
n
n=0
(1+x) =
Exponential function:
ex =
x2
x3
xn
=1+x+
+
+
n!
2!
3!
n=0
Natural logarithm:
log(1 x) =
xn
n
n=1
all for x
( )
n
( 1) ( n + 1)
k+1
=
.
=
k
n!
n
k=1
For instance, with the rst several terms written out explicitly for the common square root cases, is:
for |x| < 1
1 3
5
7
x 128
x4 + 256
x5
(1+x)0.5 = 1+ 21 x 81 x2 + 16
5 3
35 4
63 5
(1+x)0.5 = 1 12 x+ 38 x2 16
x + 128
x 256
x +
Trigonometric functions:
x3 x5
(1)n 2n+1
sin x =
x
= x +
(2n + 1)!
3! 5!
n=0
construct the Taylor series of a function, by virtue of Taylor series being power series. In some cases, one can also
derive the Taylor series by repeatedly applying integration
by parts. Particularly convenient is the use of computer
algebra systems to calculate Taylor series.
all for x
(1)n 2n
x2 x4
cos x =
x = 1 +
(2n)!
2! 4!
n=0
3
5
n
n
x 2x
(1)n E2n 2n
sec x =
x
for |x| <
one may rst rewrite the function as
(2n)!
2
n=0
arcsin x =
arccos x =
arctan x =
(2n)!
x2n+1
n (n!)2 (2n + 1)
4
n=0
for |x| 1
(2n)!
arcsin x =
x2n+1big O
fornotation)
|x| 1
n
2
2
2 n=0 4 (n!) (2n + 1)
(1)n 2n+1
x
2n + 1
n=0
for |x| 1, x = i
Hyperbolic functions:
x2n+1
x3 x5
= x+ + +
(2n + 1)!
3! 5!
n=0
cosh x =
x2
x4
x2n
=1+
+
+
(2n)!
2!
4!
n=0
tanh x =
log(1 + x) = x
x2
x3
+
+ O(x4 )
2
3
sinh x =
2
4
6
all for x cos x 1 = x + x x + O(x8 )
2
24 720
= x x3 + x5
x + for |x| <
3
15
315
2
arsinh(x) =
(1)n (2n)!
x2n+1
n (n!)2 (2n + 1)
4
n=0
artanh(x) =
x2n+1
2n + 1
n=0
all for x
for |x| 1
for |x| 1, x = 1
+ O(x )
+
+ O(x )
=
2
24 720
2
2
24
x2
x4
x6
x4
x6
x6
= +
+ O(x8 )
2
24 720
8
48 24
x2
x4
x6
=
+ O(x8 ).
2
12 45
Since the cosine is an even function, the coecients for
all the odd powers x, x3 , x5 , x7 , ... have to be zero.
7
We have for the exponential function
ex =
x2
x3
x4
xn
=1+x+
+
+
+
n!
2!
3!
4!
n=0
xn xn+1
xn xn+
+
=1+
+
n! n=0 n!
n! n=0 n!
n=0
n=1
(
)
xn xn
1
1
=1+
+
=1+
+
x
n! n=1 (n 1)!
n! (n 1)!
n=1
n=1
(1 + x)ex = ex + xex =
=1+
cos x = 1
x2
x4
+
2!
4!
n+1 n
x
n!
n=1
n+1 n
x .
n!
n=0
Classically, algebraic functions are dened by an algeThen multiplication with the denominator and substitubraic equation, and transcendental functions (including
tion of the series of the cosine yields
those discussed above) are dened by some property that
holds for them, such as a dierential equation. For example, the exponential function is the function which is equal
x
2
3
e = (c0 + c1 x + c2 x + c3 x + ) cos x
to its own derivative
everywhere, and assumes the value
(
)
(
)
x2 1 at
x4the origin. However, one may equally well dene an
2
3
4
= c0 + c1 x + c2 x + c3 x + c4 x +
1
+
2! analytic
4! function by its Taylor series.
c0
c0
c1
c1
c2 4 c2 6
c3
c3
3
= c0 x2 + x4 + c1 x x3 + x5 + c2 x2 Taylor
x series
+ x
c3 xto
+ x7 and
+ "operators"
are+
used
denex5functions
2
4!
2
4!
2
4!
2
4!
in diverse areas of mathematics. In particular, this is true
in areas where the classical denitions of functions break
Collecting the terms up to fourth order yields
down. For example, using Taylor series, one may dene
analytical
functions of matrices and operators, such as the
)
(
c1 ) 3 (
c2
cmatrix
c0 ) 2 (
0
4 exponential or matrix logarithm.
x + c3
x + c4
+
x +
= c0 +c1 x+ c2
2
2
2
4!
In other areas, such as formal analysis, it is more conveComparing coecients with the above series of the ex- nient to work directly with the power series themselves.
ponential function yields the desired Taylor series
Thus one may dene a solution of a dierential equation
as a power series which, one hopes to prove, is the Taylor
series of the desired solution.
ex
2x3
x4
2
=1+x+x +
+
+ .
cos x
3
2
7.3
Third example
(a
nd
n1
x
n
!
n
!
x
x
1
d
(1 + x)e
1
d
n1 =0 n2 =0
nd =0
as a Taylor series in x, we use the known Taylor series of
function ex :
= f (a1 , . . . , ad ) +
j=1
Thus,
x2
x3
x4
xn
=1+x+
+
+
+ .
n!
2!
3!
4!
n=0
xj
(xj aj )
1 2 f (a1 , . . . , ad )
+
(xj aj )(xk ak )
2! j=1
xj xk
d
ex =
f (a1 , . . . , ad )
k=1
d
d
d
1 3 f (a1 , . . . , ad )
+
(xj aj )(xk ak )(xl al ) + . . .
3! j=1
xj xk xl
k=1 l=1
= 0,
(x,y)=(0,0)
ex
f (a, b) + (x a) fx (a, b) + (y b) fy (a, b)
fy (a, b) =
= 1,
1 + y (x,y)=(0,0)
]
1 [
2
2
+
(x a) fxx (a, b) + 2(x a)(y b) fxy (a, b) + (y b) fyy (a, b)
2!
fxx (a, b) = ex log(1 + y)
= 0,
where the subscripts denote the respective partial deriva(x,y)=(0,0)
tives.
ex
= 1 ,
A second-order Taylor series expansion of a scalar-valued fyy (a, b) = (1 + y)2
(x,y)=(0,0)
function of more than one variable can be written com
pactly as
ex
fxy (a, b) = fyx (a, b) =
= 1.
1 + y (x,y)=(0,0)
T (x) = f (a)+(xa)T Df (a)+
(x a)
T (x) =
( f )(a) ,
!
||0
1[
0(x 0)2 + 2(x 0)(y 0) +
2
y2
+ .
2
9.1
Example
ex log(1 + y) = y + xy
y2
+
2
6
z
4
2
0
-2
-4
-6
1.5
-1
1
-0.5
0.5
0.5
x
0
1.5
2-0.5
Second-order Taylor series approximation (in orange) of a function f (x, y) = ex log (1 + y) around origin.
9
Indeed, the computation of Taylor series requires
the knowledge of the function on an arbitrary small
neighbourhood of a point, whereas the computation
of the Fourier series requires knowing the function
on its whole domain interval. In a certain sense one
could say that the Taylor series is local and the
Fourier series is global.
The Taylor series is dened for a function which has
innitely many derivatives at a single point, whereas
the Fourier series is dened for any integrable function. In particular, the function could be nowhere
dierentiable. (For example, f(x) could be a
Weierstrass function.)
The convergence of both series has very dierent
properties. Even if the Taylor series has positive
convergence radius, the resulting series may not coincide with the function; but if the function is analytic then the series converges pointwise to the function, and uniformly on every compact subset of the
convergence interval. Concerning the Fourier series, if the function is square-integrable then the series converges in quadratic mean, but additional requirements are needed to ensure the pointwise or
uniform convergence (for instance, if the function
is periodic and of class C1 then the convergence is
uniform).
Finally, in practice one wants to approximate the
function with a nite number of terms, lets say with
a Taylor polynomial or a partial sum of the trigonometric series, respectively. In the case of the Taylor
series the error is very small in a neighbourhood of
the point where it is computed, while it may be very
large at a distant point. In the case of the Fourier series the error is distributed along the domain of the
function.
11
See also
Laurent series
Madhava series
[3] Neither Newton nor Leibniz The Pre-History of Calculus and Celestial Mechanics in Medieval Kerala. MAT
314. Canisius College. Retrieved 2006-07-09.
[4] S. G. Dani (2012).
Ancient Indian Mathematics
A Conspectus.
Resonance 17 (3): 236246.
doi:10.1007/s12045-012-0022-y.
[5] Taylor, Brook, Methodus Incrementorum Directa et Inversa [Direct and Reverse Methods of Incrementation]
(London, 1715), pages 2123 (Proposition VII, Theorem
3, Corollary 2). Translated into English in D. J. Struik,
A Source Book in Mathematics 12001800 (Cambridge,
Massachusetts: Harvard University Press, 1969), pages
329332.
[6] Rudin, Walter (1980), Real and Complex Analysis, New
Dehli: McGraw-Hill, p. 418, Exercise 13, ISBN 0-07099557-5
[7] Feller, William (1971), An introduction to probability theory and its applications, Volume 2 (3rd ed.), Wiley, pp.
230232.
[8] Hille, Einar; Phillips, Ralph S. (1957), Functional analysis and semi-groups, AMS Colloquium Publications 31,
American Mathematical Society, p. 300327.
[9] Most of these can be found in (Abramowitz & Stegun
1970).
13 References
Abramowitz, Milton; Stegun, Irene A. (1970),
Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables, New York:
Dover Publications, Ninth printing
Thomas, George B. Jr.; Finney, Ross L. (1996),
Calculus and Analytic Geometry (9th ed.), Addison
Wesley, ISBN 0-201-53174-7
Greenberg, Michael (1998), Advanced Engineering
Mathematics (2nd ed.), Prentice Hall, ISBN 0-13321431-1
14 External links
12
Notes
10
Another Taylor visualisation where you can
choose the point of the approximation and the number of derivatives
Taylor series revisited for numerical methods at
Numerical Methods for the STEM Undergraduate
Cinderella 2: Taylor expansion
Taylor series
Inverse trigonometric functions Taylor series
14
EXTERNAL LINKS
11
15
15.1
Taylor series Source: http://en.wikipedia.org/wiki/Taylor%20series?oldid=660459897 Contributors: AxelBoldt, Mav, Tarquin, Jeronimo, XJaM, Miguel~enwiki, Roadrunner, Stevertigo, Patrick, Michael Hardy, Dominus, Zeno Gantner, GTBacchus, Loisel, Eric119,
Minesweeper, Ejrh, Cyp, Djmutex, Ideyal, Charles Matthews, RickK, Dysprosia, Jitse Niesen, Rudminjd, Doradus, Populus, Fibonacci,
McKay, Fvw, Cmcb, Robbot, Fredrik, Gandalf61, Sverdrup, Puckly, Hadal, Pps, Tobias Bergemann, Giftlite, BenFrantzDale, Lethe,
Frencheigh, Alberto da Calvairate~enwiki, LucasVB, Antandrus, Melikamp, Goh wz, Tsemii, Mh, Mjec, JimJast, Wclark, Shinglor,
Mecanismo, Sam Derbyshire, Paul August, Bender235, Reneld, Flxmghvgvk, Alamino, Jane Fallen, Haham hanuka, Msh210, LucaB~enwiki, Sligocki, Schapel, RJFJR, WojciechSwiderski~enwiki, Duplode, Oleg Alexandrov, Mwilde, Linas, Bo198214~enwiki, Pinkisnt-well, Guardian of Light, Stealth HR, Graham87, BD2412, Pranathi, Jclemens, Rjwilmsi, OneWeirdDude, FlaBot, VKokielov, Mathbot,
RexNL, Fresheneesz, Glenn L, Jatinshah, Chobot, Flcelloguy, Banaticus, Wavelength, Deeptrivia, Hede2000, JabberWok, CambridgeBayWeather, NawlinWiki, Qualc1, Wiki alf, Diotti, CecilWard, Schmock, Lumaga, Genjix, Zzuuzz, Gulliveig, Petri Krohn, Mordacil, Fram,
Naught101, Gesslein, GrinBot~enwiki, Bo Jacoby, Dudzcom, SmackBot, Staplesauce, InverseHypercube, KocjoBot~enwiki, Jagged 85,
Alksub, Jwestbrook, Chris the speller, Audacity, Cjs young, Silly rabbit, RayAYang, Gracenotes, Shalom Yechiel, PoiZaN, Jaro.p~enwiki,
Perlmonger42, Vanished User 0001, Goodale, Jmnbatista, Daqu, Nakon, Druseltal2005, PouyaDT, Ugur Basak Bot~enwiki, Ohconfucius,
Lambiam, Harryboyles, Richard L. Peterson, Dlimpid, Stikonas, Dalstadt, Elb2000, Datoews, Xionbox, Gijs.peek, Tikai, Bill Malloy,
Tawkerbot2, JRSpriggs, CBM, Runningonbrains, Ali Obeid, Myasuda, Rudminjw@jmu.edu, Wowus, Doctormatt, Benzi455, Cario24,
Gogo Dodo, Xantharius, Talgalili, Thijs!bot, Epbr123, LaGrange, Akitchin, Headbomb, Nadav1, AbcXyz, BigJohnHenry, LachlanA, El
Jogg, Orionus, JAnDbot, RETROFUTURE, Olaf, Igodard, BlueSoxSWJ, AdamGomaa, Bgagan911, Jakob.scholbach, Baccyak4H, Reminiscenza, David Eppstein, Error792, Pruthvi.Vallabh, Pomte, Slash, J.delanoy, Jesper Carlstrom, Krishnachandranvn, Gombang, Policron, Clarknj, Rinn0, WinterSpw, VolkovBot, Thomas.W, Jimothy 46, Je G., Fundamental metric tensor, LokiClock, TXiKiBoT, Sloq,
Anonymous Dissident, Drzib, RRRR0000RRRR, Xem 007, Janzz2k, RageGarden, Cwkmail, RJaguar3, Je223~enwiki, James T Curran,
Reinderien, Randomblue, ClueBot, Akhtarphysic, Plastikspork, Jasanwiki, Gtstricky, Basketball110,
, Dspdude, Johnuniq, Egmontaz, Darkicebot, InternetMeme, XLinkBot, Forbes72, SilvonenBot, Addbot, Nilesj, Avarice2593, Hjpotter92, BepBot, Glane23, Barak Sh,
AgadaUrbanit, Numbo3-bot, Robertsrap111, Tide rolls, Lightbot, Zorrobot, Luckas-bot, Yobot, JackPotte, Randomath~enwiki, KamikazeBot, Tskuzzy, AnomieBOT, Autarkaw, Rubinbot, 1exec1, Jim1138, Lupamind, Davidpar, E2eamon, Qorilla, Xqbot, Bdmy, Txebixev,
Musicguyguy, Shanman7, TheQuestionGuy, Twisterplus, Acky69, Zhefurui, Qgluca, Sawomir Biay, Mfwitten, Elocute, Cannolis, Citation bot 1, DrilBot, Stephen.metzger, Jschnur, RedBot, Jixzad123, Thesevenseas, Tcnuk, Plasticspork, Kallikanzarid, FoxBot, Sajoka,
Estherholleman, Red Denim, Vrenator, FootballHK, Troubled asset, GA bot, Laurifer, Netheril96, Slawekb, Holger Flier, Kepke, Preetum, Quondum, CornellRunner314, ResearchRave, Support.and.Defend, ClueBot NG, Fx0701, Chetrasho, Arthurcburigo, EnglishTea4me,
Vincius Machado Vogt, Joel B. Lewis, Hulaxhula15, Helpful Pixie Bot, Jaylowblow, Byakuya1995, Abcwikip, Koertefa, BG19bot, Walrus068, HGK745, MusikAnimal, Mhallwiki, Brad7777, Pmonaragala, Klilidiplomus, Alansfault, IkamusumeFan, NitRav, Dexbot, Frosty,
A.entropy, Faizan, Pagerank980, CsDix, SucreRouge, Unlikelyuser, Neuroxic, Mathmensch, Blackbombchu, Ugog Nizdast, The Herald,
Yumichael, K.richardson.math, Biblioworm, Jorge Guerra Pires, Xeus666, Rangdor, M7660115, Iamswag45, EmJayGee and Anonymous:
471
15.2
Images
12
15
15.3
Content license