Jacobian matrix and determinant

In vector calculus, the Jacobian matrix (/dkobin/, 1 Jacobian matrix

/jkobin/) is the matrix of all rst-order partial deriva-
tives of a vector-valued function. When the matrix is a The Jacobian generalizes the gradient of a scalar-valued
square matrix, both the matrix and its determinant are function of multiple variables, which itself generalizes the
referred to as the Jacobian in literature.[1] derivative of a scalar-valued function of a single vari-
Suppose f : n m is a function which takes as input able. In other words, the Jacobian for a scalar-valued
the vector x n and produces as output the vector f(x) multivariate function is the gradient and that of a scalar-
m . Then the Jacobian matrix J of f is an mn matrix, valued function of single variable is simply its derivative.
usually dened and arranged as follows: The Jacobian can also be thought of as describing the
amount of stretching, rotating or transforming that
a transformation imposes locally. For example, if (x, y)
f1 f1 = f(x, y) is used to transform an image, the Jacobian J (x,

[ ] x1 xn y), describes how the image in the neighborhood of (x, y)
f f ..
..
J= = .
..
. . is transformed.
x1 xn f
m fm If a function is dierentiable at a point, its derivative

x1 xn is given in coordinates by the Jacobian, but a function
or, component-wise: doesn't need to be dierentiable for the Jacobian to be
dened, since only the partial derivatives are required to
exist.
fi
Jij = . If p is a point in n and f is dierentiable at p, then its
xj
derivative is given by J (p). In this case, the linear map
This matrix, whose entries are functions of x, is also de- described by J (p) is the best linear approximation of f
noted by Df, J , and (f 1 ,...,fm)/(x1 ,...,xn). (Note that near the point p, in the sense that
some literature denes the Jacobian as the transpose of
the matrix given above.)
The Jacobian matrix is important because if the function f(x) = f(p) + Jf (p) (x p) + o(x p)
f is dierentiable at a point x (this is a slightly stronger for x close to p and where o is the little o-notation (for x
condition than merely requiring that all partial derivatives p) and x p is the distance between x and p.
exist there), then the Jacobian matrix denes a linear map
n m , which is the best (pointwise) linear approxi- Compare this to a Taylor series for a scalar function of a
mation of the function f near the point x. This linear map scalar argument, truncated to rst order:
is thus the generalization of the usual notion of derivative,
and is called the derivative or the dierential of f at x.
f (x) = f (p) + f (p)(x p) + o(x p).
If m = n, the Jacobian matrix is a square matrix, and its
determinant, a function of x1 , , xn, is the Jacobian In a sense, both the gradient and Jacobian are "rst deriva-
determinant of f. It carries important information about tives"the former the rst derivative of a scalar function
the local behavior of f. In particular, the function f has lo- of several variables, the latter the rst derivative of a vec-
cally in the neighborhood of a point x an inverse function tor function of several variables.
that is dierentiable if and only if the Jacobian determi- The Jacobian of the gradient of a scalar function of sev-
nant is nonzero at x (see Jacobian conjecture). The Ja- eral variables has a special name: the Hessian matrix,
cobian determinant also appears when changing the vari- which in a sense is the "second derivative" of the func-
ables in multiple integrals (see substitution rule for mul- tion in question.
tiple variables).
If m = 1, f is a scalar eld and the Jacobian matrix is
reduced to a row vector of partial derivatives of fi.e. 2 Jacobian determinant
the gradient of f.
These concepts are named after the mathematician Carl If m=n, then f is a function from n to itself and the Ja-
Gustav Jacob Jacobi (18041851). cobian matrix is a square matrix. We can then form its

1
2 5 EXAMPLES

The (unproved) Jacobian conjecture is related to global

invertibility in the case of a polynomial function, that is
f a function dened by n polynomials in n variables. It as-
serts that, if the Jacobian determinant is a non-zero con-
stant (or, equivalently, that it does not have any complex
zero), then the function is invertible and its inverse is a
polynomial function.

A nonlinear map f : R2 R2 sends a small square to a distorted

parallelogram close to the image of the square under the best 4 Critical points
linear approximation of f near the point.
Main article: Critical point
determinant, known as the Jacobian determinant. The
Jacobian determinant is occasionally referred to as the If f : n m is a dierentiable function, a critical point
Jacobian. of f is a point where the rank of the Jacobian matrix is not
maximal. This means that the rank at the critical point is
The Jacobian determinant at a given point gives impor-
lower than the rank at some neighbour point. In other
tant information about the behavior of f near that point.
words, let k be the maximal dimension of the open balls
For instance, the continuously dierentiable function f is
n contained in the image of f; then a point is critical if all
invertible near a point p if the Jacobian determinant
minors of rank k of f are zero.
at p is non-zero. This is the inverse function theorem.
Furthermore, if the Jacobian determinant at p is positive, In the case where 1 = m = n = k, a point is critical if the
then f preserves orientation near p; if it is negative, f re- Jacobian determinant is zero.
verses orientation. The absolute value of the Jacobian
determinant at p gives us the factor by which the function
f expands or shrinks volumes near p; this is why it occurs 5 Examples
in the general substitution rule.
The Jacobian determinant is used when making a change 5.1 Example 1
of variables when evaluating a multiple integral of a func-
tion over a region within its domain. To accommodate for Consider the function f : 2 2 given by
the change of coordinates the magnitude of the Jacobian
determinant arises as a multiplicative factor within the in-
tegral. This is because the n-dimensional dV element is [ ]
x2 y
in general a parallelepiped in the new coordinate system, f(x, y) = .
5x + sin y
and the n-volume of a parallelepiped is the determinant
of its edge vectors. Then we have
The Jacobian can also be used to solve systems of dier-
ential equations at an equilibrium point or approximate
solutions near an equilibrium point. f1 (x, y) = x2 y

and
3 Inverse
According to the inverse function theorem, the matrix in- f2 (x, y) = 5x + sin y
verse of the Jacobian matrix of an invertible function is
and the Jacobian matrix of F is
the Jacobian matrix of the inverse function. That is, if the
Jacobian of the function f : n n is continuous and
nonsingular at the point p in n , then f is invertible when
f1 f1
restricted to some neighborhood of p and [ ]
x y x2
Jf (x, y) = = 2xy
f2 f2 5 cos y
1 x y
J 1 f = Jf .
f

Conversely, if the Jacobian determinant is not zero at a and the Jacobian determinant is
point, then the function is locally invertible near this point,
that is, there is a neighbourhood of this point in which the
function is invertible. det(Jf (x, y)) = 2xy cos y 5x2 .
5.5 Example 5 3

5.2 Example 2: polar-Cartesian transfor-

mation
y1 = x1
The transformation from polar coordinates (r, ) to y2 = 5x3
Cartesian coordinates (x, y), is given by the function F: y3 = 4x22 2x3
+ [0, 2) 2 with components:
y4 = x3 sin x1

is
x = r cos ;
y = r sin .

y1 y1 y1
x x
[ ] x1 x2 x3
r
J(r, ) = = cos r sin y2 y2 y2

y y sin r cos 1 0 0
x x3
1 x2 0 0 5
r JF (x1 , x2 , x3 ) = = .
y3 y3 y3 0 8x2 2
The Jacobian determinant is equal to r. This can be used
x1 x2
x3 x 3 cos x1 0 sin x1
to transform integrals between the two coordinate sys-
y4 y4 y4
tems:
x1 x2 x3

This example shows that the Jacobian need not be a

f (x, y) dx dy = f (r cos , r sin ) r dr d. square matrix.
F(A) A

5.3 Example 3: spherical-Cartesian trans- 5.5 Example 5

formation
The Jacobian determinant of the function F : 3 3
The transformation from spherical coordinates (r, , ) with components
to Cartesian coordinates (x, y, z), is given by the function
F: + [0, ] [0, 2 ) 3 with components:
y1 = 5x2
y2 = 4x21 2 sin(x2 x3 )
x = r sin cos ;
y3 = x2 x3
y = r sin sin ;
z = r cos . is

The Jacobian matrix for this coordinate change is

 
 0 5 0   
  5 0 
8x1
2x cos(x x ) 2x cos(x x ) = 8x  = 40x1 x
 3 2 3 2 2 3  1
x3 x2 
x x x  0 x3 x2 
r

y y y sin cos r cos cos r sin sin
From this we see that F reverses orientation near those
JF (r, , ) = = sin sin r cos sin r sin cos .
r points where x1 and x2 have the same sign; the function
cos r sin 0
z z z is locally invertible everywhere except near points where
r x1 = 0 or x2 = 0. Intuitively, if one starts with a tiny object
around the point (1, 2, 3) and apply F to that object, one
The determinant is r sin . As an example, since dV = dx will get a resulting object with approximately 40 1 2 =
2

dy dz this determinant implies that the dierential volume 80 times the volume of the original one, with orientation
element dV = r2 sin dr d d. Unlike for a change of reversed.
Cartesian coordinates, this determinant is not a constant,
and varies with coordinates (r and ).
6 Other uses
5.4 Example 4
The Jacobian serves as a linearized design matrix in sta-
The Jacobian matrix of the function F : 3 4 with tistical regression and curve tting; see non-linear least
components squares.
4 10 EXTERNAL LINKS

6.1 Dynamical systems

Consider a dynamical system of the form x = F(x), where
x is the (component-wise) time derivative of x, and F :
n n is dierentiable. If F(x0 ) = 0, then x0 is a
stationary point (also called a critical point; this is not to
be confused with xed points). The behavior of the sys-
tem near a stationary point is related to the eigenvalues
of JF(x0 ), the Jacobian of F at the stationary point.[2]
Specically, if the eigenvalues all have real parts that
are negative, then the system is stable near the station-
ary point, if any eigenvalue has a real part that is positive,
then the point is unstable. If the largest real part of the
eigenvalues is zero, the Jacobian matrix does not allow
for an evaluation of the stability.

6.2 Newtons method

A system of coupled nonlinear equations can be solved
iteratively by Newtons method. This method uses the
Jacobian matrix of the system of equations.

7 See also
Hessian matrix

Pushforward (dierential)

8 References
[1] Mathworld

[2] Arrowsmith, D. K.; Place, C. M. (1992). Section 3.3.

Dynamical Systems. London: Chapman & Hall. ISBN
0-412-39080-9.

9 Further reading
Gandolfo, Giancarlo (1996). Economic Dynamics
(Third ed.). Berlin: Springer. pp. 305330. ISBN
3-540-60988-1.

10 External links
Hazewinkel, Michiel, ed. (2001), Jacobian,
Encyclopedia of Mathematics, Springer, ISBN 978-
1-55608-010-4

Mathworld A more technical explanation of Jaco-

bians
 5

11 Text and image sources, contributors, and licenses

11.1 Text
Jacobian matrix and determinant Source: https://en.wikipedia.org/wiki/Jacobian_matrix_and_determinant?oldid=780616932 Contrib-
utors: AxelBoldt, Jqt, Michael Hardy, Delirium, Looxix~enwiki, Kevin Baas, AugPi, Charles Matthews, Dysprosia, Jitse Niesen, Zero0000,
Alembert~enwiki, Robbot, Josh Cherry, YahoKa, Fredrik, Rasmus Faber, Robinh, Isopropyl, Giftlite, BenFrantzDale, Everyking, Edrex,
Vadmium, Leonard Vertighel, J. 'mach' wust, Karol Langner, Discospinster, ObsessiveMathsFreak, Bender235, Kwamikagami, Wood
Thrush, Tsirel, Varuna, Jumbuck, Keenan Pepper, Voltagedrop, Oleg Alexandrov, Thryduulf, LizardWizard, Kzollman, Mpatel, Tabletop,
Bluemoose, Rjwilmsi, HappyCamper, R.e.b., Bushido Hacks, FlaBot, Mathbot, Sanpaz, WriterHound, YurikBot, Wavelength, Dotanco-
hen, Archelon, Pmdboi, Syth, Tong~enwiki, Snaxe920, Teply, Zvika, SmackBot, Eskimbot, BiT, PeterSymonds, Njerseyguy, Silly rabbit,
Complexica, Nbarth, DHN-bot~enwiki, John Reaves, Bsilverthorn, Spakoj~enwiki, Kleuske, Rpf, Lambiam, Maverick starstrider, Jim.belk,
Asyndeton, JoeBot, JKrabbe, Paul Matthews, Cydebot, Justincop, Mathmoclaire, SGGH, Batmanshotokan, Ben pcc, Joe Schmedley, JAnD-
bot, Deective, MSBOT, VoABot II, Stevvers, User A1, J.delanoy, Melink14, LordAnubisBOT, Policron, Fylwind, LokiClock, Rei-bot,
Cloudswrest, Dch888, Geometry guy, Graymornings, Jhawkinson, Scottkosty, Ivan tambuk, YonaBot, Cwkmail, Yintan, Smarchesini,
OKBot, Blacklemon67, Randomblue, Denisarona, UKoch, Rubybrian, Qwfp, Tomhosking, MystBot, ClculIntegral, Addbot, Fgnievin-
ski, Topology Expert, EconoPhysicist, TStein, Ozob, Wikomidia, Leycec, Luckas-bot, Garymm, , AnomieBOT, Hebr138, NickK,
Kike123unc, ArthurBot, Bdmy, Etoombs, DBAllan, GrouchoBot, BradLipovsky, Point-set topologist, Kristjan.Jonasson, FrescoBot,
Agepap, TobeBot, Wcedric, Jangekheid, RjwilmsiBot, 123Mike456Winston789, Sweethaws, EmausBot, Gpooseh, D.Lazard, Mirko.raca,
U+003F, IznoRepeat, Heinzhill, ClueBot NG, PGulf, Wcherowi, Z423x5c6, Neojamaj, Dayson39, Anelson0528, ChrisGualtieri, Makecat-
bot, Mihaif7, CsDix, Zoydb, Mi scot87, AioftheStorm, Ingmar Schuster, ArchimedesZ, Phleg1, Nigelinux, JBrobst, Hayazin, Arturodon-
juan, Adam9007, Gholleywiki, Brownzb, Fmadd, Tony.Lee, Balping and Anonymous: 178

11.2 Images
File:Jacobian_determinant_and_distortion.svg Source: https://upload.wikimedia.org/wikipedia/en/9/96/Jacobian_determinant_and_
distortion.svg License: CC-BY-SA-3.0 Contributors:
created in inkscape
Original artist:
Blacklemon67

11.3 Content license

Creative Commons Attribution-Share Alike 3.0