View source for Curvature

{{about|mathematics and related concepts in geometry}}
{{short description|Mathematical measure of how much a curve or surface deviates from flatness}}
{{cs1 config |mode=cs2 }}
{{use dmy dates |cs1-dates=sy |date=November 2024}}

[[File:Cell-Shape-Dynamics-From-Waves-to-Migration-pcbi.1002392.s007.ogv|thumb|A migrating wild-type ''[[Dictyostelium discoideum]]'' cell whose boundary is colored by curvature. Scale bar: 5&nbsp;μm.]]

In [[mathematics]], '''curvature''' is any of several strongly related concepts in [[geometry]] that intuitively measure the amount by which a [[curve]] deviates from being a [[straight line]] or by which a [[Surface (mathematics)|surface]] deviates from being a [[plane (geometry)|plane]]. If a curve or surface is contained in a larger space, curvature can be defined ''extrinsically'' relative to the ambient space. [[Curvature of Riemannian manifolds]] of dimension at least two can be defined ''intrinsically'' without reference to a larger space.

For curves, the canonical example is that of a [[circle]], which has a curvature equal to the [[Multiplicative inverse|reciprocal]] of its [[radius]]. Smaller circles bend more sharply, and hence have higher curvature. The curvature ''at a point'' of a [[differentiable curve]] is the curvature of its [[osculating circle]] — that is, the circle that best approximates the curve near this point. The curvature of a straight line is zero. In contrast to the [[tangent]], which is a vector quantity, the curvature at a point is typically a [[scalar (mathematics)|scalar]] quantity, that is, it is expressed by a single [[real number]].

For surfaces (and, more generally for higher-dimensional [[manifold]]s), that are [[embedding|embedded]] in a [[Euclidean space]], the concept of curvature is more complex, as it depends on the choice of a direction on the surface or manifold. This leads to the concepts of ''maximal curvature'', ''[[Minimal surface|minimal curvature]]'', and ''[[mean curvature]]''.

==History==
The history of curvature began with the [[Ancient Greece|ancient Greeks]]' basic distinction between straight and circular lines, with the concept later developed by figures like [[Aristotle]] and [[Apollonius]]. The development of calculus in the 17th century, particularly by Newton and Leibniz, provided tools to systematically calculate curvature for curves. Euler then extended the study to surfaces, followed by Gauss's crucial insight of "intrinsic" curvature, which is independent of how a surface is embedded in space, and Riemann's generalization to higher dimensions.<ref>{{Cite web |title=The History of Curvature |url=https://homepage.villanova.edu/robert.jantzen/maple/misc/history_of_curvature/k.htm#:~:text=For%20example,%20a%20curve%20that,and%20mixed%20(Coolidge%20375). |access-date=2025-10-14 |website=homepage.villanova.edu}}</ref>

In ''Tractatus de configurationibus qualitatum et motuum,''<ref>{{citation |last=Clagett |first=Marshall |author-link=Marshall Clagett |title=Nicole Oresme and the Medieval Geometry of Qualities and Motions |year=1968 |publisher=University of Wisconsin Press |isbn=978-0-299-04880-8 |place=Madison, WI }}</ref> the 14th-century philosopher and mathematician [[Nicole Oresme]] introduces the concept of curvature as a measure of departure from straightness; for circles he has the curvature as being inversely proportional to the radius; and he attempts to extend this idea to other curves as a continuously varying magnitude.<ref>{{citation |last1=Serrano |first1=Isabel M. |last2=Suceavă |first2=Bogdan D. |author-link2=Bogdan Suceavă |year=2015 |title=A Medieval Mystery: Nicole Oresme's Concept of Curvitas |journal=Notices of the AMS |volume=62 |issue=9 |pages=1030–1034 |doi=10.1090/noti1275 |doi-access=free}}</ref>

The curvature of a [[differentiable curve]] was originally defined through [[osculating circle]]s. In this setting, [[Augustin-Louis Cauchy]] showed that the center of curvature is the intersection point of two infinitely close [[normal (geometry)|normal lines]] to the curve.<ref name="Cauchy">{{citation |last1=Borovik |first1=Alexandre |author-link1=Alexandre Borovik |last2=Katz |first2=Mikhail G. |author-link2=Mikhail Katz |year=2011 |title=Who Gave You the Cauchy–Weierstrass Tale? The Dual History of Rigorous Calculus |journal=Foundations of Science |volume=17 |issue=3 |pages=245–276 |arxiv=1108.2885 |doi=10.1007/s10699-011-9235-x }}</ref>

==Plane curves==
Intuitively, the curvature describes for any part of a curve how much the curve direction changes over a small distance travelled (e.g. angle per length in [[Radians per metre|{{math|rad/m}}]]), so it is a measure of the [[instantaneous rate of change]] of ''direction'' of a point that moves on the curve: the larger the curvature, the larger this rate of change. In other words, the curvature measures how fast the unit tangent vector to the curve at point {{mvar|P}} rotates<ref>{{citation |last=Pressley |first=Andrew |url=https://books.google.com/books?id=UXPyquQaO6EC |title=Elementary Differential Geometry |date=2001 |publisher=Springer |isbn=978-1-85233-152-8 |place=London |page=29 }}</ref> when point {{mvar|P}} moves at unit speed along the curve. In fact, it can be proved that this instantaneous rate of change of direction is exactly the curvature. 

More precisely, suppose that the point is moving on the curve at a constant speed of one unit per time, that is, the position of the point {{math|''P''(''s'')}} is a unit-speed function of the parameter {{mvar|s}}. (The parameter may be thought as time or as [[arc length]] from a given origin.) Let {{math|'''T'''(''s'')}} be a [[unit tangent vector]] of the curve at {{math|''P''(''s'')}}, which is also the [[derivative]] of {{math|''P''(''s'')}} with respect to {{mvar|s}}. Then, the derivative of {{math|'''T'''(''s'')}} with respect to {{mvar|s}} is a vector that is normal to the curve and whose length is the curvature.

To be meaningful, the definition of the curvature and its different characterizations require that the curve is [[continuously differentiable]] near {{mvar|P}}, for having a tangent that varies continuously; it requires also that the curve is twice differentiable at {{mvar|P}}, for insuring the existence of the involved limits, and of the derivative of {{math|'''T'''(''s'')}}.

The characterization of the curvature in terms of the derivative of the unit tangent vector is probably less intuitive than the definition in terms of the osculating circle, but formulas for computing the curvature are easier to deduce. Therefore, and also because of its use in [[kinematics]], this characterization is often given as a definition of the curvature.

===Osculating circle===
[[File:Osculating.svg|alt=|right|250x250px]]
Historically, the curvature of a differentiable curve was defined through the [[osculating circle]], which is the circle that best approximates the curve at a point. More precisely, given a point {{mvar|P}} on a curve, every other point {{mvar|Q}} of the curve defines a circle (or sometimes a line) passing through {{mvar|Q}} and [[tangent (geometry)|tangent]] to the curve at {{mvar|P}}. The osculating circle is the [[limit (mathematics)|limit]], if it exists, of this circle when {{mvar|Q}} tends to {{mvar|P}}. Then the ''center'' and the ''radius of curvature'' of the curve at {{mvar|P}} are the center and the radius of the osculating circle. The curvature is the [[multiplicative inverse|reciprocal]] of radius of curvature. That is, the curvature is 
: <math> \kappa = \frac{1}{R},</math>
where {{mvar|R}} is the radius of curvature<ref>{{harvnb|Kline|1998|page=458}}</ref> (the whole circle has this curvature, it can be read as turn {{math|2π}} over the length {{math|2π}}{{mvar|R}}).

This definition is difficult to manipulate and to express in formulas. Therefore, other equivalent definitions have been introduced.

===In terms of arc-length parametrization===
Every [[differentiable curve]] can be [[parametric representation|parametrized]] with respect to [[arc-length parametrization|arc length]].<ref>{{citation |type=Website
 | url = https://sites.google.com/site/johnkennedyshome/home/class-downloads
 | last1 = Kennedy
 | first1 = John
 | title = The Arc Length Parametrization of a Curve
 | year = 2011
 | access-date = 2013-12-10
 | archive-date = 2015-09-28
 | archive-url = https://web.archive.org/web/20150928030020/https://sites.google.com/site/johnkennedyshome/home/class-downloads
 | url-status = dead
 }}</ref> In the case of a plane curve, this means the existence of a parametrization {{math|'''γ'''(''s'') {{=}} (''x''(''s''), ''y''(''s''))}}, where {{mvar|x}} and {{mvar|y}} are real-valued differentiable functions whose derivatives satisfy
:<math>\|\boldsymbol{\gamma}'\| = \sqrt{x'(s)^2+y'(s)^2} = 1.</math>
This means that the tangent vector
:<math>\mathbf T(s)=\bigl(x'(s),y'(s)\bigr)</math>
has a length equal to one and is thus a [[unit tangent vector]].

If the curve is twice differentiable, that is, if the second derivatives of {{mvar|x}} and {{mvar|y}} exist, then the derivative of {{math|'''T'''(''s'')}} exists. This vector is normal to the curve, its length is the curvature {{math|''κ''(''s'')}}, and it is oriented toward the center of curvature. That is,
:<math>\begin{align}
\mathbf{T}(s) &= \boldsymbol{\gamma}'(s), \\[8mu]
\|\mathbf{T}(s)\|^2 &= 1 \ \text{(constant)} \implies \mathbf{T}'(s)\cdot \mathbf{T}(s) = 0, \\[5mu]
\kappa(s) &= \|\mathbf{T}'(s)\| = \|\boldsymbol{\gamma}''(s)\| = \sqrt{x''(s)^2+y''(s)^2}
\end{align}</math>

Moreover, because the radius of curvature is (assuming ''𝜿''(''s'') ≠ 0)
:<math>R(s)=\frac{1}{\kappa(s)},</math>
and the center of curvature is on the normal to the curve, the center of curvature is the point 
:<math> \mathbf{C}(s)= \boldsymbol{\gamma}(s) + \frac 1{\kappa(s)^2}\mathbf{T}'(s).</math> 
(In case the curvature is zero, the center of curvature is not located anywhere on the plane '''''R'''''<sup>2</sup> and is often said to be located "at infinity".)

{{anchor|signed curvature}}
If {{math|'''N'''(''s'')}} is the [[unit normal vector]] obtained from {{math|'''T'''(''s'')}} by a counterclockwise rotation of {{sfrac|{{pi}}|2}}, then 
:<math>\mathbf{T}'(s)=k(s)\mathbf{N}(s),</math>
with {{math|1= ''k''(''s'') = ± ''κ''(''s'')}}. The real number {{math|''k''(''s'')}} is called the '''oriented curvature''' or '''signed curvature'''. It depends on both the orientation of the plane (definition of counterclockwise), and the orientation of the curve provided by the parametrization. In fact, the [[change of variable]] {{math|''s'' → –''s''}}   provides another arc-length parametrization, and changes the sign of {{math|''k''(''s'')}}.

With the above, the center of curvature can be expressed as: 
:<math>\mathbf{C}(s)= \boldsymbol{\gamma}(s) + 
R(s)\mathbf{N}(s).</math>

=== In terms of a general parametrization ===
Let {{math|'''γ'''(''t'') {{=}} (''x''(''t''), ''y''(''t''))}} be a proper [[parametric representation]] of a twice differentiable plane curve. Here ''proper'' means that on the [[domain of a function|domain]] of definition of the parametrization, the derivative {{math|{{sfrac|''d'''''γ'''|''dt''}}}}
is defined, differentiable and nowhere equal to the zero vector.

With such a parametrization, the signed curvature is 
:<math>k = \frac{x'y''-y'x''}{\bigl({x'}^2+{y'}^2\bigr)\vphantom{'}^{3/2}},</math>
where primes refer to derivatives with respect to {{mvar|t}}. The curvature {{mvar|''κ''}} is thus
:<math>\kappa = \frac{\left|x'y''-y'x''\right|}{\bigl({x'}^2+{y'}^2\bigr)\vphantom{'}^{3/2}}.</math>

These can be expressed in a [[coordinate-free]] way as
:<math>
k = \frac{\det\left(\boldsymbol{\gamma}',\boldsymbol{\gamma}''\right)}{\|\boldsymbol{\gamma}'\|^3},\qquad
\kappa = \frac{\left|\det\left(\boldsymbol{\gamma}',\boldsymbol{\gamma}''\right)\right|}{\|\boldsymbol{\gamma}'\|^3}.
</math>

These formulas can be derived from the special case of arc-length parametrization in the following way. The above condition on the parametrisation imply that the arc length {{mvar|s}} is a differentiable [[monotonic function]] of the parameter {{mvar|t}}, and conversely that {{mvar|t}} is a monotonic function of {{mvar|s}}. Moreover, by changing, if needed, {{mvar|s}} to {{math|–''s''}}, one may suppose that these functions are increasing and have a positive derivative. Using notation of the preceding section and the [[chain rule]], one has 
:<math>\frac{d\boldsymbol{\gamma}}{dt}= \frac{ds}{dt}\mathbf T,</math>
and thus, by taking the norm of both sides 
:<math> \frac{dt}{ds}= \frac 1{\|\boldsymbol{\gamma}'\|},</math>
where the prime denotes differentiation with respect to {{mvar|t}}.

The curvature is the norm of the derivative of {{math|'''T'''}} with respect to {{mvar|s}}. By using the above formula and the chain rule this derivative and its norm can be expressed in terms of {{math|'''γ'''′}} and {{math|'''γ'''″}} only, with the arc-length parameter {{mvar|s}} completely eliminated, giving the above formulas for the curvature.

===Graph of a function===
The [[graph of a function]] {{math|''y'' {{=}} ''f''(''x'')}}, is a special case of a parametrized curve, of the form
:<math>\begin{align}
x&=t\\
y&=f(t).
\end{align}</math>
As the first and second derivatives of {{mvar|x}} are 1 and 0, previous formulas simplify to 
:<math>\kappa = \frac{\left|y''\right|}{\bigl(1+{y'}^2\bigr)\vphantom{'}^{3/2}},</math>
for the curvature, and to
:<math>k = \frac{y''}{\bigl(1+{y'}^2\bigr)\vphantom{'}^{3/2}},</math>
for the signed curvature.

In the general case of a curve, the sign of the signed curvature is somewhat arbitrary, as it depends on the orientation of the curve. In the case of the graph of a function, there is a natural orientation by increasing values of {{mvar|x}}. This makes significant the sign of the signed curvature.

The sign of the signed curvature is the same as the sign of the second derivative of {{mvar|f}}. If it is positive then the graph has an upward concavity, and, if it is negative the graph has a downward concavity. If it is zero, then  one has an [[inflection point]] or an [[undulation point]].

When the [[slope]] of the graph (that is the derivative of the function) is small, the signed curvature is well approximated by the second derivative. More precisely, using [[big O notation]], one has 
:<math>k(x)=y'' \Bigl(1 + O\bigl({\textstyle y'}^2\bigr) \Bigr).</math>
 
It is common in [[physics]] and [[engineering]] to approximate the curvature with the second derivative, for example, in [[beam theory]] or for deriving the [[wave equation]] of a string under tension, and other applications where small slopes are involved. This often allows systems that are otherwise [[Nonlinear system|nonlinear]] to be treated approximately as linear.

===Polar coordinates===

If a curve is defined in [[polar coordinates]] by the radius expressed as a function of the polar angle, that is {{mvar|r}} is a function of {{mvar|θ}}, then its curvature is
:<math>\kappa(\theta) = \frac{\left|r^2 + 2{r'}^2 - r\, r''\right|}{\bigl(r^2+{r'}^2 \bigr)\vphantom{'}^{3/2}}</math>
where the prime refers to differentiation with respect to {{mvar|θ}}.

This results from the formula for general parametrizations, by considering the parametrization 
:<math>\begin{align}
x&=r\cos \theta\\
y&=r\sin \theta
\end{align}</math>

=== Implicit curve ===

For a curve defined by an [[implicit equation]] {{math|''F''(''x'', ''y'') {{=}} 0}} with [[partial derivatives]] denoted  {{mvar|F<sub>x</sub>}} , {{mvar|F<sub>y</sub>}} , {{mvar|F<sub>xx</sub>}} , {{mvar|F<sub>xy</sub>}} , {{mvar|F<sub>yy</sub>}} ,
the curvature is given by<ref>{{citation |last=Goldman |first=Ron |year=2005 |title=Curvature formulas for implicit curves and surfaces |journal=Computer Aided Geometric Design |volume=22 |issue=7 |pages=632–658 |citeseerx=10.1.1.413.3008 |doi=10.1016/j.cagd.2005.06.005}}</ref>
:<math>\kappa = \frac{\left|F_y^2F_{xx}-2F_xF_yF_{xy}+F_x^2F_{yy}\right|}{\bigl(F_x^2+F_y^2\bigr)\vphantom{'}^{3/2}}.</math>

The signed curvature is not defined, as it depends on an orientation of the curve that is not provided by the implicit equation. Note that changing {{mvar|F}} into {{math|–''F''}} would not change the curve defined by {{math|''F''(''x'', ''y'') {{=}} 0}}, but it would change the sign of the numerator if the [[absolute value]] were omitted in the preceding formula.

A point of the curve where {{math|1=''F<sub>x</sub>'' = ''F<sub>y</sub>'' = 0}} is a [[singular point of a curve|singular point]], which means that the curve is not differentiable at this point, and thus that the curvature is not defined (most often, the point is either a crossing point or a [[cusp (singularity)|cusp]]).

The above formula for the curvature can be derived from the expression of the curvature of the graph of a function by using the [[implicit function theorem]] and the fact that, on such a curve, one has 
:<math>\frac {dy}{dx}=-\frac{F_x}{F_y}.</math>

=== Examples ===
It can be useful to verify on simple examples that the different formulas given in the preceding sections give the same result.

====Circle====
A common parametrization of a [[circle]] of radius {{mvar|r}} is {{math|1='''γ'''(''t'') = (''r'' cos ''t'', ''r'' sin ''t'')}}. The formula for the curvature gives
:<math>k(t)= \frac{r^2\sin^2 t + r^2\cos^2 t}{\bigl(r^2\cos^2 t+r^2\sin^2 t\bigr)\vphantom{'}^{3/2}} = \frac 1r.</math>

It follows, as expected, that the radius of curvature is the radius of the circle, and that the center of curvature is the center of the circle.

The circle is a rare case where the arc-length parametrization is easy to compute, as it is
:<math>\boldsymbol\gamma(s)= \left(r\cos \frac sr,\, r\sin \frac sr\right).</math>
It is an arc-length parametrization, since the norm of
:<math>\boldsymbol\gamma'(s) = \left(-\sin \frac sr,\, \cos \frac sr\right)</math>
is equal to one. This parametrization gives the same value for the curvature, as it amounts to division by {{math|''r''<sup>3</sup>}} in both the numerator and the denominator in the preceding formula.

The same circle can also be defined by the implicit equation {{math|1= ''F''(''x'', ''y'') = 0}} with {{math|1=''F''(''x'', ''y'') = ''x''{{sup|2}} + ''y''{{sup|2}} – ''r''{{sup|2}}}}. Then, the formula for the curvature in this case  gives
:<math>\begin{align}
\kappa &= \frac{\left|F_y^2F_{xx}-2F_xF_yF_{xy}+F_x^2F_{yy}\right|}{\bigl(F_x^2+F_y^2\bigr)\vphantom{'}^{3/2}}\\
&=\frac{8y^2 + 8x^2}{\bigl(4x^2+4y^2\bigr)\vphantom{'}^{3/2}}\\
&=\frac {8r^2}{\bigl(4r^2\bigr)\vphantom{'}^{3/2}} =\frac1r.\end{align}</math>

====Parabola====
Consider the [[parabola]] {{math|''y'' {{=}} ''ax''<sup>2</sup> + ''bx'' + ''c''}}.

It is the graph of a function, with derivative {{math|2''ax'' + ''b''}}, and second derivative {{math|2''a''}}. So, the signed curvature is 
:<math>k(x)=\frac{2a}{ \bigl(1+\left(2ax+b\right)^2\bigr)\vphantom{)}^{3/2}}.</math>
It has the sign of {{mvar|a}} for all values of {{mvar|x}}. This means that, if {{math|''a'' > 0}}, the concavity is upward directed everywhere; if {{math|''a'' < 0}}, the concavity is downward directed; for {{math|1=''a'' = 0}}, the curvature is zero everywhere, confirming that the parabola degenerates into a line in this case.

The (unsigned) curvature is maximal for {{math|1=''x'' = –{{sfrac|''b''|2''a''}}}}, that is at the [[stationary point]] (zero derivative) of the function, which is the [[vertex (curve)|vertex]] of the parabola.

Consider the parametrization {{math|'''γ'''(''t'') {{=}} (''t'', ''at''<sup>2</sup> + ''bt'' + ''c'') {{=}} (''x'', ''y'')}}. The first derivative of {{mvar|x}} is {{math|1}}, and the second derivative is zero. Substituting into the formula for general parametrizations gives exactly the same result as above, with {{mvar|x}} replaced by {{mvar|t}}. If we use primes for derivatives with respect to the parameter {{mvar|t}}.

The same parabola can also be defined by the implicit equation {{math|1= ''F''(''x'', ''y'') = 0}} with {{math|''F''(''x'', ''y'') {{=}} ''ax''<sup>2</sup> + ''bx'' + ''c'' – ''y''}}. As {{math|1=''F{{sub|y}}'' = –1}}, and {{math|1=''F{{sub|yy}}'' = ''F{{sub|xy}}'' = 0}}, one obtains exactly the same value for the (unsigned) curvature. However, the signed curvature is meaningless here, as {{math|1=–''F''(''x'', ''y'') = 0}} is a valid implicit equation for the same parabola, which gives the opposite sign for the curvature.

===Frenet–Serret formulas for plane curves===
[[File:FrenetTN.svg|thumb|right|350px|The vectors {{math|'''T'''}} and {{math|'''N'''}} at two points on a plane curve, a translated version of the second frame (dotted), and {{math|''δ'''''T'''}} the change in {{math|'''T'''}}. Here {{mvar|δs}} is the distance between the points. In the limit {{math|{{sfrac|''d'''''T'''|''ds''}}}} will be in the direction {{math|'''N'''}}. The curvature describes the rate of rotation of the frame.]]
The expression of the curvature [[#In terms of arc-length parametrization|In terms of arc-length parametrization]] is essentially the [[Frenet–Serret formulas|first Frenet–Serret formula]]
:<math>\mathbf T'(s) = \kappa(s) \mathbf N(s),</math>
where the primes refer to the derivatives with respect to the arc length {{mvar|s}}, and {{math|'''N'''(''s'')}} is the normal unit vector in the direction of {{math|'''T'''′(s)}}.

As planar curves have zero [[torsion of curves|torsion]], the second Frenet–Serret formula provides the relation
:<math>\begin{align}
\frac {d\mathbf{N}}{ds} &= -\kappa\mathbf{T},\\
            &= -\kappa\frac{d\boldsymbol{\gamma}}{ds}.
\end{align}</math>

For a general parametrization by a parameter {{mvar|t}}, one needs expressions involving derivatives with respect to {{mvar|t}}. As these are obtained by multiplying by {{sfrac|''ds''|''dt''}} the derivatives with respect to {{mvar|s}}, one has, for any proper parametrization
:<math>
\mathbf{N}'(t) = -\kappa(t)\boldsymbol{\gamma}'(t).
</math>

===Curvature comb===
[[File:Curvature comb.png|alt=Curvature comb|thumb|248x248px|Curvature comb]]
A ''curvature comb''<ref>{{citation |last=Farin |first=Gerald |date=November 2016 |title=Curvature combs and curvature plots |journal=Computer-Aided Design |volume=80 |pages=6–8 |doi=10.1016/j.cad.2016.08.003}}</ref> can be used to represent graphically the curvature of every point on a curve. If <math>t \mapsto x(t)</math> is a parametrised curve its comb is defined as the parametrized curve
:<math> t \mapsto x(t) + d\kappa(t)n(t)</math>
where <math>\kappa, n</math> are the curvature and normal vector respectively, and <math>d</math> is a scaling factor (to be chosen as to enhance the graphical representation).

Curvature combs are useful when combining two different curves in CAD environments. They provide a visual representation of the continuity between the curves. The continuity can be defined as being in one of four levels.

G0 : The 2 curvature combs are at an angle at the junction.

G1 : The teeth of the 2 combs are parallel at the junction but are of different length.

G2 : The teeth are parallel and of the same length. However the tangents of the 2 combs are not the same.

G3 : The teeth are parallel and of the same length and the tangents of the 2 combs are the same.

The above image shows a G2 continuity at the 2 junctions.

==Space curves==
[[File:Torus-Knot uebereinander animated.gif|thumb|upright|Animation of the curvature and the acceleration vector {{math|'''T'''′(''s'')}}]]
As in the case of curves in two dimensions, the curvature of a regular [[space curve]] {{mvar|C}} in three dimensions (and higher) is the magnitude of the acceleration of a particle moving with unit speed along a curve. Thus if {{math|'''γ'''(''s'')}} is the arc-length parametrization of {{mvar|C}} then the unit tangent vector {{math|'''T'''(''s'')}} is given by
:<math>\mathbf{T}(s) = \boldsymbol{\gamma}'(s)</math>
and the curvature is the magnitude of the acceleration:
:<math>\kappa(s) = \|\mathbf{T}'(s)\| = \|\boldsymbol{\gamma}''(s)\|.</math>
The direction of the acceleration is the unit normal vector {{math|'''N'''(''s'')}}, which is defined by
:<math>\mathbf{N}(s) = \frac{\mathbf{T}'(s)}{\|\mathbf{T}'(s)\|}.</math>

The plane containing the two vectors {{math|'''T'''(''s'')}} and {{math|'''N'''(''s'')}} is the [[osculating plane]] to the curve at {{math|'''γ'''(''s'')}}. The curvature has the following geometrical interpretation. There exists a circle in the osculating plane tangent to {{math|'''γ'''(''s'')}} whose [[Taylor series]] to second order at the point of contact agrees with that of {{math|'''γ'''(''s'')}}. This is the [[osculating circle]] to the curve. The radius of the circle {{math|''R''(''s'')}} is called the [[radius of curvature]], and the curvature is the reciprocal of the radius of curvature:

:<math>\kappa(s) = \frac{1}{R(s)}.</math>

The tangent, curvature, and normal vector together describe the second-order behavior of a curve near a point. In three dimensions, the third-order behavior of a curve is described by a related notion of [[torsion of a curve|torsion]], which measures the extent to which a curve tends to move as a helical path in space. The torsion and curvature are related by the [[Frenet–Serret formulas]] (in three dimensions) and [[differential geometry of curves|their generalization]] (in higher dimensions).

===General expressions===
For a parametrically defined space curve in three dimensions given in Cartesian coordinates by {{math|'''γ'''(''t'') {{=}} (''x''(''t''), ''y''(''t''), ''z''(''t''))}}, the curvature is

:<math>
\kappa=\frac{\sqrt{\bigl(z''y'-y''z'\bigr)\vphantom{'}^2+\bigl(x''z'-z''x'\bigr)\vphantom{'}^2+\bigl(y''x'-x''y'\bigr)\vphantom{'}^2}}
  {\bigl({x'}^2+{y'}^2+{z'}^2\bigr)\vphantom{'}^{3/2}},
</math>

where the prime denotes differentiation with respect to the parameter {{mvar|t}}. This can be expressed independently of the [[coordinate system]] by means of the formula<ref>A proof of this can be found at [https://mathworld.wolfram.com/Curvature.html the article on curvature] at [[Wolfram MathWorld]].</ref>

:<math>\kappa = \frac{\bigl\|\boldsymbol{\gamma}' \times \boldsymbol{\gamma}''\bigr\|}{\bigl\|\boldsymbol{\gamma}'\bigr\|\vphantom{'}^3}</math>

where × denotes the [[cross product|vector cross product]]. The following formula is valid for the curvature of curves in a Euclidean space of any dimension:

:<math>
\kappa = \frac{\sqrt{ \bigl\|\boldsymbol{\gamma}'\bigr\|\vphantom{'}^2 \bigl\|\boldsymbol{\gamma}''\bigr\|\vphantom{'}^2- \bigl(\boldsymbol{\gamma}'\cdot \boldsymbol{\gamma}''\bigr)\vphantom{'}^2 } }
{\bigl\|\boldsymbol{\gamma}'\bigr\|\vphantom{'}^3}.
</math>

===Curvature from arc and chord length===
Given two points {{mvar|P}} and {{mvar|Q}} on {{mvar|C}}, let {{math|''s''(''P'',''Q'')}} be the arc length of the portion of the curve between {{mvar|P}} and {{mvar|Q}} and let {{math|''d''(''P'',''Q'')}} denote the length of the line segment from {{mvar|P}} to {{mvar|Q}}. The curvature of {{mvar|C}} at {{mvar|P}} is given by the limit{{citation needed|date=December 2010}}

:<math>\kappa(P) = \lim_{Q\to P}\sqrt\frac{24\bigl(s(P,Q)-d(P,Q)\bigr)}{s(P,Q)\vphantom{Q}^3}</math>

where the limit is taken as the point {{mvar|Q}} approaches {{mvar|P}} on {{mvar|C}}. The denominator can equally well be taken to be {{math|''d''(''P'',''Q'')<sup>3</sup>}}. The formula is valid in any dimension. Furthermore, by considering the limit independently on either side of {{mvar|P}}, this definition of the curvature can sometimes accommodate a singularity at {{mvar|P}}. The formula follows by verifying it for the osculating circle.

==Surfaces==
{{broader|Differential geometry of surfaces}}
The curvature of curves drawn on a [[surface (mathematics)|surface]] is the main tool for the defining and studying the curvature of the surface.

=== Curves on surfaces===
For a curve drawn on a surface (embedded in three-dimensional [[Euclidean space]]), several curvatures are defined, which relates the direction of curvature to the surface's unit [[normal vector]], including the:
* [[normal curvature]]
*[[geodesic curvature]] 
*[[geodesic torsion]] 
Any non-singular curve on a smooth surface has its tangent vector {{math|'''T'''}} contained in the [[tangent plane]] of the surface. The '''normal curvature''', {{math|''k''<sub>n</sub>}}, is the curvature of the curve projected onto the plane containing the curve's tangent {{math|'''T'''}} and the surface normal {{math|'''u'''}}; the '''geodesic curvature''', {{math|''k''<sub>g</sub>}}, is the curvature of the curve projected onto the surface's tangent plane; and the '''geodesic torsion''' (or '''relative torsion'''), {{math|''τ''<sub>r</sub>}}, measures the rate of change of the surface normal around the curve's tangent.

Let the curve be [[arc-length parametrization|arc-length parametrized]], and let {{math|'''t''' {{=}} '''u''' × '''T'''}} so that {{math|'''T''', '''t''', '''u'''}} form an [[orthonormal basis]], called the '''[[Darboux frame]]'''. The above quantities are related by:
:<math>\begin{pmatrix} \mathbf{T}'\\ \mathbf{t}'\\ \mathbf{u}' \end{pmatrix} =
\begin{pmatrix}
0&\kappa_\mathrm{g}&\kappa_\mathrm{n}\\
-\kappa_\mathrm{g}&0&\tau_\mathrm{r}\\
-\kappa_\mathrm{n}&-\tau_\mathrm{r}&0
\end{pmatrix}
\begin{pmatrix} \mathbf{T}\\ \mathbf{t}\\ \mathbf{u} \end{pmatrix}</math>

====Principal curvature====
[[File:Minimal surface curvature planes-en.svg|thumb|300px|right|[[Saddle surface]] with normal planes in directions of principal curvatures]]
{{Main|Principal curvature}}
All curves on the surface with the same tangent vector at a given point will have the same normal curvature, which is the same as the curvature of the curve obtained by intersecting the surface with the plane containing {{math|'''T'''}} and {{math|'''u'''}}. Taking all possible tangent vectors, the maximum and minimum values of the normal curvature at a point are called the '''principal curvatures''', {{math|''k''<sub>1</sub>}} and {{math|''k''<sub>2</sub>}}, and the directions of the corresponding tangent vectors are called '''principal normal directions'''.

===Normal sections===
Curvature can be evaluated along surface [[normal section]]s, similar to {{section link||Curves on surfaces}} above (see for example the [[Earth radius of curvature]]).

===Developable surfaces===
Some curved surfaces, such as those made from a smooth sheet of paper, can be flattened down into the plane without distorting their intrinsic features in any way. Such [[developable surface]]s have zero Gaussian curvature (see below).<ref>[https://mathworld.wolfram.com/DevelopableSurface.html developable surface], Mathworld. (Retrieved 11 February 2021)</ref>

===Gaussian curvature===
{{main|Gaussian curvature}}
In contrast to curves, which do not have intrinsic curvature, but do have extrinsic curvature (they only have a curvature given an embedding), surfaces can have intrinsic curvature, independent of an embedding. The [[Gaussian curvature]], named after [[Carl Friedrich Gauss]], is equal to the product of the principal curvatures, {{math|''k''<sub>1</sub>''k''<sub>2</sub>}}. It has a dimension of length<sup>−2</sup> and is positive for [[sphere]]s, negative for one-sheet [[hyperboloid]]s and zero for planes and [[Cylinder#Cylindrical surfaces|cylinders]]. It determines whether a surface is [[locally]] [[:wikt:convex|convex]] (when it is positive) or locally saddle-shaped (when it is negative).

Gaussian curvature is an ''intrinsic'' property of the surface, meaning it does not depend on the particular [[embedding]] of the surface; intuitively, this means that ants living on the surface could determine the Gaussian curvature. For example, an ant living on a sphere could measure the sum of the interior angles of a triangle and determine that it was greater than 180 degrees, implying that the space it inhabited had positive curvature. On the other hand, an ant living on a cylinder would not detect any such departure from [[Euclidean geometry]]; in particular the ant could not detect that the two surfaces have different mean curvatures (see below), which is a purely extrinsic type of curvature.

Formally, Gaussian curvature only depends on the [[Riemannian metric]] of the surface. This is [[Carl Friedrich Gauss|Gauss]]'s celebrated [[Theorema Egregium]], which he found while concerned with geographic surveys and mapmaking.

An intrinsic definition of the Gaussian curvature at a point {{mvar|P}} is the following: imagine an ant which is tied to {{mvar|P}} with a short thread of length {{mvar|r}}. It runs around {{mvar|P}} while the thread is completely stretched and measures the length {{math|''C''(''r'')}} of one complete trip around {{mvar|P}}. If the surface were flat, the ant would find {{math|''C''(''r'') {{=}} 2π''r''}}. On curved surfaces, the formula for {{math|''C''(''r'')}} will be different, and the Gaussian curvature {{mvar|K}} at the point {{mvar|P}} can be computed by the [[Bertrand–Diguet–Puiseux theorem]] as

:<math> K = \lim_{r\to 0^+} 3\left(\frac{2\pi r-C(r)}{\pi r^3}\right).</math>

The [[integral]] of the Gaussian curvature over the whole surface is closely related to the surface's [[Euler characteristic]]; see the [[Gauss–Bonnet theorem]].

The discrete analog of curvature, corresponding to curvature being concentrated at a point and particularly useful for [[polyhedra]], is the [[defect (geometry)|(angular) defect]]; the analog for the [[Gauss–Bonnet theorem]] is [[Defect (geometry)#Descartes.27 theorem|Descartes' theorem on total angular defect]].

Because (Gaussian) curvature can be defined without reference to an embedding space, it is not necessary that a surface be embedded in a higher-dimensional space in order to be curved. Such an intrinsically curved two-dimensional surface is a simple example of a [[Riemannian manifold]].

===Mean curvature===
{{main|Mean curvature}}
The mean curvature is an ''extrinsic'' measure of curvature equal to half the sum of the [[principal curvature]]s, {{math|{{sfrac|''k''<sub>1</sub> + ''k''<sub>2</sub>|2}}}}. It has a dimension of length<sup>−1</sup>. Mean curvature is closely related to the first variation of [[surface area]]. In particular, a [[minimal surface]] such as a [[soap film]] has mean curvature zero and a [[soap bubble]] has constant mean curvature. Unlike Gauss curvature, the mean curvature is extrinsic and depends on the embedding, for instance, a [[cylinder (geometry)|cylinder]] and a plane are locally [[isometry|isometric]] but the mean curvature of a plane is zero while that of a cylinder is nonzero.

=== Second fundamental form ===
{{main|Second fundamental form}}

The intrinsic and extrinsic curvature of a surface can be combined in the second fundamental form. This is a [[quadratic form]] in the tangent plane to the surface at a point whose value at a particular tangent vector {{math|'''X'''}} to the surface is the normal component of the acceleration of a curve along the surface tangent to {{math|'''X'''}}; that is, it is the normal curvature to a curve tangent to {{math|'''X'''}} (see [[#Curves on surfaces|above]]). Symbolically,
:<math>\operatorname{I\!I}(\mathbf{X},\mathbf{X}) = \mathbf{N}\cdot (\nabla_\mathbf{X} \mathbf{X})</math>
where {{math|'''N'''}} is the unit normal to the surface. For unit tangent vectors {{math|'''X'''}}, the second fundamental form assumes the maximum value {{math|''k''<sub>1</sub>}} and minimum value {{math|''k''<sub>2</sub>}}, which occur in the principal directions {{math|'''u'''<sub>1</sub>}} and {{math|'''u'''<sub>2</sub>}}, respectively. Thus, by the [[principal axis theorem]], the second fundamental form is
:<math>\operatorname{I\!I}(\mathbf{X},\mathbf{X}) = k_1\left(\mathbf{X}\cdot \mathbf{u}_1\right)^2 + k_2\left(\mathbf{X}\cdot \mathbf{u}_2\right)^2.</math>
Thus the second fundamental form encodes both the intrinsic and extrinsic curvatures.

=== Shape operator ===
{{further|Shape operator}}
An encapsulation of surface curvature can be found in the shape operator, {{math|''S''}}, which is a [[Self-adjoint operator|self-adjoint]] [[linear operator]] from the tangent plane to itself (specifically, the differential of the [[Gauss map]]).

For a surface with tangent vectors {{math|'''X'''}} and normal {{math|'''N'''}}, the shape operator can be expressed compactly in [[Einstein notation|index summation notation]] as

:<math>\partial_a \mathbf{N} = -S_{ba} \mathbf{X}_{b} .</math>
(Compare the [[#Frenet–Serret formulas for plane curves|alternative expression]] of curvature for a plane curve.)

The [[Weingarten equations]] give the value of {{math|''S''}} in terms of the coefficients of the [[first fundamental form|first]] and [[second fundamental form]]s as
:<math>S= \left(EG-F^2\right)^{-1}\begin{pmatrix}
eG-fF& fG-gF \\
fE-eF & gE- fF\end{pmatrix}.</math>
The principal curvatures are the [[Eigenvalues and eigenvectors|eigenvalues]] of the shape operator, the principal curvature directions are its [[Eigenvalues and eigenvectors|eigenvectors]], the Gauss curvature is its [[determinant]], and the mean curvature is half its [[trace (linear algebra)|trace]].

==Curvature of space{{anchor|Space}}==
{{further|Curvature of Riemannian manifolds|Curved space}}
{{distinguish-redirect|Curvature of space|Curvature of space-time}}

By extension of the former argument, a space of three or more dimensions can be intrinsically curved. The curvature is ''intrinsic'' in the sense that it is a property defined at every point in the space, rather than a property defined with respect to a larger space that contains it. In general, a curved space may or may not be conceived as being embedded in a higher-dimensional [[ambient space]]; if not then its curvature can only be defined intrinsically.

After the discovery of the intrinsic definition of curvature, which is closely connected with [[non-Euclidean geometry]], many mathematicians and scientists questioned whether ordinary physical space might be curved, although the success of Euclidean geometry up to that time meant that the radius of curvature must be astronomically large. In the theory of [[general relativity]], which describes [[gravity]] and [[physical cosmology|cosmology]], the idea is slightly generalised to the "curvature of [[spacetime]]"; in relativity theory spacetime is a [[pseudo-Riemannian manifold]]. Once a time coordinate is defined, the three-dimensional space corresponding to a particular time is generally a curved Riemannian manifold; but since the time coordinate choice is largely arbitrary, it is the underlying spacetime curvature that is physically significant.

Although an arbitrarily curved space is very complex to describe, the curvature of a space which is locally [[isotropic]] and [[Homogeneous space|homogeneous]] is described by a single Gaussian curvature, as for a surface; mathematically these are strong conditions, but they correspond to reasonable physical assumptions (all points and all directions are indistinguishable). A positive curvature corresponds to the inverse square radius of curvature; an example is a sphere or [[hypersphere]]. An example of negatively curved space is [[hyperbolic geometry]] (see also: [[non-positive curvature]]). A space or space-time with zero curvature is called '''''flat'''''.{{anchor|Flat space}} For example, [[Euclidean space]] is an example of a flat space, and [[Minkowski space]] is an example of a flat spacetime. There are other examples of flat geometries in both settings, though. A [[torus]] or a [[Cylinder (geometry)|cylinder]] can both be given flat metrics, but differ in their [[topology]]. Other topologies are also possible for curved space {{xref|(see also: [[Shape of the universe]])}}.

==Generalizations==
[[File:Parallel Transport.svg|thumb|Moving a vector along a curve from A → N → B → A produces another vector. The inability to return to the initial vector is measured by the holonomy of the surface. In a space with no curvature, the angle α is 0 degrees, and in a space with curvature, the angle α is greater than 0 degrees. The more space is curved, the greater the magnitude of the angle α. ]]
The mathematical notion of ''curvature'' is also defined in much more general contexts.<ref>{{citation |last1=Kobayashi |first1=Shōshichi |author-link1=Shoshichi Kobayashi |title=Foundations of Differential Geometry |title-link=Foundations of Differential Geometry |last2=Nomizu |first2=Katsumi |author-link2=Katsumi Nomizu |publisher=Interscience |year=1963 |isbn=978-0-470-49647-3 |place=New York |chapter=2–3}}</ref> Many of these generalizations emphasize different aspects of the curvature as it is understood in lower dimensions.

One such generalization is kinematic. The curvature of a curve can naturally be considered as a kinematic quantity, representing the force felt by a certain observer moving along the curve; analogously, curvature in higher dimensions can be regarded as a kind of [[tidal force]] (this is one way of thinking of the [[sectional curvature]]). This generalization of curvature depends on how nearby test particles diverge or converge when they are allowed to move freely in the space; see [[Jacobi field]].

Another broad generalization of curvature comes from the study of [[parallel transport]] on a surface. For instance, if a vector is moved around a loop on the surface of a sphere keeping parallel throughout the motion, then the final position of the vector may not be the same as the initial position of the vector. This phenomenon is known as [[holonomy]].<ref>{{citation |last1=Henderson |first1=David W. |author-link1=David W. Henderson |title=Experiencing Geometry: Euclidean and Non-Euclidean with History |last2=Taimin̦a |first2=Daina |author-link2=Daina Taimiņa |publisher=Pearson Prentice Hall |year=2005 |isbn=978-0-13-143748-7 |edition=3rd |place=Upper Saddle River, NJ |pages=98–99 |doi=10.3792/euclid/9781429799850 |doi-access=free }}</ref> Various generalizations capture in an abstract form this idea of curvature as a measure of holonomy; see [[curvature form]]. A closely related notion of curvature comes from [[gauge theory]] in physics, where the curvature represents a field and a [[vector potential]] for the field is a quantity that is in general path-dependent: it may change if an observer moves around a loop.

Two more generalizations of curvature are the [[scalar curvature]] and [[Ricci curvature]]. In a curved surface such as the sphere, the area of a disc on the surface differs from the area of a disc of the same radius in flat space. This difference (in a suitable limit) is measured by the scalar curvature. The difference in area of a sector of the disc is measured by the Ricci curvature. Each of the scalar curvature and Ricci curvature are defined in analogous ways in three and higher dimensions. They are particularly important in relativity theory, where they both appear on the side of [[Einstein's field equations]] that represents the geometry of spacetime (the other side of which represents the presence of matter and energy). These generalizations of curvature underlie, for instance, the notion that curvature can be a property of a [[measure (mathematics)|measure]]; see [[curvature of a measure]].

Another generalization of curvature relies on the ability to [[comparison theorem|compare]] a curved space with another space that has ''constant'' curvature. Often this is done with triangles in the spaces. The notion of a triangle makes senses in [[metric space]]s, and this gives rise to [[CAT(k) space|{{math|CAT(''k'')}} spaces]].

==See also==
* [[Curvature form]] for the appropriate notion of curvature for [[vector bundle]]s and [[principal bundle]]s with [[Connection (mathematics)|connection]]
* [[Curvature of a measure]] for a notion of curvature in [[measure theory]]
* [[Parametric surface#Curvature|Curvature of parametric surfaces]]
* [[Curvature of Riemannian manifolds]] for generalizations of Gauss curvature to higher-dimensional [[Riemannian manifold]]s
* [[Curvature vector]] and [[geodesic curvature]] for appropriate notions of curvature of ''curves in'' Riemannian manifolds, of any dimension
* [[Degree of curvature]]
* [[Differential geometry of curves]] for a full treatment of curves embedded in a Euclidean space of arbitrary dimension
* [[Dioptre]], a measurement of curvature used in optics
* [[Evolute]], the locus of the centers of curvature of a given curve
* [[Fundamental theorem of curves]]
* [[Gauss–Bonnet theorem]] for an elementary application of curvature
* [[Gauss map]] for more geometric properties of Gauss curvature
* [[Gauss's principle of least constraint]], an expression of the [[Principle of Least Action]]
* [[Mean curvature]] at one point on a surface
* [[Minimum railway curve radius]]
* [[Radius of curvature (mathematics)|Radius of curvature]]
* [[Second fundamental form]] for the extrinsic curvature of hypersurfaces in general
* [[Sinuosity]]
* [[Torsion of a curve]]

==Notes==
{{Reflist}}

==References==
* {{citation |last=Coolidge |first=Julian L. |author-link=Julian Coolidge |title=The Unsatisfactory Story of Curvature |journal=American Mathematical Monthly |volume=59 |issue=6 |date=June 1952 |pages=375–379 |doi=10.2307/2306807 |jstor=2306807 }}
* {{SpringerEOM |title=Curvature |id=Curvature |oldid=12026 |first=Dmitriĭ Dmitrievich |last=Sokolov }}
* {{citation |last=Kline |first=Morris |author-link=Morris Kline |url=https://books.google.com/books?id=TzNwDwAAQBAJ |title=Calculus: An Intuitive and Physical Approach |date=1998 |publisher=Dover |isbn=978-0-486-40453-0 |edition=2nd |place=Mineola, NY |pages=457–461}} ({{Google books|YdjK_rD7BEkC|restricted online copy|page=457}})
* {{citation |first=A. Albert |last=Klaf |title=Calculus Refresher |publisher=Dover |date=1956 |isbn=978-0-486-20370-6 |pages=[https://archive.org/details/calculusrefreshe00klaf/page/151 151–168] |url-access=registration |url=https://archive.org/details/calculusrefreshe00klaf/page/151 }} ({{Google books |NR6ZuvBP3zwC |restricted online copy |page=151 }})
* {{citation |last=Casey |first=James |url=https://books.google.com/books?id=HzoDCAAAQBAJ |title=Exploring Curvature |date=1996 |publisher=Vieweg |isbn=978-3-528-06475-4 |series=Vieweg Mathematics |place=Braunschweig}}

==External links==
{{Wiktionary|curvature}}
{{Commonscat|Curvature}}
*[https://feynmanlectures.caltech.edu/II_42.html The Feynman Lectures on Physics Vol. II Ch. 42: Curved Space]
*[https://web.archive.org/web/20071106083431/http://www3.villanova.edu/maple/misc/history_of_curvature/k.htm The History of Curvature]
*[http://www.mathpages.com/rr/s5-03/5-03.htm Curvature, Intrinsic and Extrinsic] at MathPages

{{curvature}}
{{Curves}}
{{Authority control}}

[[Category:Curvature (mathematics)| ]]
[[Category:Multivariable calculus]]
[[Category:Articles containing video clips]]