University of Utah Undergraduate Colloquium

Integral Geometry & Geometric Probability

Andrejs Treibergs
University of Utah
Wednesday, October 1, 2008
2. References
Presented to the Undergraduate Colloquium, University of Utah,
Salt lake City, Utah on October 7, 2008.
This talk has also been presented to the 2008 Summer Mathematics
Research Experience for Undergraduates (REU) Seminar at Brigham
Young University, Provo, Utah on July 29, 2008.
URL of Beamer Slides: Integral Geometry and Geometric Probability
http://www.math.utah.edu/treiberg/IntGeomSlides.pdf
Some excellent references to Integral Geometry.
Luis A. Santalo, Integral Geometry and Geometric Probability,
Addison-Wesley, Reading, MA, 1976.
Herbert Solomon, Geometric Probability (CBMS-NSF Regional
Conference Series in Applied Mathematics 28), Society for Industrial
and Applied Mathetaics, Philadelphia, 1978.
Wilhelm Blaschke, Vorlesungen uber Integralgeometrie I, II, Chelsea,
New York, 1949, (2nd ed. orig. pub. B. G. Teubner, Leipzig, 1935.)
3. Outline.
Integral Geometry, known in applied circles as Geometric Probability, is
somewhat of a mathematical antique (and therefore it is a favorite of
mine!) From it developed many modern topics: geometric measure
theory, stereometry, tomography, characteristic classes. . .
1 Integral geometry examples:
Buons needle problem.
Firerys dice problem
2 Kinematic measure.
3 Poincares Formula for average number of intersections of curves.
4 Cauchys Formula for the average projected length.
5 Croftons Formula for the average chord length.
6 Santalos & Blaschkes Formuls for the averages over the of the
intesection of two domains.
7 Application to the Isoperimetric Inequality.
4. Integral Geometry. First Example.
Theorem (Buons Needle Problem [1733])
Parallel lines on a wooden oor are a distance d apart from each other.
A needle of length ( < d) is randomly dropped onto the oor. Then
the probability that the needle will touch one of the lines is
P =
2
d
.
Figure: Buons Needle is randomly dropped
5. Integral Geometry. Second Example.
Theorem (Fireys Colliding Dice Problem [1974])
Suppose
1
and
2
are disjoint unit cubes in R
3
. In a random collision,
the probablity that the cubes collide edge-to-edge slightly exceeds the
probability that the cubes collide corner-to-face. Indeed,
0.54

= P(collide edge-to-edge) > P(collide corner-to-face.)

= 0.46.
Figure: Almost all random cube collisions are edge-to-edge or corner-to-face.
6. Coordinates of a line.
An unoriented line in the plane is determined by two numbers, p the
distance to the origin and , the direction to the closest point.
The variable range is 0 p and 0 < 2.
Equivalently, we may take the range < p < and 0 < .
Figure: (p, ) coordinates for the line L.
The equation of the line L(p, ) in Cartesian coordinates is
cos()x + sin()y = p (1)
7. Rigid motions of the Euclidean Plane.
A rigid motion M of a set of points is given by a rotation by an angle
followed by a translation by the vector (x
0
, y
0
). Thus
_
x

_
= M
_
x
y
_
=
_
x
0
y
0
_
+
_
cos sin
sin cos
__
x
y
_
Thus the inverse motion is therefore given by
_
x
y
_
= M
1
_
x

_
=
_
cos sin
sin cos
__
x

x
0
y

y
0
_
(2)
The mobile line L(p, ) may be thought of as moving the xed line
L(0, 0) by the translation (x, y) (x + p, y) followed by the rotation
about the origin by angle .
The rst task is to nd a measure on a set of lines that is invariant under
rigid motions. This measure will be called KINEMATIC MEASURE.
8. Kinematic measure.
The kinematic measure for lines in (p, ) coordinates is given by
dK = dp d.
To check that this measure is invariant under rigid motions, let us rst
determine how (p, ) in the equation of the line (1) is changed by a rigid
motion M. We express (x, y) in terms of (x

, y

) using (2)
p = cos()x + sin()y
= cos()
_
cos()(x

x
0
) + sin()(y

y
0
)

+ sin()
_
sin()(x

x
0
) + cos()(y

y
0
)

= [cos cos sin sin ] (x

x
0
)
+ [cos sin + sin cos ] (y

y
0
)
= cos( +)(x

x
0
) + sin( +)(y

y
0
)
or the equation of the new line L

becomes
p + cos( +)x
0
+ sin( +)y
0
= cos( +)x

+ sin( +)y

.
9. Kinematic measure is invariant under rigid motion.
Thus we read o the (p

) coordinates of the line L

= M(L).
p

= p + cos( +)x
0
+ sin( +)y
0

= +.
Then the Jacobian formula for the change in measure is
dp

= |J| dp d
where
J =
(p

)
(p, )
=

p
p

1
0 1

= 1.
Thus we have shown that the kinematic measure is invariant under rigid
motions.
10. Dierential forms version of the same computation.
We view (p

) as function (p, ). The dierentials are thus

dp

= dp +
_
sin( +)x
0
+ cos( +)y
0
_
d,
d

= d.
Recall that wedge is a skew product so that dp d = d dp and
d d = 0. Hence
dp

=
_
dp +
_
sin( +)x
0
+ cos( +)y
0
_
d
_
d
= dp d.
11. The measure of lines that meet a curve.
Let C be a piecewise C
1
curve or network (a union of C
1
curves.) Given a
line L in the plane, let n(L C) be the number of intersection points. If
C contains a linear segment and if L agrees with that segment,
n(C L) = . For any such C, however, the set of lines for which
n = has dK-measure zero.
Figure: Henri Poincare
18541912
Theorem (Poincare Formula for lines [1896])
Let C be a piecewise C
1
curve in the plane.
Then the measure of unoriented lines
meeting C, counted with multiplicity, is
given by
2 L(C) =
_
{L:LC=}
n(C L) dK(L).
12. Key idea in IG: integtrate over a set S in two dierent coordinates.
For simplicity we assume C is a C
1
curve Z(s) =
_
x(s), y(s)
_
,
parameterized by arclength. Thus there are x(s), y(s) C
1
[0, s
0
] such
that the tangent vector

Z = ( x, y) satises |

Z| = 1. By adding the
formulas for C
1
curves gives the formula for integrating a piecewise C
1
curve.
Let us consider a ag which is the set of pairs (L, Z) where L is a line in
the plane and Z L is a point. The set of lines and corresponding points
that touch C gives the subset of the ag
S = {(L, Z); L C = , Z L C}.
The line is determined by the coordinates (p, ) and the point Z L by
an arclength coordinate q along L from the foot-point (p cos , p sin ).
_
{L:LC=}
n dK =
_
{L:LC=}
_

ZLC
1
_
dK (3)
13. Compute the integral of S in dierent coordinates.
On the other hand, the set S can be determined by the point
(x, y) = Z C rst and then L can be any unoriented line through Z of
angle 0 < (positive and negative orientations give the same line).
Thus we may replace (p, ) by the coordinates (s, ). Using
p = x(s) cos + y(s) sin .
( p, ) (, ) [0, ) are same lines as (p, ) [0, ) [0, 2). So
d p =
_
x(s) cos + y(s) sin
_
ds +
_
x(s) sin + y(s) cos
_
d.
Changing to (s, ), using tangent direction ( x, y) = (cos (s), sin (s)),
d p d =

p
s
p

ds d =

cos cos + sin sin

0 1

ds d
= | cos((s) )|ds d.
14. Finish the proof of Poincares Formula.
Using Fubinis theorem (slicing formula), we may reverse the order of
integtation in (3) over the set S,
_
{L:LC=}
_

ZL
1
_
dK =
_
{Z:ZC}
_
{L:ZL}
d p d
=
s
0
_
0

_
0
| cos((s) )|d ds
= 2
_
C
ds
= 2 L(C).
15. Convex sets. First geometric probability example.
A nonempty set R
2
is convex if for every pair of points P, Q ,
the line segment PQ . The integral geometric formulas hold for
convex sets. Since n(L ) is either zero or two for dK-almost all L,
the measure of unoriented lines that meet the a convex set is given by
L() =
_
{L:L=}
dK.
The conditional probability of an event A given the event B is dened to
be P(A|B) =
P(AB)
P(B)
.
Theorem (Sylvesters Problem [1889] )
Let be two bounded convex sets in the plane. Then the
probability that a random line meets given that it meets is
P =
L()
L()
.
16. Another application to Geometric Probability.
Corollary
Let C be a piecewise C
1
curve contained in a compact convex set . Of
all random lines that meet , the expected number of intersections with
with C is
E(n) =
2 L(C)
L()
. (4)
Hence, there are lines that cut C in at least 2 L(C)/ L() points.
Proof. Since is convex, E(n) =
_
{L:LC=}
n dK
_
{L:L=}
dK
=
2 L(C)
L()
.
The maximum of n exceeds the average.
Figure: Average number of intersections L C of a line L meeting .
17. Support function and width.
Figure: Width and support function of convex in direction.
For [0, 2), the support function, h(), is the largest p such that
L(p, ) = . The width is w() = h() + h( +).
18. Another corollary: Mean projected width or Quermassintegral.
Figure: Augustin Louis Cauchy
17891857
Theorem (Cauchys Formula [1841])
Let be a bounded convex domain. Then
L() =
_
2
0
h() d =
_

0
w() d. (5)
L() =
_
{L:L=}
dK =
_
2
0
_
h()
0
dp d
=
_
2
0
h() d =
_

0
h() + h( +) d
=
_

0
w() d.
19. Area in terms of support function.
Theorem
Suppose is a compact, convex
domain with a C
2
boundary. Then
A() =
1
2
2
_
0
h ds =
1
2
2
_
0
h(h+

h) d.
(6)
Write Z() for the point
L(h(), ) . The outer normal
is n() = (cos , sin ).
Z() n() = h()
Since n = (sin , cos ), and

Z is
tangent,

h = n Z + n

Z = n Z.
Thus Z = hn +

h n. Hence,

Z =

hn + h n +

h n

hn = (h +

h) n.
Figure: Area on polar coordinates.
Thus
ds
d
= h +

h so A() =
_

dA
=
1
2
2
_
0
h ds =
1
2
2
_
0
h(h +

h) d.
20. Buons Needle Problem Solution.
Figure: (p, ) coordinates for the closest crack L.
Fix needle N, a line segment of length centered at origin. Move oor.
< d implies only the cracks closest to the origin could touch the needle.
So we consider crack lines L so that dist(L, 0)
d
2
i C L = , where
C the circle about the origin with radius
d
2
.
21. Buons Needle Problem Solution-.
Note that if L N = then n(L N) = 1. The probability of needle
hitting a crack is
P =
_
{L:LN=}
n(L N) dK(L)
_
{L:LC=}
dK(L)
=
2 L(N)
L(C)
=
2
2
d
2
=
2
d
.
An experimental determination of .
=
2
Pd

2
d

n
x
,
where x is the number of times needle touches crack in n trials.
Wolf, in Zurich (1850), tossed 5000 needles and found 3.1596.
A Scotsman, Smith (1855), repeated with n = 3204 and found
3.1553.
22. Croftons Formula.
Figure: Morgan William
Crofton 18261915.
Theorem (Croftons Formula [1868])
Let D R
2
be a domain with compact
closure, L R
2
a random line and

1
(L D) be the length (one-dimensional
measure). Then
A(D) =
_
{L:LD=}

1
(L D) dK(L).
Let the subset of the ag be
S = {(L, Z) : L D = , Z L D}.
A point in S is given by coordinates (p, )
describing the line and q, arclength in L
from the foot point.
23. Proof of Croftons Formula.
Denote the right side by I. By extending < p < , we
double-count the lines.
I =
_
{L:LD=}

1
(L D) dK(L)
=
_
{L:LD=}
__
DL
dq
_
dp d
=
_
2
0
_

0
_

DL
(q) dq dp d
=
1
2
_
2
0
_

DL
(q) dq d p d
where
DL
is the characteristic function:

DL
(q) =
_
1, if q D L;
0, if q / D L.
24. Finish the proof of Croftons Formula.
Observe that for the line L( p, ) we have
DL
(q) = 1 if and only if the
point in the plane corresponding to ( p, q) lies in D, namely
(x, y) = p(cos , sin ) + q(sin , cos )
= ( p cos q sin , p sin + q cos ) D
thus

L( p,)D
(q) =
D
(x, y).
The change of variables to (x, y) is just rotation by angle . Thus
dx dy =
_
cos()d p sin()dq

_
sin()d p + cos()dq

= d p dq.
25. Finish the proof of Croftons Formula-.
Now we think of S another way. First pick Z D and then L is any line
through Z.
I =
1
2
_
2
0
_

DL
(q) dq d p d
=
1
2
_
2
0
_

D
(x, y) dx dy d
=
1
2
_
2
0
A(D) d
= A(D).
26. Application to Geometric Probability
Figure: Two random lines that meet
Corollary (Crofton [1885])
Let be a bounded convex domain
in the plane. Then the probability
that two random lines intersect in
given that they both meet is
P =
2 A()
L()
2
.
By the isoperimetric inequality,
4 A() L()
2
with equality
only for circle, the probability
satises
P
1
2
.
Equality holds i is a round disk.
27. Compute the expected number of intersections of two lines.
Proof. Let L
1
(p
1
,
1
) and L
2
(p
2
,
2
) be two random lines whose invariant
measure is dK
1
dK
2
= dp
1
d
1
dp
2
d
2
.
View
1
= L(p
1
,
1
) as a subset. By (4), the average number of
times that a random line L(p
2
,
2
) meets
1
given that it meets is
E(n) =
2
1
_
L(p
1
,
1
)
_
L()
.
Poincares and Croftons Formul = probability that two lines meet is
P = E(n) =
_
{L
1
:L
1
=}
_
{L
2
:L
2
=}
n(
1
L
2
) dK
2
dK
1
_
{L
1
:L
1
=}
_
{L
2
:L
2
=}
dK
2
dK
1
=
_
{L
1
:L
1
=}
E(n) dK
1
_
{L
1
:L
1
=}
dK
1
=
2
_
{L
1
:L
1
=}

1
_
L(p
1
,
1
)
_
dK
1
L()
_
{L
1
:L
1
=}
dK
1
=
2 A()
L()
2
.
28. Bertrand Paradox.
What is the average length of a chord of a compact convex set ?
There are many answers. Depends on what random line means.
When is disk of radius R,
1 Uniform distance from origin and uniform angle (proportional to dK)
E(
1
) =
_
{L:L=}

1
dK
_
{L:L=}
dK
=
A()
L()
=
R
2
2 Uniform point on boundary and uniform angle
E
2
(
1
) =
1
L()
_
L()
0
_

0

1
d ds =
4R

3 Two uniform random points on the boundary

E
3
(
1
) =
1
L()
2
_
L()
0
_
L()
0

1
ds
1
ds
2
=
4R

28. Bertrand Paradox.

What is the average length of a chord of a compact convex set ?
There are many answers. Depends on what random line means.
When is disk of radius R,
1 Uniform distance from origin and uniform angle (proportional to dK)
E(
1
) =
_
{L:L=}

1
dK
_
{L:L=}
dK
=
A()
L()
=
R
2
2 Uniform point on boundary and uniform angle
E
2
(
1
) =
1
L()
_
L()
0
_

0

1
d ds =
4R

3 Two uniform random points on the boundary

E
3
(
1
) =
1
L()
2
_
L()
0
_
L()
0

1
ds
1
ds
2
=
4R

28. Bertrand Paradox.

What is the average length of a chord of a compact convex set ?
There are many answers. Depends on what random line means.
When is disk of radius R,
1 Uniform distance from origin and uniform angle (proportional to dK)
E(
1
) =
_
{L:L=}

1
dK
_
{L:L=}
dK
=
A()
L()
=
R
2
2 Uniform point on boundary and uniform angle
E
2
(
1
) =
1
L()
_
L()
0
_

0

1
d ds =
4R

3 Two uniform random points on the boundary

E
3
(
1
) =
1
L()
2
_
L()
0
_
L()
0

1
ds
1
ds
2
=
4R

28. Bertrand Paradox.

What is the average length of a chord of a compact convex set ?
There are many answers. Depends on what random line means.
When is disk of radius R,
1 Uniform distance from origin and uniform angle (proportional to dK)
E(
1
) =
_
{L:L=}

1
dK
_
{L:L=}
dK
=
A()
L()
=
R
2
2 Uniform point on boundary and uniform angle
E
2
(
1
) =
1
L()
_
L()
0
_

0

1
d ds
=
4R

3 Two uniform random points on the boundary

E
3
(
1
) =
1
L()
2
_
L()
0
_
L()
0

1
ds
1
ds
2
=
4R

28. Bertrand Paradox.

What is the average length of a chord of a compact convex set ?
There are many answers. Depends on what random line means.
When is disk of radius R,
1 Uniform distance from origin and uniform angle (proportional to dK)
E(
1
) =
_
{L:L=}

1
dK
_
{L:L=}
dK
=
A()
L()
=
R
2
2 Uniform point on boundary and uniform angle
E
2
(
1
) =
1
L()
_
L()
0
_

0

1
d ds
=
4R

3 Two uniform random points on the boundary

E
3
(
1
) =
1
L()
2
_
L()
0
_
L()
0

1
ds
1
ds
2
=
4R

28. Bertrand Paradox.

What is the average length of a chord of a compact convex set ?
There are many answers. Depends on what random line means.
When is disk of radius R,
1 Uniform distance from origin and uniform angle (proportional to dK)
E(
1
) =
_
{L:L=}

1
dK
_
{L:L=}
dK
=
A()
L()
=
R
2
2 Uniform point on boundary and uniform angle
E
2
(
1
) =
1
L()
_
L()
0
_

0

1
d ds =
4R

3 Two uniform random points on the boundary

E
3
(
1
) =
1
L()
2
_
L()
0
_
L()
0

1
ds
1
ds
2
=
4R

29. Kinematic density for a moving curve.

Let C and be two piecewise C
1
curves in
the plane. Using rigid motion, we move
around the plane

= M
a,b,
().
M
a,b,
is rotation by angle followed by
translation by vector (a, b)
x

= x cos b sin + a
y

= x sin + y cos + b
The Kinematic Density is the invariant
measure on motions of

given by
dK = da db d.
29. Kinematic density for a moving curve.
Let C and be two piecewise C
1
curves in
the plane. Using rigid motion, we move
around the plane

= M
a,b,
().
M
a,b,
is rotation by angle followed by
translation by vector (a, b)
x

= x cos b sin + a
y

= x sin + y cos + b
The Kinematic Density is the invariant
measure on motions of

given by
dK = da db d.
29. Kinematic density for a moving curve.
Let C and be two piecewise C
1
curves in
the plane. Using rigid motion, we move
around the plane

= M
a,b,
().
M
a,b,
is rotation by angle followed by
translation by vector (a, b)
x

= x cos b sin + a
y

= x sin + y cos + b
The Kinematic Density is the invariant
measure on motions of

given by
dK = da db d.
30. Poincares Formula.
Theorem (Poincares Formula for intersecting curves [1912])
Let C and be piecewise C
1
curves in the plane. Let n(C

) denote
the number of intersection points between C and a moving

. Then
_
{

:C

=}
n(C

) dK(

) = 4 L(C) L().
We show the formula for C
1
curves and add to get it for piecewise C
1
curves. We give two computations of the integral over the ag subset
S = {(

, X) : C

= , X C

}.
For simplicity, suppose the origin 0 C and 0 .
31. Coordinates for the moving curve.
Figure: Attach a unit frame to the moving curve.
Let I be the integral over S the rst way.
I =
_
{

:C

=}
n dK =
_
{

:C

=}
_

ZC

1
_
da db d (7)
For the second equivalent way, we pick a point Z common to both curves
rst and then the angle between the tangents of C and

.
32. Finish the proof of Poincares Formula.
Figure: Angle between C and

at Z.
Let s be arclength along C from the origin and t arclength along from
the origin corresponding to the common point Z C

. Let (s)
denote the tangent angle at (x(s), y(s)) C and (t) the tangent angle
at (u(t), v(t)) . The coordinates (x, y) of Z are given in two ways
x(s) = a + u(t) cos v(t) sin
y(s) = b + u(t) sin + v(t) cos
= +(t) (s)
33. Finish the proof of Poincares Formula-.
Change to (s, t, ) coordinates for S. Dierentiating the dening
equations,
x(s) ds = da +
_
u(t) cos v(t) sin

dt
_
u(t) sin + v(t) cos

d
y(s) ds = db +
_
u(t) sin + v(t) cos

dt +
_
u(t) cos v(t) sin

d
d = d +

(t) dt (s) ds
Using (cos , sin ) = ( x, y) and (cos , sin ) = ( u, v), the kinematic
density is thus da db d
=
_
x(s) ds
_
u(t) cos v(t) sin

dt +
_
u(t) sin + v(t) cos

d
_

_
y(s) ds
_
u(t) sin + v(t) cos

dt
_
u(t) cos v(t) sin

d
_

_
d

(t) dt + (s) ds
_
=
_
x
_
u sin + v cos

+ y
_
u cos v sin
_
ds dt d
= sin() ds dt d.
34. Finish the proof of Poincares Formula - -.
Using Fubinis theorem, we nd another expression for (7)
I =
_
C
_

2
_
0
da db d =
_
C
_

2
_
0
| sin()| d dt ds = 4 L(C) L().
35. Santalos Theorem for convex domains.
Figure: Luis Santalo 1911-2001.
Figure: Convex domains have convex
intersection.
Theorem (Santalos Formula for convex domains [1935])
Let
1
and
2
be convex plane domains. We assume that

2
is moving
in the plane with kinematic density dK
2
. Then
_
{

2
:

1
=}
dK
2
= 2
_
A(
1
) + A(
2
)
_
+ L(
1
) L(
2
). (8)
36. Proof of Santalos Theorem.
Figure: Extent D of moving center so
domains overlap.
h() is the support function for
1
;
g() is the support function for
2
.
We approximate by convex sets
1
and
2
with piecewise C
2
boundaries. The second domain

2
= M
2
is moved by a rotation
of angle followed by translation of
vector (a, b). The kinematic density
is dK = da db d.
Fix and consider D(), the
set of moving centers (a, b) of

2
() such that the domains
overlap:
1

2
() = .
f () = h() + g( + )
is the support function for D();
37. Proof of Santalos Theorem -.
Use (6) to integrate the area of the moving centers D().
J =
_
{

2
:
1

2
=}
dK
=
2
_
0
_
{

2
():
1

2
()=}
da db d
=
1
2
2
_
0
2
_
0
f ()
_
f () +

f ()
_
dd
=
1
2
2
_
0
2
_
0
[h() + g( + )]
_
h() + g( + )
+

h() + g( + )
_
dd
38. Proof of Santalos Theorem - -.
Using Fubinis theorem, Cauchys Formula (5) and
_
2
0

h() d = 0,
2J =
_
2
0
_
2
0
h()
_
h() +

h()
_
dd
+
_
2
0
_
2
0
g( + ) [g( + ) + g( + )] dd
+
_
2
0
_
2
0
h() [g( + ) + g( + )] dd
+
_
2
0
_
2
0
g( + )
_
h() +

h()
_
dd
= 4 A(
1
) + 4 A(
2
)
+
_
2
0
h()
_
L(
2
) + 0
_
d +
_
2
0
L(
2
)
_
h() +

h()
_
d
= 4 A(
1
) + 4 A(
2
) + L(
1
) L(
2
) + L(
2
)
_
L(
1
) + 0
_
.
39. Geometric Probability application of Poincares and Santalos Formul.
Corollary
Let
1
and
2
be bounded convex planar domains. The expected number
of intersections of
1
with a moving

2
given that

2
meets
1
is
E(n) =
4 L(
1
) L(
2
)
2
_
A(
1
) + A(
2
)
_
+ L(
1
) L(
2
)
.
If
1

=
2
are congruent, then E(n) 2 with = i
1
is a circle.
Proof. Apply Poincares and Santalos Formulas to the expectation
E(n) =
_
{

2
:
1

2
=}
n(

2
) dK
_
{

2
:
1

2
=}
dK
2
.
If
1

=
2
are congruent, the isoperimetric inequality implies
E(n) =
4 L
2
4 A+L
2

4 L
2
L
2
+L
2
= 2 with equality i
1
is circle.
40. Total curvature.
Let C be closed piecewise C
2
curve.
The curvature is =

s
,
the rate of turning, where gives
the angle via (cos , sin ) =

Z, the
direction of C at Z.
Figure: Piecewise C
2
boundary with
corners at Z
i
A piecewise C
2
boundary is the
union of n curves =
n
_
i =1
C
i
.
The total curvature is the integral of
the curvatures over the C
2
curves C
i
plus the turning angle at the vertices
Z
i
between C
i
and C
i +1
c() =
n

i =1
_
C
i
ds +
n

i =1

i
By the Gauss-Bonnet Formula, the
total curvature of a boundary is
related to the Euler Characteristic
c() = 2().
41. Blaschkes Theorem for general domains.
Figure: Wilhelm Blaschke 18851962
Theorem (Blashkes Fundamental Formula [1936])
Let
1
and
2
be plane domains bounded by nitely many oriented,
piecewise C
2
, simple, closed curves. We assume that

2
is moving in the
plane with kinematic density dK
2
. Then
_
{

2
:

1
=}
c(
1

2
) dK
2
= 2
_
A(
1
) c(
2
) + A(
2
) c(
1
)
+L(
1
) L(
2
)
_
.
42. Special Cases.
Figure: Simple boundaries: count
components of intersection.
Figure: Convex domains have
convex intersection.
Case 1. Both domains bounded by one
simple curve. Then c(
i
) = 2. Let
(
1

2
) be number of components.
_
{

2
:

1
=}
(
1

2
) dK
2
= 2
_
A(
1
) + A(
2
)
_
+ L(
1
) L(
2
).
Case 2. Both domains convex. Then
(
1

2
) = 1. We recover (8):
_
{

2
:

1
=}
dK
2
= 2
_
A(
1
) + A(
2
)
_
+ L(
1
) L(
2
).
43. Isoperimetric Inequality - - An Integral Geometric Proof
Among all domains in the plane with a xed boundary length, the circle
has the greatest area. For simplicity we focus on domains bounded by
simple curves.
Theorem (Isoperimetric Inequality.)
1 Let C be a simple closed curve in the plane whose length is L and
that encloses an area A. Then the following inequality holds
4A L
2
. (9)
2 If equality holds in (9), then the curve C is a circle.
Simple means curve is assumed to have no self intersections.
A circle of radius r has L = 2r and encloses A = r
2
=
L
2
4
.
Thus the isoperimetric Inequality says if C is a simple closed curve of
length L and encloses an area A, then C encloses an area no bigger than
the area of the circle with the same length.
44. Convex Hull
A set K E
2
is convex if for every pair of points x, y K, the straight
line segment xy from x to y is also in K, i.e., xy K.
The bounding curve of a convex set is automatically rectiable. The
convex hull of K, denoted

K, is the smallest convex set that contains K.
This is equivalent to the intersection of all halfspaces that contain K,

K =

is convex
K
=

H is a halfspace
H K
H.
A halfspace is a set of the form H = {(x, y) E
2
: ax + by c}, where
(a, b) is a unit vector and c is any real number.
45. Reduce proof of Isoperimetric Inequality to convex domain case.
Since K

K by its denition, we have A(

K) A(K).
Taking convex hull reduces the boundary length because the interior
segments of the boundary curve, the components of C

K of C are
replaced by straight line segments in

K. Thus also L(

K) L(K).
Figure: The region K and its convex hull

K.
46. Reduce proof of Isoperimetric Inequality to convex curves case.-
Thus the isoperimetric inequality for convex sets implies
4A 4

A

L
2
L
2
.
Furthermore, one may also argue that equality 4A = L
2
implies equality
4

A =

L
2
in the isoperimetric inequality for convex sets so that

K is a
circle. But then so is K.
The basic idea is to consider the the extreme points

K of

K,
that is points x

K such that if x = y + (1 )z for some y, z

K
and 0 < < 1 then y = z = x.

K is the convex hull of its extreme
points. However, the extreme points of the convex hull lie in the curve

K C

K.

K being a circle implies that every boundary point is an
extreme point, and since they come from C, it means that C is a circle.
47. Isoperimetric Inequality for convex sets
There are many proofs of the isoperimetric inequality. We shall give two
integral geometric arguments due to Luis Santalo.
1 The rst argument only establishes the inequality part 4A L
2
.
2 To show that the circle is the unique domain for which the
Isoperimetric Inequality is equality, we prove a strong isoperimetric
inequality (12) that follows from Bonnesens inequality (11). The
second argument is Santalos proof of Bonnesens inequality.
48. Santalos proof of the Isoperimetric Inequality for convex sets.
Lemma (Isoperimetric Inequality for convex sets.)
If is a convex plane domain with boundary length L and area A, then
4A L
2
. (10)
Proof. Let
1
and
2
be congruent copies of . Let m
i
denote the
measure of positions of a moving

2
for which the number of
intersections
n(
1

2
) = i .
Note that positions that have an odd or innite number of intersection
points is dK-measure zero so that
m
i
= 0 if i is odd.
49. Finish Santalos proof of the Isoperimetric Inequality.
Then by Poincares and Santalos formulas,
4 L()
2
=
_
{

2
:

1
=}
n(
1

2
) dK = 2m
2
+ 4m
4
+ 6m
6
+ ,
4 A() + L()
2
=
_
{

2
:

1
=}
dK = m
2
+ m
4
+ m
6
+ .
Subtracting,
L()
2
4 A() = m
4
+ 2m
6
+ 3m
8
+ 0,
since all the measures m
i
0.
50. Inradius / Circumradius
Let K be the region bounded by . The radius of the smallest circular
disk containing K is called the circumradius, denoted R
out
. The radius of
the largest circular disk contained in K is the inradius.
R
in
= sup{r : there is p E
2
such that B
r
(p) K}
R
out
= inf{r : there exists p E
2
such that K B
r
(p)}
Figure: The disks realizing the circumradius, R
out
, and inradius, R
in
, of K.
51. Bonnesens Inequality
Figure: T. Bonnesen 18731935
Theorem (Bonnesens Inequality [1921])
Let be a convex plane domain whose
boundary has length L and whose area is A.
Let R
in
and R
out
denote the inradius and
circumradius of the region . Then
sL A +s
2
for all R
in
s R
out
. (11)
Bonnesens strong isoperimetric inequality
follows immediately.
Corollary (Strong Isoperimetric Inequality of Bonnesen)
Let be a convex planar domain with boundary length L and area A.
Let R
in
and R
out
denote the inradius and circumradius of the . Then
L
2
4A
2
(R
out
R
in
)
2
. (12)
52. Bonnesens Inequality implies the Strong Isoperimetric Inequality
Proof of corollary. Consider the quadratic function f (s) = s
2
Ls + A.
By Bonnesens inequality, f (s) 0 for all R
in
s R
out
. Hence these
numbers are located between the zeros of f (s), namely
R
out

L +

L
2
4A
2
L

L
2
4A
2
R
in
.
Subtracting these inequalities gives
R
out
R
in

L
2
4A

,
which is (12).
53. Strong Isoperimetric Inequality implies the Isoperimetric Inequality
Obvious. The strong isoperimetric inequality (12) implies part one of the
isoperimetric inequality (10), since
2
(R
out
R
in
)
2
0.
Moreover, if equality holds in (9), then L
2
4A = 0 which implies that
R
in
= R
out
, or is a circle.
54. Santalos proof of Bonnesens inequality
Theorem (Bonnesens Inequality)
Let be a bounded convex plane domain whose boundary has length L
and whose area is A. Let R
in
and R
out
be the inradius and circumradius
of the region . Then sL A +s
2
for all R
in
s R
out
.
Proof. Let
1
= and

2
be a moving circular disk of radius s.
Because R
in
s R
out
, the sets overlap,
1

2
= , if and only if
their boundaries overlap,
1

2
= , hence the Poincare and
Blaschke integrals are taken over the same positions of

2
.
As before, let m
i
denote the measure of positions of the moving

2
for
which the number of intersections n(
1

2
) = i , i.e.,
m
i
= dK
__

2
: n(
1

2
) = i
__
.
Again, positions that have an odd or innite number of intersection
points is dK-measure zero so that m
i
= 0 if i is odd.
55. Finish Santalos proof of Bonnesens Inequality.
Then by Poincares and Santalos formulas,
8s L() =
_
{

2
:

1
=}
n(
1

2
) dK = 2m
2
+ 4m
4
+ 6m
6
+ ,
2 A() + 2
2
s
2
+ 2s L() =
_
{

2
:

1
=}
dK = m
2
+ m
4
+ m
6
+ .
Subtracting,
2
_
s L() A() s
2
_
= m
4
+ 2m
6
+ 3m
8
+ 0,
since all the measures m
i
0.
Thanks!