university of washington seattle, wa 98195-4350 usa [email protected] branko grünbaum...

95
University of Washington Seattle, WA 98195-4350 USA [email protected] on.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE, AND THE KLEIN BOTTLE

Upload: olivia-ramsey

Post on 20-Jan-2016

219 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

University of WashingtonSeattle, WA [email protected]

Branko Grünbaum

SMALL POLYHEDRAL MODELS

OF THE TORUS, THE PROJECTIVE

PLANE, AND THE KLEIN BOTTLE

Page 2: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

For which integers n are there polyhedra that are topologically

equivalent to a torus, with all n faces triangles ? Or all faces

quadrangles, or pentagons, ... ?

Page 3: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

For which integers n are there polyhedra that are topologically

equivalent to a torus, with all n faces triangles ? Or all faces

quadrangles, or pentagons, ... ?

We need to stop at hexagons, since no torus can have all faces

with seven or more sides.

Page 4: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

For which integers n are there polyhedra that are topologically

equivalent to a torus, with all n faces triangles ? Or all faces

quadrangles, or pentagons, ... ?

We need to stop at hexagons, since no torus can have all faces

with seven or more sides.

Joint work with Lajos Szilassi

Page 5: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The question is: For which n can the torus be geometrically

(i) triangulated

(ii) quadrangulated

(iii) quintangulated

(iv) hexangulated

with n faces ?

The torus should be acoptic, that is, with simple faces and

without selfintersections.

Page 6: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

TRIANGULATIONS

It is well known that combinatorial (topological) triangulations of the torus with n faces exist for all even n with n ≥ 2, and only for such n.Geometric triangulations exist if and only if n is even and n ≥ 14.

n = 14: Császár torus

Page 7: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

For triangulations with n ≥ 16:

Start with a triangulation with n–2 faces and attach a tetrahedron

on one of the faces for a net increase of two faces.

Page 8: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

For triangulations with n ≥ 16:

Start with a triangulation with n–2 faces and attach a tetrahedron

on one of the faces for a net increase of two faces.

There exist isogonal triangulations of the torus with n faces

if and only if n = 4m, m ≥ 5.

Page 9: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

OVERARCHING FACES:

The intersection of two faces has more than one connectedcomponent.

Page 10: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A far-reaching generalization of these results is the

following theorem of Archdeacon et al.:

Every topological triangulation of the torus with no

overarching faces is isomorphic to a geometric

triangulation.

Page 11: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

QUADRANGULATIONS

Quadrangulations with n non-overarching faces exist for

all n ≥ 9, except possibly for n = 10, 11.

Page 12: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

QUADRANGULATIONS

Quadrangulations with n non-overarching faces exist for

all n ≥ 9, except possibly for n = 10, 11.

Two basic constructions

First construction:

For all integers p ≥ 3 and q ≥ 3 we can construct "picture

frames" for p-sided "pictures", with q-sided cross-sections.

These give quadrangulations with n = p q, thus yielding the

values n = 9, 12, 15, 16, 18, 20, ... .

Page 13: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A picture frame for a triangular picture with pentagonal cross section, and a frame for a pentagonal picture with a triangular cross section.

Page 14: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Second construction:To a face of a given quadrangulation attach a suitable image of a cube. This increases the number of faces by 4.

From n = 9 we get n = 13 and then n = 17.Then the consecutive values n = 15, 16, 17, 18 are available, hence adding multiples of 4 yields all n ≥ 15.

Page 15: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The only still missing value is n = 14.

An example is shown here:

Page 16: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

If overarching faces are admitted, then quadrangulations are

possible for all n > 9.

Example with n = 10.

Page 17: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

An example with n = 11 faces.

Conjecture. There exist no geometric acoptic quadrangulations with n ≤ 8 quadrangles.

Page 18: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A far-reaching conjecture is:

Every topological quadrangulation with no overarching

faces can be realized geometrically.

Only a few examples of such topological

quadrangulations are known for which it has been

proved that they cannot be realized by polyhedra with

convex faces. They all seem to be realizable with simple

faces that are not necessarily convex.

Page 19: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

QUINTANGULATIONS

For every even n ≥ 12 there are convex-faced

quintangulations of the torus, except possibly for n = 14.

Basic construction:

Page 20: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Simple variants of this construction yield

quintangulations with p-fold rotational symmetry

and with n = 2 p q faces, for all p ≥ 3, q ≥ 2. In

particular, n = 12, 16, 18, 20, 24, 28, ... are obtained.

Attaching a copy of a dodecahedron yields an

increase of 10 faces, thus establishing the claim.

In the previous slide p = 7, q = 2.

Page 21: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

CONJECTURE:

There are no quintangulations with n convex faces

where n = 14 or n ≤ 10.

It is not clear what happens if one does not insist that

the faces are convex.

Page 22: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

HEXANGULATIONS

Acoptic hexangulations with n faces, with no overarching faces, exist for n = 7, and for all n = p q with p ≥ 3, q ≥ 3.

The case n = 7 is the well-known Szilassi polyhedron

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

1

2

5

3

4

14

1213

11 10

9

78

6

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

1

2

5

3

4

14

1213

11 10

9

78

6

Page 23: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Hexangulations with n = p q for p ≥ 3, q ≥ 3 can be constructedby starting with trapezohedra (Catalan polyhedra, that are polar tothe antiprisms). For each p ≥ 3 such a polyhedron P has 2p quadrangular faces, p of which meet at each of two apices of P. By intersecting such a polyhedron P with a p-sided prism, having its axis coinciding with the axis of P and rotated appropriately, the resulting tunnel has p hexagonal sides, and all 2p sides of P become hexagons as well.

Example: p = 3, q = 3 so n = 9

Page 24: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

An example with p = 5, q = 3, so n = 15.

Page 25: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

An example with p = 3, q = 5, so again n = 15.

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Page 26: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Conjecture. The only acoptic hexangulations with n faces and no overarching that have rotational symmetry of order3 or more are those with n = p q, with p, q ≥ 3.

There are hexangulations lacking such symmetry for some other values of n. These values have not been characterized.

If overarching faces are allowed,some other symmetric possibilitiesarise. Here is a hexangulation with8 hexagons. It was found by J. Schwörbel.

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Page 27: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

In contrast to the situation concerning quintangulations, it is well known that no hexangulation of the torus can have only convex faces.

Conjecture. Every hexangulation of the torus has at least six non-convex faces.

Conjecture. Polyhedra of Kepler-Poinsot type admit hexangulations with n faces for all n ≥ 7.

Page 28: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

QuickTimeª and aTIFF (Uncompressed) decompressor

are needed to see this picture.

Example. A hexangulation with n = 11 faces. It has two pairs of intersecting faces, and one selfintersecting face.

Page 29: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

One may wonder how far all these constructions can be generalized. Long ago I proposed the following

Conjecture. Every cell-complex decomposition, without overarching elements, of an orientable 2-manifold is realizable by an acoptic polyhedron.

The theorem of Archdeacon et al. confirms this in a very special case, while a result of Bokowski et al. shows that it fails for triangulations of a 6-manifold.

As far as I know, there is no information in the case overarching faces are admitted.

No characterization of spherical acoptic polyhedra is known.

Page 30: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

In the second part we shall be concerned with

polyhedral realizations of non-orientable manifolds.

By a basic topological result such polyhedra cannot

be acoptic.

Page 31: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

In this part we shall look mainly for polyhedra that

realize or represent the projective plane, the Klein

bottle, and the Möbius band. These are the simplest

non-orientable 2-manifolds.

This will lead to considerations of non-convex

polyhedra, polyhedra with selfintersections, and most

significantly, polyhedra with selfintersecting faces.

Page 32: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

During most of the Twentieth Century polyhedra with

selfintersections have been neglected, except for some

very special classes. In particular, their topological

properties have not been investigated. This is part of

our goal here.

Another part is to simply show some of these

polyhedra, in order to illustrate their mathematical and

esthetic aspects.

Page 33: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Selfintersecting polygons are of primary importance as faces of the polyhedra we consider. Here are some examples of suchpolygons, and the names I will use.

Bow-tie (quadrangle) Hexagram

Pentagram

In all cases we need to distinguish between the polygonal lineand the polygon as a 2-dimensional part of the plane.

Page 34: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

USUAL WAYS OF PRESENTING THE REAL PROJECTIVE

PLANE P2

Page 35: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

USUAL WAYS OF PRESENTING THE REAL PROJECTIVE

PLANE P2

Euclidean plane E2 together with points and line at infinity

Family of lines and planes through the origin of E3

Points and great circles of a sphere, with antipodal points identified

Circular disk in E2, with diametral points identified

Page 36: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Convex polyhedra with center, in which opposite elements (vertices, edges, faces) are identified. These are often calledhemi-polyhedra.

POLYHEDRAL MODELS OF THE PROJECTIVE PLANE

Regular dodecahedron withopposite elements identified

Page 37: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Convex polyhedra with center, in which opposite elements (vertices, edges, faces) are identified. These are often calledhemi-polyhedra.

POLYHEDRAL MODELS OF THE PROJECTIVE PLANE

Planar polyhedral maps, with boundary identification, are analogues of the circular disk model in E2, with diametral points identified.

1

2

3 4

5

1

2

34

5

Regular dodecahedron withopposite elements identified

The correspondingpolyhedral map

Page 38: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

CAN A POLYHEDRAL MAP OF THE PROJECTIVE PLANEP2 BE REALIZED BY AN ACTUAL POLYHEDRON ?

Page 39: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

This cannot be an acoptic (selfintersection-free) polyhedron, since

acoptic polyhedra are orientable, while the projective plane is

non-orientable.

Allowing intersections of faces, can all faces be simple polygons?

CAN A POLYHEDRAL MAP OF THE PROJECTIVE PLANEP2 BE REALIZED BY AN ACTUAL POLYHEDRON ?

Page 40: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

This cannot be an acoptic (selfintersection-free) polyhedron, since

acoptic polyhedra are orientable, while the projective plane is

non-orientable.

Allowing intersections of faces, can all faces be simple polygons?

CAN A POLYHEDRAL MAP OF THE PROJECTIVE PLANEP2 BE REALIZED BY AN ACTUAL POLYHEDRON ?

YES !112233

(hemi-cuboctahedron)

Page 41: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The projective plane P2 is a 2-manifold, as is the polyhedral map

with boundary identification. However, the heptahedron (or

hemi-cuboctahedron) is not a manifold, since each neighborhood

of each of its six vertices has an essential selfintersection.

Vertex figure is a bow-tie quadrangle

Such a singularpoint is known as aWhitney umbrella

Page 42: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

If the heptahedron were a 2-manifold, which manifold would it be ?

Every closed 2-manifold M is characterized by its orientability (orientable or non-orientable) and by its Euler characteristic (M) = V – E + F.

(M) = 2 ==> M is the orientable 2-sphere.

(M) = 1 ==> M is the non-orientable projective plane.

(M) = 0 ==> M is either the orientable torus, or else the non-orientable Klein bottle.

Page 43: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

If the heptahedron were a 2-manifold, which manifold would it be ?

Every closed 2-manifold M is characterized by its orientability (orientable or non-orientable) and by its Euler characteristic (M) = V – E + F.

Since (H) = 1, if H denotes the heptahedron, the only manifoldit could be is the projective plane – but the Whitney umbrella isan impediment.

(M) = 2 ==> M is the orientable 2-sphere.

(M) = 1 ==> M is the non-orientable projective plane.

(M) = 0 ==> M is either the orientable torus, or else the non-orientable Klein bottle.

Page 44: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

There are many possibilities for realizations of hemi-polyhedra(that is, the projective plane) and not all have Whitney umbrellas.

Hemi-dodecahedron

Reinhardt 1885

AMHFJDGKEBCFaces:ADGKBABEHCACFJDJEBKFGKFCHEHGDJ

M is the point at whichthree faces haveselfintersections; it isnot a vertex of the hexahedron

ADGKBEHCFJEHCFJ

Page 45: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Hemi-dodeca-hedron(L. Szilassi, 2007)

Conjecture (Szilassi): Every realization of the hemi-dodecahedron has at least one selfintersecting face.

1

2

34

510

987

6

10

9

8

7

6

Face [1,2,3,4,5] isselfintersecting,faces [1,8,9,10,2],[3,2,10,6,7],[4,5,6,10,9],[1,5,6,7,8] and[3,7,8,9,4] aresimple polygons;the last two intersect.

Page 46: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Problem: The classification of 2-manifolds implies that every

manifold is homeomorphic to one of the standard representatives.

For (M) ≥ 0 this means that M is homeomorphic to one of the

four we have listed three slides back. But just as a Whitney

umbrella cannot be homeomorphic to a disk, neither can a

selfintersecting polygon.

?

Page 47: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The problem has nothing to do with non-orientability, or the

projective plane –– it is inherent in selfintersection of faces.

The prism with bow-tie basis is isomorphic to the cube,

but is it homeomorphic ?

1 2

3

4

65

7

8

1

2

3

4

6

5 7

8

Page 48: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The regular dodecahedron and the great stellated dodecahedron are

isomorphic, as shown by the labels. But are they homeomorphic?

That is, is there is a 1-to-1 map that is continuous and has a continuous

inverse.13

52

34

2145

24

35

41

12 54

23

1551

43

25

31

53

32

14

42

12

45

23

5134

25

13

42

41

53

14

35

24

31

52

54

32

15

21

43

Page 49: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Dilemma: Do we stop thinking of polyhedra with selfintersecting

faces as manifolds (as topology does, for at least the last

century), or do we find a way to interpret selfintersecting

polygons as homeomorphic to simple polygons ?

The regular dodecahedron and the great stellated dodecahedron are

isomorphic, as shown by the labels. But are they homeomorphic?

That is, is there is a 1-to-1 map that is continuous and has a continuous

inverse.13

52

34

2145

24

35

41

12 54

23

1551

43

25

31

53

32

14

42

12

45

23

5134

25

13

42

41

53

14

35

24

31

52

54

32

15

21

43

Page 50: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A similar problem arose, and was solved, long ago in connectionwith curves and polygonal lines. It is obvious that as a set of points, the bow-tie quadrangular line is not homeomorphicto the boundary of the square (or to a circle).

12341234x

Page 51: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

To establish a homeomorphism, the concept of immersion was accepted as the solution: A crossing-point (such as x) is considered

as representing two distinct points, one on each branch of the

curve or polygonal line. This makes possible a parametrization of

the selfintersecting curve or polygonal line.

Page 52: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

To establish a homeomorphism, the concept of immersion was accepted as the solution: A crossing-point (such as x) is considered

as representing two distinct points, one on each branch of the

curve or polygonal line. This makes possible a parametrization of

the selfintersecting curve or polygonal line.

With this understanding, and obvious generalizations, selfintersection points of polygonal lines (or curves) do not

prevent homeomorphisms of the parametrized line with simple

polygonal lines, or with a circle.

Page 53: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

To establish a homeomorphism, the concept of immersion was accepted as the solution: A crossing-point (such as x) is considered

as representing two distinct points, one on each branch of the

curve or polygonal line. This makes possible a parametrization of

the selfintersecting curve or polygonal line.

With this understanding, and obvious generalizations, selfintersection points of polygonal lines (or curves) do not

prevent homeomorphisms of the parametrized line with simple

polygonal lines, or with a circle.

Note: we are dealing with a homeomorphism of the parametrized

lines or curves –– not with a homeomorphism of the plane.

Page 54: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The concept of immersion and the appropriate parametrization was long ago extended to cover segments of multiple points, aswell as whole regions of such points. An example was providedby Reinhardt, in comments to the 1886 edition of Möbius’ collected works. His drawing of the Möbius band (as proposedby Möbius) is shown at left, while a version of the band, and ofthe hemi-icosahedron obtained by constructing a pyramid overthe band, are shown in the other parts. These are taken from abook by Apéry (1987).

Page 55: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

We need to validate the use of the Euler characteristic in deciding which manifold is represented by a polyhedron that may contain selfintersecting faces. By relying on the immersion andparametrization concepts, we know that a simple polygonal line ishomeomorphic with the polygonal line that determines a face, evenif the polygonal line is selfintersecting. Now we need to extend thishomeomorphism in a way that the interior of the simple polygonmaps onto a kind of “interior” of the selfintersecting polygon.Where is the quadrangular “interior” of the bow-tie that ishomeomorphic to the interior of the square?

12341234x

Page 56: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

We need a different kind of immersion. To avoid misunder-

standings, we call it infolding. In the infolding of the bow-tie

quadrangle, or of the selfintersecting pentagon in the Szilassi

pentagon, we need each crossing point to represent a segment of

points. How such a parametrization can be done in a consistent and

bicontinuous way is illustrated in the next slide. It shows an isotopy

(continuous deformation) from a parametrized bow-tie to a square.

The point parameters throughout the isotopy are (x, c), the isotopy

parameter is t, and the points have coordinates (x, y). All

parameters go from –1 to +1. The coordinate y is given by

y = ((t – 1)x + t + 1)c/2

Page 57: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

(1,–t)

(1,t)

(-1,1)

(-1,-1)

(-1,c)

(1,t c)

t = –1

(1,–t)

(1,t)

(-1,1)

(-1,-1)

(-1,c)

(1,t c)

t = –0.5

(1,t)

(-1,1)

(-1,-1)

(-1,c)

(1,t c)

t = 0

(1,t)

(1,–t)

(-1,1)

(-1,-1)

(-1,c)

(1,t c)

t = 0.5

(-1,1) (1,1)

(-1,-1) (1,–1)

(-1,c) (1,c)

t = 1

y = ((t – 1)x + t + 1)c/2

Page 58: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

For a different variety of infolding, consider the regular

dodecahedron, and the great stellated dodecahedron.

They are isomorphic, so the great stellated dodecahedron

should be homeomorphic to a sphere –– even though

it is well known that the density at its center is 3.

13

52

34

2145

24

35

41

12 54

23

1551

43

25

31

53

32

14

42

12

45

23

5134

25

13

42

41

53

14

35

24

31

52

54

32

15

21

43

Page 59: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Here an isotopy connects a pentagram to a bow-tie quadrangle, which is isotopic to a square, which is isotopic to a pentagon. Thus the pentagram is homeomorphic to the pentagon, hence the great stellated dodecahedron is indeed homeomorphic to the Platonic dodecahedron.

Page 60: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

It can be shown that the infolding approach generalizes to all

polyhedra, and that therefore the criteria of orientability and

Euler characteristic are sufficient to establish each polyhedron

(in suitably parametrized form) as homeomorphic to a (compact)

2-manifold. All it takes is to interpret each selfintersection point as

representing a whole segment in a suitable parametrization.

A simple way to formulate this conclusion is:

For suitably parametrized polyhedra

ISOMORPHIC => HOMEOMORPHICHere are a few examples of the application of this result.

Page 61: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Examples of connels (= conical tunnels) formed by bow-ties.Like tunnels, they can be used to replace pairs of faces that are related by symmetry in a point. If the pair was part of an orientable polyhedron, the resulting polyhedron is orientable as well. Each replacement reduces the Euler characteristic by 2.A simple example is shown in the next slide.

Page 62: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Torus: 4 squares, one connel with 4 bow-tie faces, 16 edges, 8 vertices, = 0, orientable

Faces: ABFE, BCGF, CDHG, DAEH, ADFG, DCEF, CBHE, BAGH

A B B C C D D A

E F F G G H H E

A B B C C D D A

E F F GG H H E

A B

C

D

E F

G

H

A B

C

D

E

G

H

F

Page 63: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A selfdual torus, isomorphic to the preceding one: 4 rectangles, 4 bow-tie faces, 16 edges, 8 vertices (there are overarching faces).

Described by K. Merz (1935)

This torus, and the one in the preceding slide, have fewer edges than the minimum number established by U. Brehm (1990) fora class of “polyhedral maps” more restricted than allowed here.

Faces: ABHG, ADHE, CDFE, CBFG, AGFD, AEFB, CEHB, CGHDEAGCHDFBEAGCHDFB

Page 64: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

0 1

2

3

4

0 2 4 1 3 0

1 3 0 2 4 1

A selfdual torus consisting of five bow-tie quadrangles,with five vertices and ten edges. All faces are in one plane.Some pairs of faces are overarching (have intersection with more than one component).

Page 65: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A projective plane consisting of five coplanar convex quadrangles

(only one is shown filled in, for easier visualization). Some pairs

overlap on triangular regions. This is the hemi-5-sided-trapezohedron

(dual of antiprism). The center is a singular point (vertex figure has

rotation number 2).

Analogous projective planes exist for all odd n ≥ 5.

Page 66: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Projective plane

Modified from an example of F. Apéry (1987)

6 vertices, 14 edges, 9 faces (8 triangles, one bow-tie)

AAFFCCDDEBAAFFBBDDEECC

ABDEFC

ABDEFCABDEFC

Page 67: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Projective plane

Taking two copies with coinciding bow-ties, and eliminating

the bow-ties yields an ORIENTABLE polyhedron. This is a

hemi-polyhedron of a torus, which is itself a torus.

6 vertices, 14 edges, 9 faces (8 triangles, one bow-tie)

AAFFCCDDEBAAFFBBDDEECC ABDEFCABDEFC

Page 68: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

6 quadrangles (3 convex, 3 bow-tie), 12 edges, 7 vertices, = 1

Projective plane, hemi-rhombic dodecahedron

This model has fewer edges than the minimum for a more restrictedkind of “polyhedral maps” considered by U. Brehm (1990)

Page 69: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Projective plane: Heptahedron, dual to the hexahedron of the previous slide.3 bow-tie quadrangles, 4 triangles, 12 edges, 6 verticesTwo copies joined at bases give a Klein bottle, with6 bow-tie quadrangles, 6 triangles, 21 edges, 9 vertices

Page 70: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Projective plane:6 quadrangles, 1 hexagon, 15 edges, 9 vertices

Put two copies together, and eliminate hexagons:

Klein bottle with 12 quadrangles, 24 edges, 12 vertices

Page 71: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Faces: ABCD, BAEF, FEHG, GHDC, AEDH, ADEH, BCFG, BFCD

D

G

HE

FF

F

CGGF

C

C C B B

BB

A

A A

AD D

E

E

E

H H D

H

G

Klein bottle: 8 faces, 16 edges, 8 vertices, = 0, non-orientable

ABCDEFGHBACDEFGH

Page 72: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Klein bottle: 12 quadrangles, 20 edges, 8 vertices, non-orientableno overarching faces

1

2

3

45

6

7

8

9

10

11

12

1

2

3

45

6

7

8

9

10

11

12

1

2

3

45

6

7

8

9

10

11

12

Page 73: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Klein bottle: 12 quadrangles 8 quadrangles, 2 hexagons

1

2

3

45

6

7

8

9

10

11

12

1

2

3

45

6

7

8

9

10

11

12

1

2

3

45

6

7

8

9

10

11

12

1

2

3

45

6

7

8

9

10

11

12

Page 74: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Six triangles of the mantle (only one shown at

right), a connel (conical tunnel) of three bow-ties

(two shown at left), 15 edges, six vertices.

Orientable. = 0. Torus with overarching faces.

Page 75: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Two triangles, six bow-ties, 15 edges, six vertices, = –1.

Overarching faces. Non-orientable, of genus 3

Page 76: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Various interesting polyhedra are possible with more

complicated selfintersecting faces. A few examples

are shown next. For simplicity, all are isogonal, that

is, have vertices that are equivalent under symmetries

of the polyhedron.

Page 77: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

(3.6*.3#.6*)

An isogonal faceting of the truncated tetrahedron. This istopologically a torus, in which each vertex is incident with one small triangle, one large triangle and two hexagrams.

Page 78: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

(3.6*.3#.6*)

An isogonal faceting of the truncated tetrahedron. This istopologically a torus, in which each vertex is incident with one small triangle, one large triangle and two hexagrams.

2134561,62,34,5123456123456123456123456{6}1,42,53,6{6/2}

The same type of polyhedron is possible for all hexagrams in theisotopy (metamorphosis) shown.

Page 79: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

2

1

34

5

6 1,6

2,34,5

1

2

34

5

6 1

2

34

5

6 1

2

34

5

61

2

34

5

6{6}

1,4

2,5

3,6

{6/2}

A view of the metamorphosis

Page 80: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

An isogonal faceting of the truncated dodecahedron involving decagrammatical faces. Each vertex meets one triangle, oneregular pentagon, and two decagrams, as shown at left. The resulting polyhedron is shown at right without the pentagons, to make it more intelligible.

(3.10*.5.10*)

Page 81: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

(3.10*.5.10*)Pentagons not shown.

012345678901234567890123456789012345678901234567893,47,81,25,60,90123456789012345678901234567891,60,52,73,84,90123456789012345678901234567890,34,71,82,56,90123456789{10/2}=10/2-110/2-210/2-310/2-510/2-610/2-710/2-810/2-910/2-1010/2-4{10/3}=10/2-1110/2-210/2-210/2-410/2-6

As in the case with hexagons we have seen earlier, there is

a metamorphosis of decagons that connects two regular

decagons through a continuum of isogonal ones. All are

suitable to yield polyhedra of the same type.

Page 82: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Another isogonal faceting of the truncated dodecahedron

involving decagrammatical faces. At each vertex one regular

pentagon meets two decagrams. The resulting polyhedron

is shown without the pentagons, to make it more intelligible.

Orientable, 24 faces, 90 edges, 60 vertices, genus 4.

(5.10*.10*)

Page 83: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

(3.10*.10*.10*.10*)

An isogonal faceting of the truncated dodecahedron in which

each vertex is incident with one triangle and four decagrams.

Orientable, 44 faces, 150 edges, 60 vertices, genus 24.

Page 84: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Isogonal, one triangle and four bow-ties at each vertex.

60 vertices, 150 edges, 80 faces, = –10.

Non-orientable, genus 12.

Page 85: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

ö

The boundary of a Möbius band consist of a single

closed curve. A realization of the Möbius band in

which this curve is planar and bounds a topological

disk is often called a cross-cap. Any cross-cap must

have selfintersections. Here are two familiar versions.

Page 86: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Adapted fromApéry

ö

Page 87: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Each has two Whitney umbrellas. The homeomorphismto the Möbius band needs justification. This is possible,but is rarely given.

Page 88: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Each has two Whitney umbrellas. The homeomorphismto the Möbius band needs justification. This is possible,but is rarely given.

What is the importance of cross-caps?They are needed in the basic theorem on the classification of closed 2-manifolds,

Page 89: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Orientable 2-manifold M

Euler characteristic

(M) = 2 – 2g,

g = (orientable) genus of M

= (2 – )/2

Non-orientable 2-manifold M

Euler characteristic

(M) = 2 – g

g = (non-orientable) genus of M

= 2 -

Page 90: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Orientable 2-manifold M

Euler characteristic

(M) = 2 – 2g,

g = (orientable) genus of M

= (2 – )/2

Non-orientable 2-manifold M

Euler characteristic

(M) = 2 – g

g = (non-orientable) genus of M

= 2 -

Sphere: Orientable, g = 0, = 2.

Each orientable M of genus

g ≥ 0 is homeomorphic

to sphere with g handles.

Each non-orientable M of

genus g ≥ 0 is homeomorphic

to sphere with g cross-caps.

The classification of 2-manifolds:

Page 91: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A

B

C

D

E

F

ö Polyhedral Möbius band

that is a cross-cap

A

B

C

F

E

D

Page 92: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A

B

C

D

E

FA

B

C

D

E

F

A

B

C

F

E

D

A

B

C

F

E

D

A

B

C

D

E

F

A

B

C

F

E

D

=

Transition from the traditional Möbius band tothe polyhedral three-bow-ties cross-cap.

Page 93: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

A variety of cross-caps

Each is homeomorphic to the Möbius band by infolding.

Page 94: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

The three bow-tie quadrangles AA*BB*, BB*CC* and CC*AA*

overlap in the triangular region shown in white. The three bow-ties

form a cross-cap, just as in the case their crossover points coincide.

Another kind of cross-caps

A

B

C

B*

A*

C*

Page 95: University of Washington Seattle, WA 98195-4350 USA grunbaum@math.washington.edu Branko Grünbaum SMALL POLYHEDRAL MODELS OF THE TORUS, THE PROJECTIVE PLANE,

Despite the work of Brückner and others a

century or more ago (or possibly because of it !?),

the study of polyhedra with selfintersections and

with selfintersecting faces has been neglected

during much of the last 100 years. I hope the

examples of such polyhedra and their applications

to the topology of manifolds discussed here will

lead to an awakening of interest in their study.