Back to GEOMETRY GARRET

## POLYHEDRA

Alan H. Schoen

Comments are welcome!

Nested pair of truncated icosahedra, (sculptor unknown)

Photographed by Reiko Takasawa Schoen in Madison Square Park, New York City, October 21, 2012

Cloud City, a sculpture by Tomás Saraceno

Photographed by Reiko Takasawa Schoen at the Metropolitan Museum of Art, New York City, October 20, 2012

A0. Random Polyhedral Honeycomb

This fifteen convex cells in this packing are the Voronoi polyhedra of

fifteen points located at random positions in the central region of a sea

of more than one hundred random points. The computer program used

to carry out the calculations was designed by the author and was coded

in FORTRAN by Randall Lundberg in 1969. In 1971, Robert Fuller

constructed this vinyl model.

A1. Roundest Polyhedra

The polyhedron P33, which I have conjectured is theroundestpolyhedron with 33 faces,

is composed of one heptagon (green), fourteen pentagons (yellow), and eighteen hexagons (red-orange).29 of the 33 faces of

P33define a connected assembly with the same combinatorial structure as

29 of the 32 faces of the 'soccer ball' polyhedron (truncated icosahedron).

To view a sequence of eight rotated images ofP33, set the magnification at 80%.

The theory of(sometimes calledroundest polyhedra) was first treated by Simon Antoine Jean L'Huilier (1750-1840). One of several accomplishments for which L'Huilier is remembered today is his generalization tobest polyhedranon-convexpolyhedra of the famous formula of Euler forconvexpolyhedra, which Euler first described in 1750 in a letter to Goldbach:v-e+f= 2;

v,e, andfare the numbers of vertices, edges, and faces, respectively.The history of this development is summarized in George G. Szpiro's fascinating book, "Poincaré's Prize".

The subject of roundest polyhedra is conventionally referred to as

'The Isoperimetric Problem for Polyhedra'.

In the 19

^{th}century the theory of roundest polyhedra was investigated by Jakob Steiner, E. Kötter, Lorenz Lindelöf, and Hermann Minkowski, and in the 20^{th}century by Ernst Steinitz, Michael Goldberg, and Victor Klee.Lorenz Lindelöf proved a fundamental theorem about roundest polyhedra (

cf.my 1986 conference paper) — first in 1869 and again in 1899. Klee greatly deepened the analysis and also extended it to dimensions beyond three.

Jakob Steiner (1796-1863) Lorenz Lindelöf (1827-1908) Hermann Minkowski (1864-1909)

Georgy Voronoy (1868-1908) Ernst Steinitz (1871-1928) Victor Klee (1925-2007)

Ms. Ivonne Vetter, a staff member at the Mathematisches Forschungsinstitut Oberwohlfach, noticed yesterday (March 28, 2011) that Lorenz Lindelöf's picture was missing from the pantheon above. She then kindly emailed me the splendid Lindelöf photo, from the Oberwohlfach archives, that you now see here.

Thanks, Ivonne!I regret that unlike Paul Halmos (1916-2006), I never developed the habit of taking a snapshot of every mathematician I met. As a consequence, I don't have a photo of Michael Goldberg. If someone will send me one, I'll gladly post it.

My interest in roundest polyhedra was kindled in 1972 when I first saw the late Victor Klee's two films, 'Shapes of the Future — Some Unsolved Problems in Geometry'. Afterwards Prof. Klee (1925-2007) gave me copies of his filmscript booklets, Part I: Two Dimensions and Part II: Three Dimensions. With the permission of the Mathematical Association of America, I have made digital copies of both of these films and uploaded them to YouTube:

Part I: Two Dimensions (YouTube video)

and

Part II: Three Dimensions. (YouTube video)I met Michael Goldberg several times in the late sixties at AMS meetings. We talked mostly about minimal surfaces and Voronoi polyhedra. Like Victor Klee, Michael was a modest man. Unfortunately he never mentioned to me that he'd written a groundbreaking article in 1934, 'The Isoperimetric Problem for Polyhedra'. I regret that by the time (1972) I had become interested in this subject, he had already 'shuffled off this mortal coil'.

The roundest polyhedron question for R

^{3}is:Among the convex polyhedra withnfaces, which has the smallest ratio ofS^{ 3}/V^{ 2}?

(= surface area andS= volume.)VIn 1897, Minkowski proved that a roundest polyhedron exists for every

n≥ 4. In 1899 it was proved by Lorenz Lindelöf that anecessarycondition for a polyhedronto be roundest is thatP

( l)circumscribes a sphere, andP

(2) the inscribed sphere is tangent to all the faces ofat their respective centroids.P

For any polyhedron circumscribed about the unit sphere,S/V= 27V= 9S. Hence minimizingS^{ 3}/V^{ 2}is equivalent to minimizingS(orV).

It has been conjectured — but never proved — that the roundest polyhedron is alwayssimple,i.e.,that its vertices are all of degree three.My computational geometry 1986 conference paper, 'A Defect-Correction Algorithm for Minimizing the Volume of a Simple Polyhedron Which Circumscribes a Sphere', summarizes a 1986 experimental investigation of roundest polyhedra. Conjectured solutions for 4 ≤

n≤ 35 and forn= 42 are described and illustrated. A supplement (May, 1986) includes conjectured solutions for 36 ≤n≤ 41 andn= 43. The supplement includes data from William Tutte's analysis of the asymptotic number of simple polyhedra as a function of the numbernof faces.

Below are links to sets of face templates for

n= 8, 25, 33, and 44. These templates can be used to construct physical models of conjectured 'roundest' polyhedra. For the casen= 8, I have provided templates only for faces 1, 2, 3, and 4, since faces 5, 6, 7, and 8 are congruent to faces 1, 2, 3, and 4, respectively.For cases

n= 8, 25, and 44, eachvertexof the template is labeled with an integer, but for the casen= 33, eachedgeis labeled with an integer.template for

n= 8: face 1

template forn= 8: face 2

template forn= 8: face 3

template forn= 8: face 4templates for

n= 25: ALL faces

templates for

n= 33: ALL faces

templates for

n= 44: faces 1, 2, 3

templates forn= 44: faces 6, 8, 14

templates forn= 44: faces 7, 9, 10

templates forn= 44: faces 5, 11, 12, 34

templates forn= 44: faces 13, 15, 16

templates forn= 44: faces 17, 18, 41

templates forn= 44: faces 19, 20, 44

templates forn= 44: faces 21, 22, 23

templates forn= 44: faces 4, 24, 26

templates forn= 44: faces 25, 31, 33

templates forn= 44: faces 27, 28, 29, 30

templates forn= 44: faces 35, 36, 37

templates forn= 44: faces 32, 38, 40

templates forn= 44: faces 39, 42, 43

P44, [probably] the 'roundest' polyhedron with 44 faces

IBM Computational Geometry Conference, 1986

(If you can identify this participant, please tell me who she is!)

Tetsuya Hatanaka and the model he assembled in Spring, 2011

ofP44, the putative 'roundest' polyhedron with 44 facesAlthough I still have templates — plotted in 1986 — for the faces of

P44, I am unable to locate any numerical data forP44. I no longer have either numerical data or graphic images forP45andP46— the last two cases I investigated. Recently (Spring, 2011), using the face templates listed above, I have constructed a new physical model ofP44. Below is a cross-eyed stereo pair of photos of this new model. Identifying itsD_{2d}symmetry requires careful scrutiny of the polyhedron!

P44(a fresh model constructed in 2011), the putative 'roundest'

polyhedron with 44 faces (identfied in 1986). Its symmetry isD_{2d}.Below is a stereo image of a simpler convex polyhedron with

D_{2d}

symmetry. It has the simplest possible combinatorial structure of all

convex polyhedra withD_{2d}symmetry. It is shown inscribed in a

cube of edge length two. One of its two orthogonalc2axes is the

blueline, which coincides with thex-axis. The otherc2axis is the

redline, which coincides with the line between points at 0,1,1)

and (0,-1,-1).

P44also has two orthogonal planes of mirror symmetry, the horizontal

(equatorial)plane and the verticalx-yplane. It is invariant under twox-z

successive rotations—a half-turn about the [vertical]-axis, followed byz

a quarter-turn about the [horizontal]-axis.xLike the convex hexahedron shown above, if

P44is viewed from the

back, it appears the same as it does from the front except that its image

is rotated by a quarter-turn.My 1986 conference paper on roundest polyhedra lists only two other

values ofnfor which the roundestn-faced polyhedron hasD_{2d}symmetry:

8 and 20 (cf.pp. 162-164).Although a separate template is provided here for each of the forty-four faces of

P44(see above), there are only eight different face shapes — two pentagons (p_{1}andp_{2}) and six hexagons (h_{1},h_{2},h_{3},h_{4},h_{5},h_{6}). The number of specimens of each face and its symmetry are given below. Each of the asymmetrical faces occurs in two four-specimen subsets; the faces in one subset are reflected images of the faces in the other subset.

- pentagons

- 4
p_{1}:d2symmetry- 8
p_{2}:c1symmetry (i.e.no symmetry)- hexagons

- 4
h_{1}:d2symmetry- 4
h_{2}:d2symmetry- 4
h_{3}:d2symmetry- 4
h_{4}:d2symmetry- 8
h_{5}:c1symmetry- 8
h_{6}:c1symmetryThe results described in my 1986 conference paper and in the May 1986 supplement demonstrate that roundest polyhedra exhibit a great variety of symmetries. Determining the symmetry of one of these polyhedra is an interesting challenge. It is probably best accomplished by examining, from different directions, either a physical model or a virtual model — in

Mathematica, for example.In 1989, my former student David M. Aubertin wrote a computer science masters thesis in which he designed and implemented a program for obtaining and displaying solutions for the

4-dimensional versionof the isoperimetric problem for polyhedra. His thesis, which is entitled 'Optimization of Four-Dimensional Polytopes', contains stereoscopic perspective drawings of projections of the solution 4-polytopes into R^{3}. It also includes a listing of the elegantly structured FORTRAN program Dave wrote to obtain his results, together with a short BASIC program for displaying the perspective images. He has told me that you are welcome to copy and use both of these programs.

Back to GEOMETRY GARRET