Studies in Geometry

Christopher Arthur June, 2004

Tesselators

A Tesselation is commonly known as a tiling of a the plane using a single, primative geometric shape. If the shape is copied and transated, the copies interlock to cover the entire plane without any overlapping. The idea of a tesselation is expanded by Kepler to allow not just single shapes, but tilings by aggregate patterns of regular and star polygons. There is in either case a nite pattern. A tesselator is a relation that enumerates each of the copies. It serves to organize a tesselation for the purposes of projection onto manifolds. We introduce two kinds of tesselators.

1.1

Toriodal Tesselators

TT : R2 Z2 is a surjective map that essentially assigns a unique pair of integers (i, j) to each instance of the pattern. For the purpose of building aggregate tesselations, the pattern is topologically equivalent to a torus, so that we can imagine it also as a rectangle having the top meeting the bottom and the left meeting the right.

TT

(0,2) (0,1) (-2,0) (-1,0) (0,0) (1,0) (2,0) (0,-1) (0,-2)

Figure 1: Enumeration by toroidal tesselator

TH

(0,2)

(0,1) (-1,0) (0,0) (-1,-1) (-2,-1) (0,-1) (1,-1) (1,0)

(2,1) (2,0)

(0,-2)

Figure 2: Enumeration by hexagonal tesselator 2

1.2

Hexagonal Tesselators

TH : R2 Z2 is similiar to TT , only that the patterns are enumerated in staggered rows due to the nature of how hexagons interlock.

Tubes

A tube T(C, s) in R3 is a cross-section C extruded along a path s. The cross-section c : [0, 1] C R2 is a closed curve with no intersections (an injective map). The path s : A R R3 is a continuously dierentiable C 2 curve.

Figure 3: Cross-section and path

2.1

Construction

To construct a tube from its path and cross-section, we project the crosssection at each point along the path. We rst derive a basis for the plane into which we make the projection, then dene a mapping from the space of the cross-section. Let t A be arbitrary, p = s(t) and v = s (t) so that vp R3 . For the cross-section, let e1 , e2 R2 be a basis. At point p we have p a plane normal to v with basis (1 )p , (2 )p R3 . For each a C, we have e e p 3

a = xe1 + ye2 . The projection T : R2 R3 that maps the cross-section is dened as 1 1 e1 e2 x 1 2 +p T (a) = e2 e2 y 3 3 e1 e2

a' a

Figure 4: Mapping cross-section onto path, (a = T (a)) Now, as t is arbitrary, this mapping can be repeated for all t. As an example, consider the usual basis for the cross-section space, and for the planes, we derive an expression for the basis at a point by parametrizing everything in terms of t: n(t) = e1 (t) = s (t) s (t) n (t) n (t)

e2 (t) = n(t) e1 (t) We obtain a coordinate system f : A [0, 1] R3 for the entire tube surface. 4

f (t, u) = (c1 (u)1 (t) + c2 (u)2 (t) + s(t) e e

Figure 5: Complete tube T(C, s)

Regular Polyhedral Mappings

A polyhedral map P = {Pi }n is a set of transformations Pi : V Wi i=1 dened on the a bounded space V = A [0, ) where A is a polygon in R2 . A polyhedral map is regular if: 1. A is a face of a regular polyhedron. 2. P maps to the volumes above each face on the polyhedron. 3. Wi Wj = for all i, j. For example, for a cube we have P = {Pi }6 for the six sides with i=1 A = [0.5, 0.5] [0.5, 0.5]. Note that Pi is essentially a rotation followed by a translation. One application of polyhedral maps is in the construction of symmetric solids and surfaces in which the symmetry is guided by underlying polyhedra. By carefully choosing the boundary of a subset of V , P (V ) becomes a surface with continuous transitions from one face to the next. 5

P
W11 W10

W1 W2

W3 W9 V W8 W12 W4 W7 W6 W5

Figure 6: A dodecahedral map

Figure 7: A cube mapping of a cross 6

Determing Cubic Spline Functions

Arising frequently in geometry is need for a smooth curve between two endpoints, with the added requirement that the direction of the curve at its endpoints can be specied. Cubic splines provide a straightforward solution and a simple algebraic means to derive a continously dierentiable function for the curve. Let X be the set of all cubic polynomials having real coecients. A cubic spline seqment in Rn is a function s : A R Rn such that s : t (f 1 (t), f2 (t), ..., fn (t)) where {fi }n X. For isolated segments, i=1 it is convenient to consider A = [0, 1]. A cubic spline segment is a smooth curve that can be completely determined by the position and heading at its endpoints. The headings are essentially the derivatives of the curve at the endpoints (more properly, they are the limits of the derivatives).

s(1) s'(1)

s(0) s'(0)
Figure 8: Heading and position of endpoints If the endpoints and headings are known, we can determine each fi by solving a system of linear equations. Consider rst the case of a spline segment in R. Then s(t) = f (t). Let f (t) = a0 +a1 t+a2 +a3 t3 . By dierentiation 2 of f and substitution, we have s(0) = a0 s(1) = a0 + a1 + a2 + a3 s (0) = a1 s (1) = a1 + 2a2 + 3a3 7

which yields the following solutions: a0 a1 a2 a3 = s(0) = s (0) = 3s(0) + 3s(1) 2s (0) s (1) = 2s(0) 2s(1) + s (0) + s (1)

For the case in Rn , we have n systems of equations all with similar solutions. We note that i s = fi , and generate the systems accordingly. A cubic spline in Rn may be expressed as a piecewise function composed of a series of segments, typically linked together to form a continuously differentiable C 1 curve.

s4

s3 s1 s2

Figure 9: Segments chained together If S is a cubic spline having two segments s1 and s2 , then we could have, for example, the domains of s1 as [0, 1] and s2 as [1,2], and the chaining condition that s1 (1) = s2 (1) and s1 (1) = s2 (1).

Special Transformations of the Unit Circle

Let s be a parameterization of the unit circle A.

such that s : t (cos 2t, sin 2t) A transformation Tr : A B R2 is a radial modulator if there exists a function r : [0, 1] R such that Tr (A) = {r(t)s(t)|t [0, 1]} Tr transforms the circle by changing its radius at each point. We show that Tr is well-dened. Let x, y A such that x = y. Since s is injective, there exist a, b [0, 1] where a = s1 (x) and b = s1 (y). Now x = y, so s(a) = s(b) r(a)s(a) = r(b)s(b). Since, Tr s(t) = r(t)s(t)t, Tr s(a) = Tr s(b) and nally Tr (x) = Tr (y).

[0,1] rs

A Tr

Tr (A)

Figure 10: Diagram of radial modulator

Tr

Figure 11: r(t) = 1 cos 8t + 1 3 A transformation Tt is a toroidal modulator if there exists a function : [0, 1] R2 such that : t (R1 R2 cos 2nt, R2 sin 2nt) and Tt (A) = {(1 s1 , 1 s2 , 2 s3 )|t [0, 1]} with R1 , R2 R and n Z+ . For convenience we dene a function G : [0, 1] R3 such that G = (1 s1 , 1 s2 , 2 s3 ). The proof that Tt is well-dened proceeds as before, noting that for x, y A, x = y G(s1 (x)) = G(s1 (y)) T (x) = T (y).

10

[0,1] G

Tt

Tt (A)

Figure 12: Diagram of toroidal modulator

Tt

1 Figure 13: (t) = (1 1 cos 28t, 3 sin 28t) 3

11

References
[1] [2] Spivak, Michael.(1965). Calculus on Manifolds. Eric W. Weisstein. Transformation. From MathWorld. http://mathworld.wolfram.com/Transformation.html

12