Crystallographic
definitions for cube assemblies and ballpyramid assemblies from polycubes and
polyballs, respectively
By Berchtold Frank Rordorf, Château d’Epanvilliers, Musee
du Jeu Ancien, Feldpark 20, CH6300 Zug
Abstract
Definitions in accordance with crystallographic
nomenclature are proposed for all possible polycube assemblies up to 6 layer cubes
and for all ball pyramids up to 8 layer tetrahedrons. A new definition has thus also been given for
the 3 layered 27cube Soma–Cube as defined by Piet Hein, who used the term
“nonregular” which was not in agreement with crystallographic
nomenclature.
Crystallographic restrictions have been introduced for
both types of threedimensional puzzles. Only polycube assemblies of the
simple cubic structure are included. Polyball
assemblies with cubic densest ball packing and face centered cubic unit cell
(fcc) only, leading to tetrahedral pyramids, are included. It is shown that all polyball clusters to
pentamers are both members of the cubic close packing (ccp) and the hexagonal
close packing (hcp) structures.
A common nomenclature has been introduced for the
polycube and the polyball clusters. It
is used up to tetramers for polyballs and to pentamers for polycubes. The
number of possible spatial positions for these clusters in the simple cubic
structure for the polycubes and in the face centered cubic for the polyballs
has been elucidated as a preparation for computer simulations for finding puzzle
solutions.
Definitions in accordance with crystallographic
nomenclature are presented for the 3, 4 and 5layer cube assemblies of the
polycubes and for the 4, 5, 6, 7 and 8layer polyball pyramid
assemblies. Hand found solutions are presented
for the 4 and 5layer polycube assemblies and for the 5, 6, 7 and 8layer
polyball tetrahedron assemblies. The polycube clusters have been colored in a
checkered fashion and in order to increase the challenge for finding solutions.
Introduction
A large community of puzzlers and
mathematical recreation interested persons has studied the practical aspects of
polycubes and polyominoes and an overview is given on the polypages of the
internet site recmath.com [1, 2]. Most
of this material has only been published on internet sites with a few exceptions
such as the book in German by Ekkehard Künzell on games played with pentacubes
[3]. The most systematic studies have
been published by Thorston Sillke of the Mathematikum at the
On the other hand polycubes problems are widely treated in computing and combinatorics and are a popular problem in discrete geometry and an important tool in statistical physics for instance in computations related to percolation processes and branched polymers [5]. These publications are, however, in general out of scope for the puzzle and gaming community. Again for polyball puzzles, information is only available on internet, and one of the most comprehensive publications is the document “Some Notes on BallPyramid
and Related Puzzles” by Leonard Gordon [6].
The nomenclature of polyominoes for flat and polycubes for threedimensional orthogonal cube clusters of the puzzle community has been dropped [1, 2, 3] and been replaced by a harmonized nomenclature for both polycube and polyball assemblies by adopting the standard nomenclature used in chemistry for polymers. There are many excellent introductions for readers who are not familiar with crystallographic nomenclature and symmetry [7, 8] and a classic student introduction in crystallography for students [9]. As this paper should be accessible for the gaming community some basic concepts are presented here as an introduction.
Symmetry according to G. Burns [7] is a visual demonstration that an object is invariant to a given transformation. The symmetry transformations of the identity and of translation are trivial and will not be considered here. Point symmetry operations are all of the transformations which are defined with respect to at least one point of the object which does not move during the operation. These points are often, but need not be, identical with the points which define the lattices and the point groups. A lattice is defined as an array of points in space with each point having identical surroundings. The simplest way to generate such a lattice is by invoking translational invariance. Mathematically this process is described by applying translational symmetry by a primitive translation vector, which in turn is the sum of three spatial unit vectors for any three dimensional lattice.
Lattices are a mathematical concept and their introduction is essential in order to abstract from the complications of real life situations, such as encountered with molecules or crystals, and make these complex bodies amenable to symmetry considerations. As an example let us consider a wooden cube: it has 8 corners, which define 8 visible points. More important for the present consideration is the point formed by the intersection of the 4 space diagonals. In case of a homogenous cube this point defines its center of mass and can be used to describe the position of the cube in the lattice. Likewise the position of a wooden ball in a crystal lattice can be described by using its radial center point. For symmetry considerations the constituting cubes or balls can thus be replaced by their center of mass points.
Describing the constituents of polycube or polyball puzzles by resorting to such lattice descriptions has some immediate consequences: a single cube for instance, no longer has any symmetry at all. Likewise, all flat polycube or polyball clusters no longer have the mirror symmetry with respect to the plane as defined by the center of mass points.
The three unit vectors which sum up to the primitive translation vector, also define a space, called the primitive unit cell of the lattice. Primitive unit cells always contain only one single lattice point, in general the result of the sum of the contribution fractions of lattice points (for instance 8x 1/8^{th} of a lattice point for a simple cubic lattice). Often the unit cell defining a crystallographic lattice is, however, not a primitive unit cell, as for example for the face centered cubic structure of the cubic system.
Polycube puzzles
All of the polycube assemblies discussed here and all of the resulting three dimensional puzzles built up from these clusters have a lattice unit cell which has the same dimension as the unitcube of the puzzle. This is a consequence of the crystallographic restriction to orthogonal assemblies only: all neighboring cubes in all possible assemblies always share four corner points each. All assemblies are of the simple cubic structure, having one cube sitting on each corner of the cubic unit cell with a crystal lattice of the point group Pm3m (O^{1}_{h}) [7, p.193]. In the space group nomenclature of the 32 crystallographic point groups, Pm3m is the international symbol and O^{1}_{h} the SchoenfliesSymbol of the simple cubic structure.
The
primitive unit cell is identical with the lattice unit cell in the simple cubic
structure. They in turn are identical
with the space defined by the corner points of the translated unit cube,
translated so that the center of mass point comes to sit on the lattice point.
No 
Object name 
Letter & (Künzell)
notation 
Position count 
Mirror planes 
C2 
C3 
C4 
Point group 
1 
PC_1F 
1_I 
1 
4m+3m 



4mm(C4v), Pm3m (O^{1}_{h}) 
2 
PC_2F 
2_I 
3 
2m+m 
2xC2 


mm2(C_{2v}), mmm(D_{2h}) 
3 
PC_3F_180 
3_I 
3 
2m+m 
2xC2 


mm2(C_{2v}), mmm(D_{2h}) 
4 
PC_3F_90 
3_V 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
5 
PC_4F_180_180 
4_I 
3 
2m+m 
2xC2 


mm2(C_{2v}), mmm(D_{2h}) 
6 
PC_4F_180_90 
4_L 
24 
m 



1(C_{1}), m(C_{1h}) 
7 
PC_4F_90_270 
4_N 
12 
m 
1xC2 


2(C_{2}), 2/m(C_{2h)} 
8 
PC_4F_90_90 
4_O 
3 
4m+3m 
4xC2 

1xC4 
4mm(C4v), 4/mmm(D4h)

9 
PC_4F_180y90 
4_T 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
10 
PC_4T_90y0 
4_A 
8 
3m+m 

1xC3 

3m(C_{3v}), ‾6m2(D_{3h}) 
11 
PC_4R_90_0 
4_S1 
24 
m 
1xC2 


2(C_{2}), 2/m(C_{2h)} 
12 
PC_4R_270_0 
4_L1 
24 
m 
1xC2 


2(C_{2}), 2/m(C_{2h)} 
13 
PC_5F_180_180_180 
I (10) 
3 
2m+m 
2xC2 


mm2(C_{2v}), mmm(D_{2h}) 
14 
PC_5F_180_180_90 
L (11) 
24 
m 



1(C_{1}), m(C_{1h}) 
15 
PC_5F_180_90_180 
V (13) 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
16 
PC_5F_180_90_270 
N (40) 
24 
m 



1(C_{1}), m(C_{1h}) 
17 
PC_5F_90_180_90 
U (90) 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
18 
PC_5F_90_180_270 
Z (20) 
12 
m 
1xC2 


2(C_{2}), 2/m(C_{2h)} 
19 
PC_5F_90_270_90 
W (30) 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
20 
PC_5F_180_90_90 
P (60) 
24 
m 



1(C_{1}), m(C_{1h}) 
21 
PC_5F_180_180y90 
Y (12) 
24 
m 



1(C_{1}), m(C_{1h}) 
22 
PC_5F_90_180y270 
F (70) 
24 
m 



1(C_{1}), m(C_{1h}) 
23 
PC_5F_180_90y270 
T (80) 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
24 
PC_5F_90y180y270 
X (50) 
3 
4m+m 
4xC2 

1xC4 
4mm(C4v), 4/mmm(D4h) 
25 
PC_5T_90_90_0

Q (61) 
24 
1m 



m(C_{1h}) 
26 
PC_5T_90_0_90

A (37) 
24 
1m 



m(C_{1h}) 
27 
PC_5T_180_90y0

L3 (81) 
24 
1m 



m(C_{1h}) 
28 
PC_5T_180y90_0

T2 (82) 
12 
1m 



m(C_{1h}) 
29 
PC_5T_180y90y0

T1 (51) 
12 
1m 
1xC2 


mm2(C_{2v}) 
30 
PC_5R_180_90_0

L4 (41) 
24 




1(C_{1}) 
31 
PC_5R_180_270_0

J4 (42) 
24 




1(C_{1}) 
32 
PC_5R_90_270_0

N1 (31) 
24 




1(C_{1}) 
33 
PC_5R_270_90_0

S1 (32) 
24 




1(C_{1}) 
34 
PC_5R_90_180y0

J2 (72) 
24 




1(C_{1}) 
35 
PC_5R_270_180y0

L2 (71) 
24 




1(C_{1}) 
36 
PC_5R_90_270y0

N2 (33) 
24 




1(C_{1}) 
37 
PC_5R_270_90y0

S2 (34) 
24 




1(C_{1}) 
38 
PC_5R_90_180_0

J1 (22) 
12 

1xC2 


mm2(C_{2v}) 
39 
PC_5R_270_180_0

L1 (21) 
12 

1xC2 


mm2(C_{2v}) 
40 
PC_5R_90_0_270

V2 (35) 
12 

1xC2 


mm2(C_{2v}) 
41 
PC_5R_270_0_90

V1 (36) 
12 

1xC2 


mm2(C_{2v}) 
Table 1. All polycubes up to pentamers are listed. For reference, the table also lists the nomenclatures for the polycubes by letters [4] and in brackets by Künzell [3]. Underlined bold angles point to the second layer. C2, C3, C4 indicate the presence of 2, 3, and 4fold symmetry axis. The point groups of the clusters are given both with (italic) and without the mirror plane (m) parallel to the surface of the flat pieces.
In order to define the polycube clusters the following nomenclature is introduced (see Table 1): PC stands for Poly Cube, F for flat, T for three dimensional with a mirror plane, R for racemic (chiral); angles are 180 for straight and 90 / 270 for right angles, y indicates a branching, and _0 a change to the second layer for the next cube. In this case a more correct _90 , as used in case of the polyballs, has been replaced by _0 for clarity. The maximum number of assembled cubes lye on the surface and one starts out from the longest straight piece. Angles are given starting with the 3rd piece. Each cube can be treated as a point for defining the symmetry. All flat pieces would otherwise have an additional mirror plane (m) parallel to the surface of the flat pieces. The column labeled position counts indicates the number of possible spatial positions of each cluster in the cubic lattice (of course without translations). This information is useful for designing computer simulations for determining the maximum number of possible solutions to the puzzles. Figure 1 shows the different polycubes as listed in Table 1.
Table 2 presents a summary of the number of possible polymer of each type, flat, threedimensional with at least one out of plane mirror plane, or in form of raceme pairs. The number of constituting cubes is also indicated and these numbers were used for finding unique definitions for the different polycube assemblies forming cubes again.
Table 2. Summary of polycube assemblies:

Flat 
Number of cubes 
Threedim. non
chiral 
Number of cubes 
Racemes 
Number of cubes 
Total cubes 
Monomer 
1 
1 




1 
Dimer 
1 
2 




2 
Trimer 
2 
6 




6 
Tetramer 
5 
20 
1 
4 
2 
8 
32 
Pentamer 
12 
60 
5 
25 
6x2=12 
60 
145 







186 
Unique definitions for the macroscopic polycube assemblies up to the 6 layer cube:
 The 2 layer 8unit cube can be assembled from all
pieces up to trimers, while the linear trimer is replaced by a dimer.
 The 3 layer 27unit SomaCube can be assembled from the
nonregular polycube pieces up to tetramers (Piet Hein Definition).
 The 3 layer 27unit cube can also be assembled from all
flat pieces up to tetramers, while the linear tertamer is replaced by a
dimer. Therefore the clusters 1, 2x2, 3,
4, 6, 7, 8, 9 of Table 1 and Figure 1 are used.
 The 4 layer 64unit cube can be assembled from the flat
pentamers (11x4), while the linear pentamer is replaced by the linear dimer, trimer
and tetramer pieces (2+3+4 = 9).
 The 5 layer 125unit cube can be assembled from all pentamers
pieces except the four raceme pieces with a symmetry axis.
 The 6 layer 216unit cube can be assembled from all
clusters to pentamers, plus the linear and the three dimensional nonraceme
pentamer pieces.
The definitions for the 3, 4 and 5layer cube assemblies are illustrated in the scheme of Figure 1. The same numbers are used in turn in Figure 2 and 3 for presenting a solution for the 6, 5 and the 4layer cubes. The checkering patterns of the polycubes are defined by the solutions. Many solutions are possible for the 6 layer cube and the cubes under 4 layers [4].
Figure 1. The polycubes as listed in Table 1 are shown. Cubes presented by fine lines indicate cubes in the bottom layer, while bold stands for cubes in the upper layer. Note that the bold cubes hide a cube in the bottom layer in all cases, except for the last two cubes of the racemes 40 and 41. Raceme pairs are indicated by lines for the mirror planes. The brackets illustrate the above definitions for the 3, 4, 5 and 6layer cubes. The total numbers of unitcubes are also indicated, e.g. 125 cubes for the 5layer cube.
Figure 2. Solution for the 6layer 216unit cube, showing the cube layers from top to bottom. The pieces 13 and 25 to 29 are used two times and this is indicated by the primes. For clarity all clusters are once more shown on the bottom (see Table 1 and Figure 2).

Figure 3. Solutions of 4layer, top, and 5 layer polycube assemblies. The cubes are presented by layers from top down. The numbers correspond to the numbers in Table 1 and Figure 1. The solutions for the 4 and 5layered cubes shown define one possible coloring of the clusters. Agreement in the coloring of all pieces is achieved for the 4, 5 and 6layer solutions, except for number 22 where two differently checkered individuals are needed. A mirror image solution of the 5layered cube would also be possible, if all raceme pairs where colored to form mirror image pairs. This condition turns out to be only fulfilled for the pair 36/37 in the solution presented. 





Polyball puzzles
The crystallographic situation for the polyball puzzles is more complicated than in case of the polycubes. Only the cubic densest ball packing (layers a b c a b c...) is considered here, as a crystallographic restriction, thus excluding the hexagonal densest ball packing (layers a b a b…). The cubic densest ball packing is the structure of metallic copper and is of point group Fm‾_{3}m(O^{5}_{h}). One possible macroscopic manifestation of this crystal structure is the famous Prussian army canon ball pyramid with its tetrahedral appearance. The three crystallographic axes are equal, orthogonal to each others and are parallel to the three four fold axis. The 3fold axis point along <111> and there are mirror planes perpendicular to <100> and <110>. The unit cell of the face centered cubic lattice (fcc) is not a primitive unit cell as it contains 4 lattice points with 8x 1/8 + 6 x 1/2 points.
Each ball is surrounded by 3+6+3 balls in the triclinic direction, or by 4+4+4 balls in the cubic direction, forming 6 octahedral and 8 tetrahedral holes. This leads to the formation of triclinic layers in the planes which are parallel to the macroscopic tetrahedron faces and cubic layers which are parallel to the opposite nonintersecting edge pairs of the tetrahedron. This is the reason why cubic (90º and 270 º) and triclinic (60 º and 300 º, 120º and 240 º) angles are intermixed in the polyball clusters.
A closer look at the buildup of the densest, or closest, ball packing reveals that the cubic (layers a b c a b c..) and hexagonal (layers a b a b…) closest packing are two fully interchangeable crystal structures. If a first plane is built up by arranging 6 balls around each ball on a flat surface, a hexagonal twodimensional lattice is obtained (layer a). A second layer of the same type can be added by adding balls into the dips of the first layer, by forming tetrahedron holes (layer b). Note that placing the first ball, defines the translation vector of the second layer with respect to the first one. Indeed every second dip stays empty. Both layers a and b are thus part of the hexagonal (hcp – P_{6/}mmc (D^{2}_{h}) ) and the cubic closest packing (ccp – Fm‾_{3}m(O^{5}_{h}) ). The third layer can now be built by placing the first ball exactly over a ball of layer a, thus leading to the hexagonal structure abab… Alternatively the first ball of the third layer may be placed over a tetrahedron hole of the first layer thus leading to the cubic structure abc…
Note, however, that the same choice exists when building the fourth layer, thus permitting to build up sequences abab and abca, or alternatively abac and abcb, and so on. A free interchange of hcp and ccp layer sequences is thus possible and the hexagonal and the cubic structures are fully interchangeable. The packing densities of the two structures are also identical. Note that the octahedron holes arise in the ccp structure when placing the third layer, while there are hexagonal tubes in a pure hcp crystal structure across the complete crystal.
The rare metals Au, Ag and Pt, for instance, crystallize in the cubic structure, like also copper. Be, Mg, Zn, Cd, for instance crystallize in the hexagonal structure. Cobalt can exist in either the cubic as αCo, or hexagonal structure as βCo, but the layers can also freely interchange resulting in a mixed form of cobalt [9].
This consideration shows that in order to find out if the polyball clusters considered here, namely up to 4ball assemblies, can be part of both structures we only have to test if these assemblies can be placed in up to two layers. This is obvious for all the linear, or all of the flat pieces which contain no 90degree angles. All of the flat pieces which contain 90degree angles can be placed in two layers. The same holds for all of the threedimensional 4ball assemblies – they can also be accommodated in just two layers. This proves that all of the polyball assemblies up to tetramers can be used in both the hcp and the ccp structure.
Only the tetrahedron ball pyramids are considered here and this goes beyond a limiting to cubic close packed structure for polyball assemblies as other macroscopic structures are possible for the face centered cubic lattice: cube, rhombdodecahedron, pentagondodecahedron, tritetrahedron, deltoiddodecahedron, or tetrahedronpentagondodecahedron.
For convenience a trigonal 60º coordinate system in accordance with the macroscopic tetrahedron is used to describe the puzzle assemblies, while the face centered cubic crystal lattice is aligned with the opposite non intersecting edges of the tetrahedron. All 90º angles of polyball clusters are aligned with the fcclattice.
In order to define the polyball clusters the following nomenclature is introduced: PB stands for Poly Ball, F is flat, T is three dimensional with a mirror plane, R are racemes or chiral pieces. Angles are 180 for straight and 60 / 300 for 60 degree, 120 / 240 for 120 degrees and 90 / 270 for right angles. Branching is indicated by y and is only introduced, if a cluster can not be described by a single continuous ball string. The maximum number of assembled balls lye on the surface and one starts with the longest straight piece. The ball carrying the number in Figure 4 is the starting point.
Racemes are placed for having the maximum number of balls on the bottom surface and permitting a description with the angles ordered in the sense 30º (300º), 90º (270º), 60º (300º), and last 120º (240º). For instance No 26 is called PB_4R_90_120 and not PB_4R_120_90. Angles are given starting with the 3^{rd} ball. Each ball is treated as a point for defining the symmetry. All flat pieces would otherwise have an additional mirror plane. All possible spatial positions in the cubic lattice are summed up in the position count. C2, C3, C4 stand for 2, 3, 4fold symmetry axis.
No 
Object name 
Gordon notation 
Position count 
Mirror plane 
C2 
C3 
C4 
Point group 
1 
PB_1F 

1 
nm 
n 
n 
n 
1(C_{1}), sphere 
2 
PB_2F 
I2 
6 
2m+∞m 
3xC2 


mm2(C_{2v}), mmm(D_{∞h}) 
3 
PB_3F_180 
I3 
6 
2m+∞m 
3xC2 


mm2(C_{2v}), mmm(D_{∞h}) 
4 
PB_3F_60 
D3 
8 
3m+m 
3xC2 
1xC3 

3m(C_{3v}), ‾6m2(D3h) 
5 
PB_3F_120 
C3 
24 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
6 
PB_3F_90 
L3 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
7 
PB_4F_180_180 
I4 
6 
2m+∞m 
3xC2 


mm2(C_{2v}), mmm(D_{∞h}) 
8 
PB_4F_180_60 
P4 
48 
m 



1(C_{1}), m(C_{1h}) 
9 
PB_4F_180_120 
J4 
48 
m 



1(C_{1}), m(C_{1h}) 
10 
PB_4F_180_90 
L4 
24 
m 



1(C_{1}), m(C_{1h}) 
11 
PB_4F_120_120 
C4 
24 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
12 
PB_4F_60_120 
D4 
12 
2m+m 
3xC2 


mm2(C_{2v}), mmm(D_{2h}) 
13 
PB_4F_120_240 
W4 
24 
m 
1xC2 


2(C_{2}), 2/m(C_{2h)} 
14 
PB_4F_90_90 
O4 
3 
4m+m 
4xC2 

1xC4 
4mm(C_{4v}), 4/mmm(D_{4h}) 
15 
PB_4F_90_270 
S4 
12 
m 
1xC2 


2(C_{2}), 2/m(C_{2h)} 
16 
PB_4F_120y240 
Y4 
8 
3m+m 
3xC2 
1xC3 

3m(C_{3v}), ‾6m2(D_{3h}) 
17 
PB_4F_90y180 
T4 
12 
1m+m 
1xC2 


2/m(C_{2h}), mm2(C_{2v}) 
18 
PB_4T_60_60 
 
2 
6m 
3xC2 
4xC3 

23(T) 
19 
PB_4T_60_240 
 
24 
1m 



m(C_{1h}) 
20 
PB_4T_60_90 
 
24 
2m 
1xC2 


mm2(C_{2v}) 
21 
PB_4T_90y240 
 
48 
1m 



m(C_{1h}) 
22 
PB_4R_120_120 
 
24 

1xC2 


mm2(C_{2v}) 
23 
PB_4R_240_240 
 
24 

1xC2 


mm2(C_{2v}) 
24 
PB_4R_120_240 
 
24 

1xC2 


mm2(C_{2v}) 
25 
PB_4R_240_120 
 
24 

1xC2 


mm2(C_{2v}) 
26 
PB_4R_90_120 
 
24 




1(C_{1}) 
27 
PB_4R_270_240 
 
24 




1(C_{1}) 
28 
PB_4R_90_240 
 
24 




1(C_{1}) 
29 
PB_4R_270_120 
 
24 




1(C_{1}) 
30 
PB_4R_60_90 
 
24 




1(C_{1}) 
31 
PB_4R_300_270 
 
24 




1(C_{1}) 
Table 3. All polyball clusters up to tetramers are
listed. Underlined bold angles point to the second layer. Gordon [6] named only flat polyball
assemblies. C2, C3, C4 indicate the presence of
2, 3, and 4fold symmetry axis.
The point groups of the clusters are given both with (italic) and
without the mirror plane (m) parallel
to the surface of the flat pieces.
Only clusters with mutually touching balls are considered here. Clusters which include second to nearest neighbors, using stick connections, have indeed been introduced in some puzzles [1, 2, 3, 6]. The following listing gives unique definitions for the macroscopic tetrahedral polyball assemblies up to the 8 layer ball pyramids:
 The 2 layer 4unit ball pyramid can be built from two dimer
clusters.
 The 3 layer ball pyramid can be built from all trimers,
which do not have a 3fold symmetry axis, plus a single ball.
 The 4 layer 20unit ball pyramid can be built
from all trimers in addition to the linear 4ball piece.
 More challenging, the 4 layer 20unit ball
pyramid can also be built from all nonraceme tetramers, in addition to
the linear 4ball piece, however, the 2edged pyramid must be split in to the trigonal
3ball cluster plus a single ball.
 The 5 layer 35unit ball pyramid can be built
from all monomer to trimer clusters plus the nonlinear flat tetramers which
have at least one plane of symmetry. Note: the mirror plane parallel to the
base surface of the pieces does not count.
 The 6 layer 56unit ball pyramid can be built from all flat trimers and
tetramers.
 The 7 layer 84unit ball pyramid can be built from
all flat 3 to 4ball pieces which have at least one two fold symmetry axis, in
addition to all raceme pieces.
 The 8 layer 120unit ball pyramid can be built from all parts up to tetramer clusters plus all pyramids up to the 2layer one (the 4ball tetrahedron plus a single ball).
These definitions are illustrated in the scheme of Figure 3 for the 4, 5, 6, 7 and 8layer ball pyramids. Please note that according the above definition the polymers 8, 9, and 10 are not part of the 7layer polyball tetrahedron assembly. On the other hand they are of course part of the set forming the 6layer pyramid. The linear tetramer is shown twice in Figure 3, as it is also part of the 4layer pyramid. The 8layer 120ball pyramid, of course contains the linear tetramer only once as indicated by the bracket.
Figure 5 and 6 present possible solutions for the different pyramids. In order to permit an easier reading of these figures the various ballpolymers are once more presented in the same order as given in Table 3. As all clusters plus the monomer only add up to 115 balls, the missing 5 balls have been brought in by adding all pyramids to 2 layers, thus considering the monomer as a one layer pyramid. In the solution of Figure 4 these balls are indicated by 1’ and 18’ balls. The solutions are presented by showing each layer from top to bottom.
Figure 6 shows the solution for the smaller ball pyramids, again according to the definitions given above. So far no systematic searches for further solutions have been undertaken. It is clear that the polyball puzzles are less restrictive than the polycube puzzles presented as the balls are not colored in a checkered pattern in this case.
Figure 4. The polyballs as listed in Table 3 are shown. All balls presented by fine lines are in the bottom layer, while bold stands for balls in the upper layer. Raceme pairs are indicated by lines for the mirror planes. The brackets illustrate the above definitions for the 4, 5, 6, 7 and 8layer ball pyramids. The total number of balls is also indicated, e.g. 120 balls for the 8layer tetrahedron.
Figure 5. Solution for the 8layer 120ball pyramid, showing the pyramid layers from top to bottom. The pieces 1 and 18 are used two times and this is indicated by the primes. For clarity all clusters are once more shown on the bottom. The numbers correspond to the numbers in Table 3 and Figure 3.
Figure 6. From top to bottom the solutions of the 7, 6, 5, and 4layer ball pyramids. The numbers correspond to the numbers in Table 3 and Figure 3 and 4.
References:
[1] RECMATH, archive of recreational mathematics, polypages,
http://www.recmath.com/PolyPages/index.htm (accessed on 18 February, 2015).
[2] Kadon Enterprises, Inc. ©19982015 http://www.gamepuzzles.com/polycube.htm (accessed on 18 February, 2015).
[3] Künzell, E. (1992). Games with Pentacubes. Heuristic Programming in Artificial Intelligence 3: the third computer olympiad (eds. H.J. van den Herik and L.V. Allis), pp.
[4] Sillke, T., “Tiling and Packing results  Polycubes Problems” http://www.mathematik.unibielefeld.de/~sillke/results.html (accessed on 18 February, 2015).
[5] Wang, H., He, Y., Li, X. , Gu, X., Qin, H., Polycube Splines, ComputerAided Design, Elsevier, 2008.
[6] Gordon, L.,
(accessed on 18 February, 2015).
[7] Burns, G., Glazer, A.M., Space Groups for
[8] Kinsey, L.C.,
[9] Kleber, W., Einführung in die Kristallographie, VEB
Verlag
Poly Cube Puzzles:
development source code in:
WEBGL2x2x2cube
3x3x3SomaCube
Short instructions on how to use the graphical user interface is given here although most of the functionality can easily be grasped by playing around with the user interface. It is best to open the 2x2x2Cube in order to get acquainted with the graphical user interface.
Below the canvas one will find a number of buttons and sliders to allow the 3D manipulation of the poly cube clusters. The first and most important button on the top left side indicates Rotate and permits to switch the movement mode from Transfer to Rotate mode. In the transfer mode the canvas background is black and in the rotate mode it is blue.
Mouse controlled transfer or rotation movements are activated upon clicking as chosen cluster. Upon a second click of the mouse the motion control by the mouse is terminated but the cluster remains activated as indicated by the red status line on the button of the canvas. It indicates Translate active cluster n, or Rotate active cluster n, respectively, with n indicating the chosen cluster number and name. The three sliders labeled X, Y and Z are active in this state for the active cluster. A cluster remains active until the canvas background or another cluster is clicked.
A major problem of playing 3D puzzles on computers stem from the fact that computer monitors are two dimensional only. One of the dimensions is thus in principal always hidden and it is important to be able to freely turning view around the to be built up puzzle. This is achieved by the View buttons +90 +45 Tilt 45 90 , which allow to turn the world by plus or minus 90 or 45 degrees. The Tilt button toggles between 45 degree views from above, from below, or from front. The tilted view is reset to front view upon changing the angle of the view. Constantly switching back and forth the viewing angle is important while building up the puzzle in order to observe the spatial position of a cluster to be placed on the to be completed poly cube.
The Align button next to the Z slider is an essential help in order to precisely position a cluster on the growing poly cube. Below the YSlider is another very important toggle 2 states button X>Z and Z>X, respectively, which permits to interchange the horizontal mouse movement between X and Z control. Please note that the the X and Y ax is are coupled in the 45, 135, 225 and 315 degree diagonal view positions. This means that a horizontal mouse movement impacts both the X and Z position at the same time.
The Lock and Block buttons, plus the number scroll field, allows to lock selected clusters in a chosen block from 1 to 5, where 1 is reserved for the final solution. The locking status is indicated on the red status line on the canvas. Locking selected clusters in blocks allows to transfer them as an assembly upon clicking the button Block. Typing LOCK into the File text field discussed below allows to lock all clusters (the finished assembly) to block 1. Typing UNLOCK allows to unlock all clusters again.
The last two rows of controls permit to store intermediate solutions, which is essential when trying to solve one of the challenging puzzles 5x5x5Cube or 6x6x6Cube. Two storage methods are implemented; the rather volatile local storage mechanism in the browser itself. Please test it with your browser on the 2x2x2 cube as does not work with all browsers. Store, Load and Erase are present for this local storage and concern this storage and not the cluster assembly shown on the canvas. The File text field permits to give a name to the local stored game. Typing SOLUTION into the text field brings forth the solution of each cube puzzle.
Clicking the Load button and then the Text button also permit to display a JASON (Java Script Object Notation) version of your progressing solution of the puzzle. Copy this text from [[ to ]] with Ctrl C and paste it into MSNotepad or similar pure text editor with Ctrl V. Save this your computer as a text file (ie 2x2x2cube.txt). You can then recover your saved puzzle at any time by clicking the Browse... button and selecting your file in your file system. This second non volatile storage method is more complicated as Java Script does not allow us to directly save content to the file system for security reasons.
For educational purposes all code has been included into each time a single html page. No file includes or APIs have been used. Rather standard Java Script has been used in combination with the 2015 introduced WEBGL version of Open Graphic Language (OpenGL). The puzzles run therefore only on recent browsers. I have tested the code on Mozzila Firefox, Chrome, normal and Development versions, on Opera and on MS Explorer and Microsoft Edge. When playing on Android mobile devices, please use an USB Mouse connected to the phone via an OTG (on the go) plug. This will permit to play the games using a VRgoggle (I am presently programming the VR extension to obtain two renderings sidebyside with a slight shift in the camera angle).
The following books have been used to develop the code:
E. Angel and D. Shreiner, Interactive Computer Graphics, Pearson, 7th Ed. 2015, ISBN 10: 9781292019345.
D. Cantor and B. Jones, WEBGL Beginners Guide, 2012, PACKT, ISBN 9781849691727.
K. Matsuda and R. Lea, WebGL Programming Guide, 2015, AdisonWesley, ISBN 9780321902924