Design for Muqarnas

Here's a scaling girih tile design in the shape of muqarnas. First is the design with tile edges, then just the girih tile strapwork.

Girih tiling based on a decagonal fractal

My last entry described a girih tiling based on a pentagonal fractal. It's also possible to base a fractal tiling on decagons. This process is simply nesting or subdividing a decagon with five smaller decagons. Starting with a regular decagon, place five smaller regular decagons, edge-to-edge, within the original decagon. One vertex each of the smaller decagons should coincide with a vertex of the larger. Two sides each of the smaller decagons should be edge-to-edge with another small decagon. The lengths of the edges of the smaller decagon are 1/(2x(1+cosine(72))) of the larger. This is the same as calculating the sides of the five smaller decagons by dividing the larger decagon sides by the golden ratio, twice. That is, divide the length of a side of the original decagon by approximately 1.618, then divide the result again by approximately 1.618. The result is a measure for the side of a smaller decagon to nest and repeat inside a larger. The process of subdividing or nesting decagons with five smaller decagons may be repeated infinitely. Following are diagrams of nesting decagons, and a fractal border.


Here's a completed girih tiling.

Girih tiling based on a pentagonal fractal

These girih tile drawing are loosely based on a fractal, a variant of the Koch snowflake, but pentagonal. The inside edges of the outer border of kite-shaped polygons follow a fractal. The original pentagon edges are broken three times to generate the final image. Pentagons may be used in place of the kites.

Each successive transformation of an edge into four edges was done by creating two new edges that are 72 degrees to the prior edge, and the lengths of the four new edges are 1/(2x(1+cosine(72))) of the prior edge.

Starting with a pentagon and using an angle of 72 degrees allows us to fill the interior with girih tiles. In the examples below I used a couple of scaling girih tiles.

Here is the girih line drawing I generated from the tiling. Yes, it has gaps.


Here are two more images following the same process.

More Scaling Girih Tiles

My scaling girih tiles project is inspired by a remarkable Islamic art patterning system that originated by the year 1200, and was rediscovered by Peter J. Lu [1] in 2005. Islamic artisans used girih [2], from the Persian word for “knot”, to develop intricate patterns from just five tiles decorated with lines. The lines rather than the tile edges become the pattern. I’ve extended the system for design purposes to allow for scale and density variations. This innovation would have been inappropriate before computers. The remarkable complexity of design in Islamic girih tiles doesn’t suggest the need for extension by multiple levels of self-similarity. Nevertheless, by expanding the system I hope to create patterns with the look and spirit of one facet of Islamic art while providing new design possibilities. The key is the simple addition of scaling tiles and a self-similar tile set.

By taking the basic girih tile set and extending it to include scaling tiles I’ve added a level of complexity that’s easily managed with digital media. With up to 32 tiles in my tile set, this would have been impractical and unnecessary for Islamic artisans. I decorate these extra tiles with lines (girih strap work) like the five girih tiles, but unlike the girih tiles the scaling tiles are not equilateral. These tiles also differ in that they may lack girih lines on some sides, but otherwise I preserve the angles of girih tiles. I generate multiple sets of the girih tiles, scaled to match the two different sides of each scaling tile. With the complete set I can create fractal-like drawings in which tile patches at different scales are similar.

Since this is art, not math nor historical architecture, I’m free to add tiles to the origianl girih tile set. Sometimes I add a second rhombus that is not in the basic set of five girih tiles. My designs are not all always tessellations – gaps and boundaries are OK. I choose to work with tilings that are always edge-to-edge, meaning adjacent tiles always share full sides. I often create symmetrical designs, but infinitely changing asymmetrical patterns are also possible. I’m searching for interesting designs whether or not they fill the plane. I’m not attempting to stay true to the historical methods.

It’s interesting though not surprising that in some of the tile sets in this project the long and short sides of the scaling tile are in the ratio of 1.61803…, the golden ratio.






References

1. Peter J. Lu and Paul J. Steinhardt (2007). "Decagonal and Quasi-crystalline Tilings in Medieval Islamic Architecture". http://peterlu.org/content/decagonal-and-quasicrystalline-tilings-medieval-islamic-architecture.

2. Wikipedia contributors, "Girih tiles," Wikipedia, The Free Encyclopedia,http://en.wikipedia.org/wiki/Girih_tiles(accessed November 11, 2010).

Asymmetrical Scaling Girih Tiling

Below are three renderings of an asymmetrical scaling girih tiling. This tiling demonstrates that the scaling girih tiles might be extended to fill the plane, and that the scaling could continue as well. In this case I have intentionally ended the boundary with pentagons and decagons at the same scale to indicate the degree of control possible. Depending on requirements, a scaling girih tiling could be made to create controlled patterns, fill a variety of forms, or control a range of line-shape densities.

Scaling Girih Tilings

This series of drawings is based on the discovery by Peter J. Lu [1] of girih tiles [2], a set of decorated tiles used by Islamic architects for centuries. I’ve taken the basic girih tile set and extended it to include scaling tiles. I decorate these extra tiles with lines (girih strapwork) like the five girih tiles, but unlike the girih tiles, the scaling tiles are not equilateral. The scaling tiles lack girih lines on two sides, but otherwise preserve the angles of girih tiles.


The scaling tile sides match the sides of two sets of girih tiles. I use multiple scaled sets of the tiles to generate fractal-like drawings in which tile patches at different scales are similar.

Since this is art, not math nor archeology, I’m free to add tiles, ignore symmetry, and disregard normal restrictions on tessellations. I’m searching for interesting designs whether or not they fill the plane. I’m not attempting to stay true to the historical methods.





References
1. Peter J. Lu and Paul J. Steinhardt (2007). "Decagonal and Quasi-crystalline Tilings in Medieval Islamic Architecture". http://peterlu.org/content/decagonal-and-quasicrystalline-tilings-medieval-islamic-architecture.
2. Wikipedia contributors, "Girih tiles," Wikipedia, The Free Encyclopedia, http://en.wikipedia.org/wiki/Girih_tiles (accessed November 11, 2010).

Pentagons and Isosceles Trapezoids

This dot pattern diagram from the series, Self-similar Boundaries, is based on pentagons and isosceles trapezoids. The trapezoids are the scaling polygons. They share non-parallel sides with a pentagon and parallel sides with scaled pentagons. The trapezoids generate one pentagon that scales down and one that scales up. The initial pentagon generates one trapezoid, with two scaled pentagons. Each subsequent iteration could double the number of pentagons, except that many are duplicates. In this example I follow three scaling iterations which might have produced 1 + 2 + 4 + 8 = 15 different pentagons, but since some are equal the pentagons produced are 10.

Pentagons and Scalene Triangles

This dot pattern diagram from the series, Self-similar Boundaries, is based on pentagons and scaling scalene triangles. Neither the design of the tile set nor the arrangement of tiles makes any attempt to create symmetry or even a tiling that fills the plane. Nevertheless, the tile set and tiling are interesting in that they can be used to generate tendrils that extend indefinitely. In this case the tile set scales down seven times.

Scaling Triangles

These two new dot patterns from the series, Self-similar Boundaries, are based on different tiles sets. They have in common triangles for scaling tiles. In the first example below the tile set is three rhombi and an isosceles triangle scaling tile. The second image tile set is a pentagon and two mirror image scalene triangle scaling tiles. I've included the tilings for the dot patterns.

Affine Transformations

This entry is about a series of drawings from tilings. It has a few simple math terms, but it's about art, not math. These drawings are dot pattern diagrams I call Self-similar Boundaries.

These dot patterns, structured with tilings dependent on affine transformations, reveal characteristic grids and alignments. An affine transformation is in this case a rotation or scaling followed by a translation. I'm applying the term to the placement of individual tiles, as opposed to patches of tiles. Each dot pattern hints at the design of the tile set because the patterns reveal all possible affine transformations from each tile, with all possible scaling, rotations, and translations predetermined by the tile set. The tile sets are simple polygons. The tilings or tile patches may or may not be symmetrical. They usually have gaps but no overlaps, are always edge-to-edge, and never fill the plane. Each possible tiling from a tile set is a collection of scaling, rotation, and translation of tiles in the set. There is an ordered selection of tile from the tile set, and an affine transformation of each tile to a position edge-to-edge with another tile.

The tile sets always include one or two scaling tiles with a side that matches the length of a polygon in the tile set and a shorter side that determines the affine transformation scaling. The scaling tile may be the only tile in the tile set, or it may be one of several tiles.

These tilings, or tile patches, are one solution to a problem I encountered while generating tilings as structures for a series of vertex pattern diagrams. I needed a way to create greater density changes than I was getting with other tilings. Self-similar tile sets solve the problem. Scaling tiles provide the mechanism.

The three boundary diagrams below are from the same tile set. They represent regular and irregular versions of tilings and dot patterns. The dot patterns reveal the fact that each tile set determines a limited number of possible affine transformations for each tile added. There are usually many possible ways to add a new tile, and the dots represent the vertices of all possible additions. But the design of the tile set limits the pattern of vertices to visible grids and alignments. These are just three examples of the dot patterns I generate from the tilings. You can view the tilings as well here.

Irregular Boundary Diagrams

This boundary diagram is based on a pentagon and three triangles. Two of the triangles are scalene, and the tile structure underneath is irregular, with gaps. The dot pattern transitions from regular to highly irregular. Nevertheless, from the regular bottom of the image to the irregular top the pattern retains some common horizontal and angular alignments.

This is one of several boundary diagrams that I've done combining tiles that would easily be regular tessellations with a transition triangle or two. Previously, I have designed self-similar scaling tile sets that suggest a regular if not gap-less tiling. In this case, the scalene triangles make the transition of pentagons and isosceles triangles from one scale to the next.

Here's a similar example, but with three rhombi and an isosceles triangle making the transition.

Boundary Diagrams

I've created a series of Self-Similar Boundary diagrams. I'm using original self-similar tile sets as the structure for these dot pattern diagrams. I begin the process of creating each diagram by designing a self-similar tile set. Then I create a tiling that emphasizes the transition from the largest tile to a potentially infinite boundary. Finally I generate vertices and extrinsic vertices for the tiling. The vertices or dot pattern recedes from open to closed at the boundary. I've included images of the dot pattern and the tiling structure for the vertices.

Examples of Self-Similar Tile Sets (Part 3)

These are examples of tilings from a variety of polygons. My purpose for making these is art, not math. These tilings, or tile patches, are one solution to a problem I encountered while generating tilings as structures for a series of vertex pattern diagrams. I needed a way to create greater density changes than I was getting with other tilings. Self-similar tile sets solve the problem. See Part 1 and Part 2 for detailed descriptions of two of these examples.

The tile set I describe in Part 1 is closely related to, and can be derived from tilings described by Robert W. Fathauer. Fathauer found two families of self-similar tilings based on segments of regular polygons [See the references below]. He went on to discover a great variety of fractal tilings that, as tilings, are more interesting than most of those I describe here. Unlike Fathauer’s discoveries the tilings below are either trivial, or they require more than one prototile. My method for developing a tiling is not necessarily straight forward or fixed. My purpose is not to generate interesting fractal diagrams, but to develop often asymmetrical structures for vertex pattern diagrams.

I've imposed a few restrictions on these tilings. That is, they should be edge-to-edge, with no overlaps. Tilings may not fill the plane, but should be infinitely scalable at the boundaries, as in a fractal. In some examples there might be gaps, but these gaps might be bounded by infinitely scaleable tiles forming singularities. These are actually tile patches, not tessellations. In most of these examples the tile set could be extended by adding other polygons, especially triangles.

1: A square and a triangle. The triangle is the s=4 case described by Fathauer, and I have retained the square for which the triangle is a segment. In most of these examples I'm showing one possible tiling that might extend indefinitely in a similar way.



2: A pentagon and a triangle. The triangle is the s=10 case described by Fathauer, and I have retained the pentagon that remains after removing five s=10 tiles from a regular decagon. This is the tile set I describe in Part 1.



3: A regular hexagon and two triangles. The isosceles triangle subdivides the equilateral triangle, which subdivides the hexagon. Therefore, a similar tiling could be accomplished with just an isosceles triangle.



4: Two trapezoids and two triangles. Other tilings could be made with just the two trapezoids or one trapezoid and a triangle.



5: A dart and a triangle. In this example the triangle is necessary to continue the scaling when the darts fold in on themselves. Darts alone without the triangles would overlap.



6: Another dart. Unlike the previous example, no triangle is necessary. When the darts fold in on themselves the gap can be bounded by infinitely scalable darts.



7: Another dart and triangle. A right isosceles triangle or square is necessary to fill a gap when the darts fold in on themselves.



8. A hexagon. This example is trivial, but the hexagon prototile can be combined with triangles that subdivide the hexagon to create interesting asymmetries. The hexagon fits inside a regular pentagon. This is the tile set I describe in Part 2.



9: A pentagon and two triangles. The pentagon fits inside a regular hexagon.



10: Golden Rectangles. The lengths of the two longest sides of each L-shaped hexagon are in the golden ratio.



References

Fathauer, Robert W. (2000). "Self-similar Tilings Based on Prototiles Constructed from Segments of Regular Polygons," presented at the Bridges Conference (July 28-30, 2000, Southwestern College, Winfield, Kansas).
http://www.mathartfun.com/shopsite_sc/store/html/Compendium/Bridges2000.pdf
Also see:
http://www.mathartfun.com/shopsite_sc/store/html/Compendium/encyclopedia.html

Self-similarity in Architecture

Most tessellations, especially in architecture, limit the available diversity within the chosen system. Part-to-whole relationships are singular when they could be scalable. Self-similarity solves the problem by allowing a potentially infinite range of parts within one whole. It's possible to design chaotic part-to-whole relationships that are flexible but impractical. Self-similarity restores the practicality while maintaining acceptable flexibility.



Other self-similar tile sets can be based on regular polygons including squares or hexagons. Here’s an animation using three scaled sets of three prototiles: http://joebartholomew.com/aniVertices_HSS9.html

Self-Similar Vertices, Part 2

This is the second of two methods that I use to create dot patterns with self-similar tile sets. My interest in tilings is relative to art, not math. These tilings, or tile patches, are one solution to a problem I encountered while generating tilings as structures for a series of vertex pattern diagrams. I needed a way to create greater density changes than I was getting with other tilings. Self-similar tile sets solve the problem.

The tile set described in Part 1 is closely related to, and can be derived from tilings described by Robert W. Fathauer. Fathauer found two families of self-similar tilings based on segments of regular polygons (see Reference below). One family includes an 18-18-144 triangle that is a segment of a regular decagon. The triangle is the s=10 prototile described by Fathauer. The tile set that I use includes a pentagon as well as the triangle. The pentagon is the shape that remains after removing five s=10 tiles from a regular decagon. The tile set I describe here in Part 2 is also self-similar, but doesn’t include Fathauer’s prototiles.

The method I describe here is based on an infinitely self-similar tile set consisting of a partially concave hexagon and two isosceles triangle prototiles. The hexagon is congruent to a regular pentagon except one side of the pentagon has been made concave with two smaller sides forming interior angles of 72, 252, and 72 degrees in addition to three 108 degree interior angles. The triangles are 72-72-36 and 36-36-108 isosceles triangles.

The initial hexagon and triangle prototiles are sized so the two long sides of the 72-72-36 triangle are equal to the long sides of the hexagon. The single long side of the 36-36-108 triangle is also equal to the long sides of the hexagon. The 36-36-108 triangle fits exactly in the concave area of the hexagon, and together the 36-36-108 triangle and the hexagon form a regular pentagon. The tile set in Part 1 was infinitely scalable pentagon-triangle pairs by setting the next pentagon side equal to the short side of the previous triangle. The tile set describe here is infinitely scalable by setting the next hexagon long sides equal to a short side of the previous pentagon.

Tilings should be edge-to-edge, with no overlaps. Gaps are always fillable. Singularities are not necessary. Tilings may not fill the plane, but could be infinitely scalable at the boundaries, as in a fractal. These are actually tile patches, not tessellations. A radially symmetrical tiling of just hexagons looks like the following figure.

The tilings are a structure for the finished diagram of vertices and what I call extrinsic vertices. This demonstrates the advantage of using a tiling with properties of self-similarity. The changing density of tiles translates to a greater variation in vertex density.

It's possible to create numerous symmetrical tilings with these tiles, but I often choose to create asymmetrical diagrams. The processes, lattices, and patterns I use are not math. I'm influenced by structures in math, science, architecture, and design, but unconstrained by the rigorousness of math. These diagrams have no practical use or purpose other than art.

Reference

Fathauer, Robert W. (2000). "Self-similar Tilings Based on Prototiles Constructed from Segments of Regular Polygons," presented at the Bridges Conference (July 28-30, 2000, Southwestern College, Winfield, Kansas).
http://www.mathartfun.com/shopsite_sc/store/html/Compendium/Bridges2000.pdf
Also see:
http://www.mathartfun.com/shopsite_sc/store/html/Compendium/encyclopedia.html

Self-Similar Vertices

I've been using the process described here to create tilings from a pentagon and triangle. These tilings, or tile patches, are one solution to a problem I encountered while generating tilings as structures for a series of vertex pattern diagrams. I needed a way to create greater density changes than I was getting with other tilings. Self-similar tile sets solve the problem.
My interest in tilings is more art than math. The tile set I’m using is closely related to, and can be derived from tilings described by Robert W. Fathauer (see Reference below). Fathauer found two families of self-similar tilings based on segments of regular polygons. One family includes an 18-18-144 triangle that is a segment of a regular decagon. The triangle is the s=10 prototile described by Fathauer. The tile set that I use includes a pentagon as well as the triangle. The pentagon is the shape that remains after removing five s=10 tiles from a regular decagon. Other hexagonal or square prototiles can be combined with triangles to make self-similar tile sets.

So, this method is based on an infinitely self-similar tile set consisting of pentagons and 18-18-144 triangle prototiles. Tilings should be edge-to-edge, with no overlaps. Gaps are inevitable, but they should allow lining with infinitely scaled tiles. Tilings will not fill the plane but should be infinitely scalable at the boundaries, as in a fractal. These are actually tile patches, not tessellations.

The initial pentagon and triangle prototiles are sized so the long side of the 18-18-144 triangle is equal to the pentagon side. Each subsequent pentagon-triangle pair is scaled so that the next pentagon side is equal to the short side of the previous triangle. Using this scheme, as the boundaries of the tiling grow outwards they form singularities, or gaps surrounded by tiles. The inside edges of these gaps can be continuously and infinitely lined with scaling pentagons and 18-18-144 triangles, or just 18-18-144 triangles.

An interesting feature of these prototiles is that the ratio of the areas of each pentagon to the next smaller pentagon (or triangle to triangle) is always 3.618... or 2 plus Phi.

It's possible to create numerous symmetrical tilings with these tiles, but I often choose to create asymmetrical diagrams. The processes, lattices, and patterns I use are not math. I'm influenced by structures in math, science, architecture, and design, but unconstrained by the rigorousness of math. These diagrams have no practical use or purpose other than art.

Reference:

Fathauer, Robert W. (2000). "Self-similar Tilings Based on Prototiles Constructed from Segments of Regular Polygons," presented at the Bridges Conference (July 28-30, 2000, Southwestern College, Winfield, Kansas).
http://www.mathartfun.com/shopsite_sc/store/html/Compendium/Bridges2000.pdf
Also see:
http://www.mathartfun.com/shopsite_sc/store/html/Compendium/encyclopedia.html