Pythia · 55 families

Tiling families

Shapes, original sources, and what we have checked. Review statuses describe the recorded scope and its limits. SS/S/A ranks rate artwork appearance separately.

Verified variants use explicit conventions where the originals leave choices open or conflict. Needs work includes partial reviews and known repairs.

Showing 55 of 55 families

12-fold Shuriken

Source-checked

Exact geometry is proven. Symmetric child frames follow the attributed Ginden reconstruction; the original raster cannot independently certify Paz’s decorated hierarchy.

Review scope: All eight dissections have exact oriented boundary cancellation. Every one of fourteen states is checked through two native generations, including ordered promotions and rotated/reflected frames.

Review evidence

Ammann-Beenker

Source-checked

Reconstructed the original three-state rhomb/triangle mesh and physical sizes, repaired the orange-parent child state, and corroborated all recursive rules in a third-generation original patch. All nine native root/pose cases pass three subdivisions.

Review scope: The original three-state rhomb/triangle source identity, exact physical sizes, source colors and ordered similarities; 169 declared original raster polygons with 157 holdouts. Original patch has 12 controls and 157 holdouts; bounded raster regions only.

Review evidence

CAST 3

Source-checked

Corrected physical prototype sizes and source identity. All original directed-edge rules and recursive native frames agree.

Review scope: All five original prototypes, fifteen child placements, sixty directed-edge marks, shared physical size ratios, and all fifteen native root/pose cases through three subdivisions. Complete original marked rules; later patch retained without registration.

Review evidence

CAST 5.1

Source-checked

Restored the large square and its promotion; original marked rules and later patch agree with native recursion.

Review scope: All three original states, missing square promotion restored, thirty rule children and thirty-eight corner marks; all nine native root/pose cases through four applications and at least two subdivisions. Separate original next-generation panel has fifty-seven held-out polygons and sixty-six corner marks.

Review evidence

Crown

Source-checked

Exact triangle and crown rules replace the dormant implementation, correcting child reflections that changed later generations.

Review scope: All 52 child frames; exact shape and partition checks. Native second-generation output matches 880 polygons from the crown root and 423 visible polygons from the triangle root in the original PDF. The separate source image resolves the triangle corner outside the PDF crop.

Review evidence

Danzer

Source-checked

Reconstructed the paper’s single marked 1-2-2 triangle and all nine direct child frames; removed an unsupported two-state refinement; confirmed three native recursive levels and an independently registered later original patch.

Review scope: One exact shape, inflation 3, direct recursive frames, all canonical/transformed/reflected native cases; bounded geometric agreement with a clipped original vector-patch region.

Review evidence

Danzer 7-fold Original

Source-checked

Corrected relative prototype sizes while preserving established ordered recursive shapes and frames. Every kind now uses one physical source scale.

Review scope: All 9 root/pose cases through 4 applications, 41505 descendant size checks, and old-size/prototype-only negative controls. Unchanged source registration has 59 holdouts with maximum error 0.385 pixels. Bounded source regions only.

Review evidence

Danzer 7-fold Variant

Source-checked

Corrected relative prototype sizes while preserving the previously repaired lime-triangle frames. Every kind now uses one physical source scale.

Review scope: All 9 root/pose cases through 4 applications, 6666 descendant size checks, and old-size/prototype-only negative controls. Unchanged source registration has 9 holdouts with maximum error 0.426 pixels and rejects all 7 alternative lime frames. Bounded source regions only.

Review evidence

Danzer Non-FLC 5

Verified variant

The five-state recursion matches 130 original atlas tiles under one fixed affine map. Every recursive frame is checked, and an exact fault-line proof establishes non-FLC.

Review scope: Verified 60-degree representative of the affine family. The original atlas does not specify an exact metric angle; this variant does not claim that angle. All five native roots pass canonical, transformed and reflected poses through 12 applications.

Review evidence

DTO C5

Source-checked

Source half-arrows correct 166 child orientations and two exchanged tile types. The existing first-step outlines are preserved.

Review scope: Fourteen independent rule occurrences and 2,056 arrow segments; 390 source polygons check a second-generation pentagon and all five promotions followed by subdivision. Every state passes native shape and partition checks. Coordinates remain numerical reconstructions.

Review evidence

Generalized Binary

Source-checked

Four exact rhomb rules match the source and its marked directions. The inflation multiplier is corrected without changing existing artwork geometry.

Review scope: All four n=9 rules; eight independent third-generation patch regions containing 436 marked tiles, including rotated and reflected native output. Whole children intentionally replace the straight parent boundary.

Review evidence

Girih-inspired 14-fold

Source-checked

Eleven exact rules reproduce the original vector figures. Whole tiles intentionally extend across the parent boundary.

Review scope: All 849 source children; every root through two native subdivisions with transformed and reflected frames. Four complete independent patch regions cover 124 tiles. Higher-generation neighbor patches are not certified duplicate-free.

Review evidence

Golden Triangle

Source-checked

Corrected the red-to-blue prototype size ratio; original altitude rule and recursive source patch agree with native output.

Review scope: All 2 roots in canonical, transformed and reflected native poses through 5 applications, with at least 2 substantive subdivisions on every branch. Original raster regions contain 13 polygon observations (13 unique components), 6 fixed controls and 7 registration holdouts. Bounded source regions only.

Review evidence

Hudson 7-fold

Source-checked

All six rules use exact shared-edge geometry from the original vector patch, removing the traced gaps and overlaps.

Review scope: Every root through three native generations, including exact shapes, partitions and reflected frames. Source comparisons cover six rule regions and an independent 148-tile second-generation heptagon; other recursive source markings remain outside that comparison.

Review evidence

Nischke-Danzer 6-fold-2

Source-checked

Four source states and twelve marked child frames independently reconstructed from the original rule and SVG metadata. All native roots and three poses pass three subdivisions and four complete source SVG regions pass held-out comparisons.

Review scope: Physical triangle sizes, marked source state assignments, recursive frames, and bounded original SVG agreement.

Review evidence

Nischke-Danzer-Deltoid 7-fold-2-2

Source-checked

All five source states and fifteen directed child frames independently reconstructed. Native recursion passes every root in three poses through three subdivisions and matches five complete original PDF regions with held-out polygons.

Review scope: Source rule arrows, exact pi/7 triangle ratios, recursive child states and frames; bounded agreement with the original PDF artwork.

Review evidence

Pinwheel

Source-checked

Correct relative prototype sizes and serialize independently derived ordered-similarity rules. Old bytes already match recursive geometry with separate root registration. The demonstrated defect is relative prototype size, not subdivision footprints.

Review scope: All 9 native root/pose cases through 3 applications. Geometry is certified after erasing cosmetic labels. Later-image orientation colors are not claimed to match the application palette refinement.

Review evidence

Pinwheel 1-2

Source-checked

The missing second cut in the large triangle is restored: nine children now follow the original two-step substitution.

Review scope: Both source states in canonical, transformed and reflected native poses through four applications; one original PDF depth-three supertile with 441 unique polygons, using 16 registration controls and 425 held-out polygons. Unmarked triangles use intrinsic corner frames. This checks a bounded PDF region, not the entire published patch.

Review evidence

Pinwheel 2-2

Source-checked

The exact Sadun Til(2) formula and per-child operators match the complete recursive rule and promotion. No further geometry change was needed.

Review scope: All 6 native root/pose cases through 4 applications and 1824 intrinsic frames. Exact mathematical rule verified from authored TeX and Figures 3, 9 and 10. The earlier Bielefeld JPEG registration remains rejected.

Review evidence

Pinwheel Cubic

Source-checked

Repaired relative prototype sizes. The original patch swaps the rule colors; correcting that interpretation resolves both recursive endpoint choices. Existing native corner frames already agree.

Review scope: All 18 native root/pose cases through 8 applications and 1122 polygon/parent checks. Two frozen source regions pass 144 interior and 432 edge probes at the unchanged 3-pixel tolerance; all three alternate conventions fail. Corroboration follows discovery, not a whole-patch certificate.

Review evidence

Pinwheel Square

Source-checked

Corrected the two triangle-born square reflections. Original patch color and edge probes distinguish all eight recursive conventions; repaired native descendants follow the selected source rule.

Review scope: All 6 root/pose cases through 4 substantive applications. A frozen 125-tile region has 416 interior and 1248 boundary probes at the unchanged 3-pixel tolerance. Corroboration follows discovery; this is a bounded source-region check.

Review evidence

Pythia 3.1

Source-checked

Correct relative prototype sizes and serialize independently derived ordered-similarity rules. Old bytes already match recursive geometry with separate root registration. The demonstrated defect is relative prototype size, not subdivision footprints.

Review scope: All 9 native root/pose cases through 3 applications. Original PDF validation covers one complete depth-three root-1 supertile. The primary-paper construction establishes all three size states; native tests cover all roots and poses.

Review evidence

Pythia 5.1

Source-checked

Correct relative prototype sizes and serialize independently derived ordered-similarity rules. Old bytes already match recursive geometry with separate root registration. The demonstrated defect is relative prototype size, not subdivision footprints. Correct the Pythagoras-5-1 source URL to Pythia-5-1.

Review scope: All 15 native root/pose cases through 3 applications. Original PDF validation covers one complete depth-three root-1 supertile. The primary-paper construction establishes all five size states; native tests cover all roots and poses.

Review evidence

Robinson

Source-checked

Shared physical prototype sizes corrected; exact colored source rules and ordered recursive placements match the original patch regions.

Review scope: All four roots in canonical, transformed and reflected native poses through four subdivisions. Four original raster regions contain 178 polygons, 48 fixed controls and 130 held-out polygons. This checks declared raster regions, not the whole published patch.

Review evidence

Sadun

Source-checked

Correct relative prototype sizes and serialize independently derived ordered-similarity rules. Old bytes already match recursive geometry with separate root registration. The demonstrated defect is relative prototype size, not subdivision footprints.

Review scope: All 12 native root/pose cases through 3 applications. Original PDF validation covers one complete depth-three δ supertile. Primary-paper Figure 1 establishes all four rules; native tests cover all roots and poses.

Review evidence

Say-Awen 17-fold

Verified variant

Repaired two T6 child frames to match all 862 visible descendants in the registered atlas region. The two unspecified T4 frames use an explicit half-turn convention.

Review scope: Completed atlas-patch variant with exact 17-fold rotational and non-FLC certificates. The source has contradictory T5/T6 rules and corner caps; the selected convention is not claimed as a unique authored answer. All 24 native root/pose cases pass two subdivisions.

Review evidence

Say-Awen 21-fold

Verified variant

The marked Figure 5 construction is complete with nine explicit apex conventions. Independent exact certificates establish 21-fold rotational symmetry and non-FLC.

Review scope: Verified marked-rule variant. Original local descendants reject the frame reversal required by Table 7; the whole Figure 7 hierarchy is not claimed. Unspecified apices remain source conventions, not recovered author marks. All 30 native root/pose cases pass two subdivisions.

Review evidence

Self-Tiling Tile Set 2 (Sallows)

Source-checked

Five corrected recursive child frames now match the original corner marks and later PDF generations.

Review scope: All four source rule panels and marked recursive frames; all four roots in canonical, transformed and reflected poses through three subdivisions; independent original PDF depth-three regions and adjacent depth-four supertile, 256 unique polygons with separate registration controls. Five child-frame corrections preserve all first-step footprints.

Review evidence

Socolar 7-fold

Source-checked

Exact shared edges preserve all 129 source placements. Two independent patch regions confirm the half-turn where the rule sheet disagrees.

Review scope: All roots through two native generations, including ordered and reflected frames and neighboring clusters. Two held-out 46-tile regions resolve the disputed corner; complete whole-patch compatibility is not certified.

Review evidence

Squeeze

Source-checked

The 187 by 161 original raster supplies all rule states, shared vertices, and unique geometric corner frames. The exact similar-triangle derivation extends the diagram to its stated free parameter. The retained 1200 by 1200 later patch has not been independently registered; the verdict covers the complete rule construction.

Review scope: All three original states and complete free-parameter rule construction, including fixed original raster corners and geometric recursive frames. This right-angle-at-apex parameter matches all nine native root/pose cases through four applications and at least two subdivisions. Later patch retained without registration.

Review evidence

Squeeze (Bielefeld 2-shape)

Source-checked

The 187 by 161 original raster supplies all rule states, shared vertices, and unique geometric corner frames. The exact similar-triangle derivation extends the diagram to its stated free parameter. The retained 1200 by 1200 later patch has not been independently registered; the verdict covers the complete rule construction.

Review scope: All three original states and complete original right-angle parameter rule; physical middle-state size corrected. Fixed original raster corners and unique geometric recursive frames agree with all nine native root/pose cases through four applications and at least two subdivisions. Later patch retained without registration.

Review evidence

Squeeze (sharp)

Source-checked

Identified as Frettloh Pythagoras sigma_3,2. Existing triangles, relative sizes, altitude split and promotions match the primary construction.

Review scope: Complete primary mathematical rule and intrinsic frames; 9 native root/pose cases through 6 applications. Squeeze Sharp is a repository alias, distinct from Bielefeld Squeeze and the diagonally flipped Pythia rules. No later-patch registration is claimed.

Review evidence

Stampfli 12-fold

Source-checked

The 16 and 19 unique source children retain full SVG precision and ordered recursive frames, including intentional boundary spill.

Review scope: Both roots through two native generations, reflected frames and neighboring clusters; source SVG placement chains agree through four applications. Source color reproduction and arbitrary depth are not certified.

Review evidence

Sub Rosa 7-fold

Source-checked

All three n=7 rules follow the original paper, including its clockwise ownership of boundary rhombs.

Review scope: All 1,064 source children; every native root through two generations in rotated and reflected poses. The three adjacent source parents produce 412,128 second-generation rhombs with no duplicate footprints or positive-area overlaps.

Review evidence

Tangram

Source-checked

Restored the original five states and both promotions. Exhaustive original-patch frame search resolves all placements up to an equivalent square coordinate convention.

Review scope: All 5 roots in canonical, transformed and reflected native poses through 6 applications, with at least 2 substantive subdivisions on every branch. Original raster regions contain 65 polygon observations (65 unique components), 8 fixed controls and 57 registration holdouts. Bounded source regions only. An exhaustive 8192-candidate frame search checks all 69 source polygons and proves the two surviving square frames equivalent. Two merged green-pair outlines are checked separately.

Review evidence

Thue-Morse 2D

Source-checked

Both source rules and the complete published cell-state pattern agree with the implementation.

Review scope: Both source states and rule panels; canonical, transformed and reflected native cases through three applications; exact original 4096-cell depth-six patch pattern with 4094 held-out cells. Published SVG has nonuniform offsets and 512 same-color overdraws; no exact shared-edge artwork claim.

Review evidence

Trapezotriangular

Source-checked

Corrected the trapezoid-to-triangle physical size ratio and source URL. Both source rules and recursive patch regions agree with native output.

Review scope: All 2 roots in canonical, transformed and reflected native poses through 3 applications, with at least 3 substantive subdivisions on every branch. Original raster regions contain 128 polygon observations (128 unique components), 16 fixed controls and 112 registration holdouts. Bounded source regions only.

Review evidence

Triangle Duo

Source-checked

Corrected the acute child marked base corner in the scalene rule. The original marked rule and later source regions agree with native output.

Review scope: All 2 roots in canonical, transformed and reflected native poses through 5 applications, with at least 5 substantive subdivisions on every branch. Original raster regions contain 64 polygon observations (48 unique components), 14 fixed controls and 50 registration holdouts. Bounded source regions only.

Review evidence

Tuebingen

Source-checked

Shared physical prototype sizes corrected; exact colored source rules and ordered recursive placements match the original patch regions.

Review scope: All four roots in canonical, transformed and reflected native poses through four subdivisions. Four original raster regions contain 178 polygons, 48 fixed controls and 130 held-out polygons. This checks declared raster regions, not the whole published patch.

Review evidence

Uberpinwheel

Source-checked

Repaired diagonals and colors match the source inside the exact triangular parents. One extra triangle outside the parent in the encyclopedia raster is omitted, following the original paper’s boundary.

Review scope: All 260 in-parent source children across four marked rules; every root through two native generations with transformed and reflected shape checks. Published patch placement and arbitrary recursion depth are not certified.

Review evidence

Watanabe-Ito-Soma 12-fold

Source-checked

The corrected January 2023 source rules and their corner marks are reproduced. Children intentionally extend beyond the straight parent.

Review scope: All three rules through three native generations, ordered transformed/reflected frames and neighboring patches. A fourth-generation square matches all 69,446 source polygons and their marked order.

Review evidence

Westend

Source-checked

The original executable FAM rules agree through six applications, including promotions and ordered child frames.

Review scope: All three original FAM prototypes and six child placements, corroborated by 30 original rule-image interior samples; all roots and canonical, transformed and reflected poses through six applications and at least two substantive subdivisions, including ordered recursive frames. Patch raster retained without a registration claim.

Review evidence