VOYAGE TWELVE · FROM SHAPE TO RELATION

When Distance Disappeared — Bridges, Holes, Knots, and the Shape of Data

Walk 269 years from a city map folded into four dots and seven lines. Follow how discarding length and angle proved an impossibility, fingerprinted surfaces and knots, and eventually tracked long-lived loops in clouds of data.

QUESTION FOR THE ROUTE

To recognize shape, what must be discarded, which of connection, orientation, dimension, and holes must remain, and how far can any invariant certify sameness?

WHAT THE LINE DOES NOT CLAIM

This line is not a direct transmission chain from one Königsberg school to Stanford. It is an edited route separating problem setting, writing, publication, seminars, independent discovery, and later computation; it does not equate graph theory with topology, homeomorphism with smooth equivalence, or persistence with meaning.

The camera rests on each city while you read, then eases through the runway between scenes. Select any marker or scene link to travel in either direction.

The same route, four questions

A lens never hides a scene or proves a cause. It changes which places you compare first, and the URL preserves your choice.

Whole route Translation networks Commercial demand Observation and institutions Patronage and power

The whole-route view keeps problem, cognitive change, place, movement, and evidence boundary at equal weight. Choose a lens when you want to test a different explanation against the same scenes.

Live map · 지도를 불러오는 중…

08 / 18 · 1895 CE

Paris

01 · 1736 CE
Königsberg · Other
Folding a City Map into Four Dots and Seven Lines — The Bridge Problem
Can a walk cross each of the seven bridges joining the two Pregel islands and riverbanks exactly once? Euler's decisive choice was not to calculate lengths, bridge shapes, or island areas. Turning the four land regions into vertices and bridges into edges erased almost everything about the city but preserved what connected to what. This pin marks the problem's setting; it is not evidence that Euler wrote there or visited in 1736.
PAUSE AND ASK
Can the walk be decided after erasing every length and geometric shape from the city?
How the idea changed
Compress a metric city map into a graph by replacing land regions with vertices and bridges with edges.
What this place made possible
The Pregel's islands and branches made a once-per-bridge walk a natural city puzzle.
How it moved
Local walking puzzle and map → problem reaching Euler → abstract model of four land regions and seven bridges
Do not overclaim
1736 is the conventional problem date; the pin marks the setting, not proof that Euler visited or wrote there.
Evidence sources
- Euler Archive — Solutio problematis ad geometriam situs pertinentis (E53)
  Supports: The four-region, seven-bridge abstraction and bibliography distinguishing 1735 writing from 1741 publication
Stable link to this scene
KönigsbergSaint Petersburg
02 · 1741 CE
Saint Petersburg · Publication
Proving a Walk Impossible without Trying Every Walk — Odd Degree
At a vertex with odd degree, bridges used for intermediate arrivals and departures cannot all be paired. All four Königsberg vertices have odd degree, while an open one-stroke trail can exempt only its start and end. Euler wrote the paper in 1735; it appeared in the Saint Petersburg Academy proceedings in 1741. It is an early foundation of graph theory and a precursor of topological thought, not the completed birth of all modern topology.
PAUSE AND ASK
How can vertex counts prove impossibility without trying every route?
How the idea changed
Turn the pairing of arrival and departure at intermediate visits into a parity condition on degree.
What this place made possible
The Saint Petersburg Academy proceedings made a distant city puzzle public as a reusable proof.
How it moved
Königsberg problem → Euler's 1735 paper → academy presentation and editing → 1741 proceedings
Do not overclaim
It is an early foundation of graph theory and precursor of topology, not the single birth of all modern topology.
Evidence sources
- Euler Archive — E53 bibliographic record and original text
  Supports: The original bridge argument based on incidences and vertex degree
Stable link to this scene
Saint PetersburgBerlin
03 · 1750 CE
Berlin · Composition
Finding a Count That Survives Distortion — V−E+F
Working in Berlin, Euler organized the observation that a polyhedron's vertices V, edges E, and faces F satisfy `V−E+F=2`. Rounding a cube does not destroy the relation among the three counts. Yet the value is not unconditionally two for every surface. Later work had to state conditions on sphere-like closed surfaces and cell decompositions, then explain how the value changes with holes, before the observation became a general invariant.
PAUSE AND ASK
What relation among vertices, edges, and faces survives when a polyhedron is rounded and distorted?
How the idea changed
Replace individual lengths and angles with the global fingerprint `V−E+F`.
What this place made possible
Euler organized the polyhedron problem while working at the Berlin Academy, though publication later occurred in a Saint Petersburg journal.
How it moved
Polyhedron calculations → 1750 Berlin writing → 1758 Saint Petersburg publication → later generalization to surfaces
Do not overclaim
`χ=2` is not an unconditional law of every space; it begins with suitable decompositions of sphere-like closed surfaces.
Evidence sources
- Euler Archive — Elementa doctrinae solidorum (E230)
  Supports: The 1750 writing, 1758 publication, and relation among polyhedral vertices, edges, and faces
Stable link to this scene
BerlinGöttingen
04 · 1848 CE
Göttingen · Publication
Studying Relations of Position instead of Measuring Position — Listing's Topologie
In his Göttingen publication Vorstudien zur Topologie, Johann Benedict Listing joined the Greek topos and logos to organize a study of relations of position. What remains under deformation without cutting or gluing became an independent question. Euler's bridges, Gauss's surfaces, and Listing's name connect historically, but not as a single straight line to today's definitions. Naming the field did not finish its axioms and tools overnight.
PAUSE AND ASK
How did a study of relations of position without measurement gain an independent name?
How the idea changed
Organize adjacency, enclosure, boundary, and orientation as objects distinct from metric geometry.
What this place made possible
Göttingen's geodesy, geometry, and university press network gave Listing a setting to publish *Topologie* as a term and program.
How it moved
Surface and geodesy questions around Gauss → Listing's research and teaching → 1848 publication of *Vorstudien zur Topologie*
Do not overclaim
Publishing the term did not mean that modern topological-space axioms or algebraic topology were already complete.
Evidence sources
- Göttinger Digitalisierungszentrum — Vorstudien zur Topologie
  Supports: The Göttingen edition, 1848 bibliography, and Listing's use of Topologie
Stable link to this scene
GöttingenLeipzig
05 · 1858 CE
Leipzig · Discovery
Following the Front and Returning without a Back — Möbius's Band
August Ferdinand Möbius studied what happens when a strip receives a half twist before its ends are joined: inside and outside no longer split into two sides. A pencil following the centerline reaches what looked like the reverse before returning. Direction and boundary, not length, become essential. Listing found the same structure independently, so the familiar eponym does not give one person sole ownership of the discovery.
PAUSE AND ASK
Why does the familiar separation of front and back fail after a strip receives a half twist?
How the idea changed
Treat whether orientation survives a complete journey as more essential than a surface's length or shape.
What this place made possible
Within Leipzig's university work in astronomy and geometry, Möbius studied the one-sided twisted strip in 1858.
How it moved
Questions of orientation in surfaces and polyhedra → Möbius's 1858 manuscript → later publication, models, and eponym
Do not overclaim
Listing found the structure independently; the name Möbius strip does not prove sole discovery.
Evidence sources
- MacTutor — August Ferdinand Möbius
  Supports: The 1858 one-sided-strip work, Leipzig setting, and chronology of independent discovery with Listing
Stable link to this scene
LeipzigGöttingen
06 · 1861 CE
Göttingen · Publication
Recording a One-Sided Surface Independently — Listing's Twisted Ring
Listing independently treated twisted rings and direction-reversing surfaces in Göttingen and published his work in 1861. Near-simultaneous discoveries suggest that the strange paper object arose naturally from contemporary questions about surfaces, joining, and orientation. In modern terms, non-orientability means a local orientation cannot be preserved consistently around every loop. The Euler characteristic of a Möbius band alone does not determine orientability.
PAUSE AND ASK
How does the discovery story change when two people find nearly the same strange surface independently?
How the idea changed
Turn one-sided surprise into the general question of non-orientability and read independent discovery as evidence of a shared problem environment.
What this place made possible
Göttingen's surface research, teaching, and publication environment preserved Listing's twisted-ring analysis for comparison.
How it moved
Listing's independent work → 1861 publication → later comparison with Möbius → stabilization of orientability as a concept
Do not overclaim
Euler characteristic, boundary count, and orientability interact, but one number does not determine every other property.
Evidence sources
- MacTutor — Johann Benedict Listing
  Supports: Listing's topology terminology, independent one-sided-surface work, and 1861 publication
Stable link to this scene
GöttingenEdinburgh
07 · 1877 CE
Edinburgh · Publication
A Classification Table That Outlived a Failed Atomic Model — Tait's Knots
Kelvin proposed atoms as vortex knots in an ether, and Edinburgh's Peter Guthrie Tait systematically drew and classified knots for the project. The vortex-atom model disappeared, while the question 'can two loops be changed into one another without cutting?' survived. A failed scientific theory can demand precise tables and operations that become independent mathematics. Tait's tables were a major beginning, not a complete or error-free classification of every knot.
PAUSE AND ASK
Why did knot tables survive as mathematics after the atomic model that demanded them failed?
How the idea changed
Turn diagrams built as candidate atoms into an independent classification of closed curves under deformation without cutting.
What this place made possible
Edinburgh's university and Royal Society network let Tait combine calculation, experiment, and tables in a public knot catalog.
How it moved
Helmholtz's vortices → Kelvin's atomic hypothesis → Tait's Edinburgh knot tables → independent knot theory
Do not overclaim
Failure of vortex atoms was not failure of knot theory, and Tait's table was neither complete nor error-free for all knots.
Evidence sources
- Royal Society of Edinburgh — P. G. Tait, On Knots (1877)
  Supports: Tait's 1877 Royal Society of Edinburgh paper, knot tables, and vortex-atom context
Stable link to this scene
EdinburghParis
08 · 1895 CE
Paris · Publication
Translating Holes into Algebra Richer than a Count — Analysis Situs
Henri Poincaré's Analysis Situs tried to compare cycles and boundaries in a space algebraically. Some closed paths contract to a point; others catch on a hole. Connecting that difference to Betti numbers and the language that became homology marked a turn toward algebraic topology. The paper itself contained errors corrected in later supplements, and today's definitions by groups and chain complexes did not arrive fully formed in one publication.
PAUSE AND ASK
How can holes in a picture become objects of calculation rather than visual guesswork?
How the idea changed
Translate relations among cycles and boundaries into numbers and algebraic structure that survive continuous deformation.
What this place made possible
Parisian lectures, journals, and mathematical-physics networks let Poincaré join cycle problems from several fields in one language.
How it moved
Complex functions, differential equations, and surfaces → 1895 *Analysis Situs* → supplements and corrections → formal homology and fundamental groups
Do not overclaim
Poincaré's turn was decisive, but modern chain complexes and groups were not finished in one paper and had precedents.
Evidence sources
- Numdam — Henri Poincaré's Analysis Situs papers
  Supports: Bibliography of the 1895 paper and supplements, their corrections, and the formation of algebraic topology
Stable link to this scene
ParisParis
09 · 1904 CE
Paris · Composition
Could Loops Identify the Three-Dimensional Space We Inhabit? — The Poincaré Conjecture
Poincaré asked whether every closed three-manifold in which every loop contracts to a point must be the three-sphere. Classification by holes is familiar for two-dimensional surfaces but became startlingly difficult in three dimensions. The scene follows his Paris work on the 1904 supplement while noting that the journal was published in Palermo. The beginner's phrase 'no holes' cannot replace the exact conditions of simple connectivity and a closed 3-manifold.
PAUSE AND ASK
Can a three-sphere be recognized solely because every loop in the space contracts to a point?
How the idea changed
Extend surface classification into a recognition problem for closed simply connected three-manifolds.
What this place made possible
Working in Paris, Poincaré wrote supplements addressing errors and exceptions; the paper appeared in a Palermo journal.
How it moved
Supplements to *Analysis Situs* → 1904 question → dimension-specific solutions → Hamilton's Ricci flow and Perelman
Do not overclaim
The phrase 'no holes' omits simple connectivity and the closed 3-manifold condition; the journal publication site was Palermo, not Paris.
Evidence sources
- Numdam — Poincaré's fifth supplement and the conjecture bibliography
  Supports: The 1904 fifth supplement, Palermo publication, and the conjecture's exact conditions
Stable link to this scene
ParisAmsterdam
10 · 1911 CE
Amsterdam · Main activity
Folding a Plane toward a Line without Making Dimension Disappear — Brouwer
Brouwer completed work showing that open spaces of different dimensions cannot be topologically equivalent. A curve can fill a square densely or even continuously cover it, but one dimension does not become two when a continuous one-to-one map with a continuous inverse is required. The theorem made intuitive coordinate count into a topological fact. It belongs to the same 1911 period as Brouwer's fixed-point theorem but is not the same result.
PAUSE AND ASK
If a line is folded intricately enough to fill a plane, do one and two dimensions become the same?
How the idea changed
Distinguish a continuous space-filling image from a homeomorphism with continuous inverse, establishing dimension as an invariant.
What this place made possible
Brouwer's University of Amsterdam research and teaching environment supported rigorous topology amid set-theoretic and geometric puzzles.
How it moved
Dimension-challenging examples of Cantor and Peano → Brouwer's topological tools → 1911 invariance proof → manifold dimension theory
Do not overclaim
Invariance of dimension and the contemporary fixed-point theorem are related early-topology results, not one theorem.
Evidence sources
- University of Amsterdam — L. E. J. Brouwer
  Supports: Brouwer's Amsterdam setting and the distinction between 1911 invariance and fixed-point work
Stable link to this scene
AmsterdamLeipzig
11 · 1914 CE
Leipzig · Publication
Discarding Even the Rubber Sheet and Keeping Rules of Nearness — Hausdorff
Felix Hausdorff's Grundzüge der Mengenlehre organized very general spaces through neighborhoods and open sets. The condition that two distinct points have disjoint neighborhoods explains why limits are unique in well-behaved spaces and where that behavior can fail. The Leipzig pin marks the book's 1914 publication, not Hausdorff's workplace at the time. Topology could now build spaces from relational rules without requiring a physical rubber sheet.
PAUSE AND ASK
Can a space be built from rules of nearness without drawing a physical surface?
How the idea changed
Replace distance formulas with neighborhoods, open sets, and separation axioms that include function spaces.
What this place made possible
A Leipzig publisher circulated Hausdorff's 1914 set-theory book; the pin does not mark his workplace at the time.
How it moved
Set theory, function theory, and point-set work → publication of *Grundzüge der Mengenlehre* → standard topological spaces and separation axioms
Do not overclaim
A Hausdorff space is not the definition of every topological space but one additional condition separating distinct points.
Evidence sources
- Universität Leipzig — Hausdorff's Grundzüge der Mengenlehre
  Supports: The 1914 Leipzig publication and the influence of neighborhoods and axiomatic spaces on topology
Stable link to this scene
LeipzigGöttingen
12 · c. 1925 CE
Göttingen · Presentation
From Numbers That Count Holes to Groups That Transform — Noether's Seminar Network
In 1920s Göttingen, seminars and conversations involving Emmy Noether and Pavel Alexandroff pushed topology from numerical Betti counts toward organizing cycles and boundaries as groups and maps. Groups preserve not just a count but structure, transformations, and relations between dimensions. That shift became modern algebraic topology's grammar. Because the influence was substantially oral, 1925 is an approximate anchor rather than a claim for one dated paper or sole inventor.
PAUSE AND ASK
What becomes visible when holes are organized by addition and maps rather than merely counted?
How the idea changed
Lift Betti counts into groups and maps of cycles and boundaries, creating a structural language for topological change.
What this place made possible
Göttingen's small seminars and algebra community let Noether and Alexandroff reshape the language before formal publication.
How it moved
Poincaré's Betti numbers and torsion → Göttingen seminars and conversations → group-and-map homology → Eilenberg–Steenrod axioms
Do not overclaim
Because the influence was substantially oral and communal, 1925 is approximate and not a sole-inventor claim tied to one paper.
Evidence sources
- American Mathematical Society — Emmy Noether and algebraic topology
  Supports: The shift from Betti numbers to groups and the influence of Noether–Alexandroff seminars in Göttingen
- Universität Göttingen — Emmy Noether
  Supports: Noether's Göttingen teaching, research, seminar activity, and institutional context
Stable link to this scene
GöttingenPrinceton
13 · 1928 CE
Princeton · Publication
Giving a Knot a Polynomial Fingerprint — The Alexander Polynomial
James Alexander compressed algebra from a knot's complement into a Laurent polynomial. Equivalent knots keep the same value as their diagrams change, so the invariant distinguishes many knots quickly. It turned visual comparison into comparison by calculation. Different knots can nevertheless share an Alexander polynomial, so this fingerprint is not a complete identity document.
PAUSE AND ASK
Which differences become calculable when a complicated knot diagram is compressed into one polynomial?
How the idea changed
Convert relations in a knot complement into a Laurent polynomial that survives diagram changes as an algebraic fingerprint.
What this place made possible
Princeton's university and journal ecosystem let a new topological invariant circulate as calculation rather than a table of pictures.
How it moved
Tait's tables → knot groups and coverings → Alexander's 1928 polynomial → later skein and homological invariants
Do not overclaim
Equivalent knots share the polynomial, but distinct knots can also share it, so it is not a complete classifier.
Evidence sources
- American Mathematical Society — Alexander, Topological invariants of knots and links
  Supports: Bibliography of the 1928 Transactions paper and its topological invariant of knots and links
Stable link to this scene
PrincetonPrinceton
14 · 1956 CE
Princeton · Publication
The Same Sphere Where Differentiation Works Differently — Milnor's Exotic 7-Sphere
John Milnor found manifolds homeomorphic to the standard seven-dimensional sphere but not equivalent through smooth coordinate changes. Preserving differentiable structure is stricter than preserving topology without tearing or gluing. Invariants distinguish kinds of 'sameness' even in seven dimensions that cannot be drawn. The Princeton pin marks the Annals of Mathematics publication and research network rather than one room and instant of discovery.
PAUSE AND ASK
What does it mean for two spheres to be topologically the same yet carry inequivalent smooth coordinates?
How the idea changed
Separate homeomorphism from diffeomorphism and make smooth structures atop a topological shape into objects of classification.
What this place made possible
Princeton's *Annals of Mathematics* and manifold community spread a seven-dimensional example into a central classification problem.
How it moved
Fiber bundles and characteristic classes → Milnor's 7-sphere construction → 1956 Annals publication → classification of exotic smooth structures
Do not overclaim
Exotic does not mean a different topological sphere; it means a different smooth structure on the same topological sphere.
Evidence sources
- American Mathematical Society — Milnor's exotic spheres
  Supports: Historical explanation of the 1956 seven-sphere paper and the distinction between homeomorphic and diffeomorphic
Stable link to this scene
PrincetonBerkeley
15 · 1961 CE
Berkeley · Main activity
Taking High-Dimensional Spaces Apart by Handles — Smale
Stephen Smale used Morse theory, critical points, and handle attachment to solve the generalized Poincaré conjecture in dimensions greater than four. Higher dimensions can paradoxically provide more room to move pieces past one another. The success did not descend automatically to dimensions four and three. Different dimensions admit different techniques and smooth structures, splitting a century-old problem into distinct branches.
PAUSE AND ASK
Why could the Poincaré problem be solved in high dimensions before the lower ones?
How the idea changed
Assemble and disassemble manifolds through critical points and handles, tracking the room available in each dimension.
What this place made possible
In Berkeley's 1960–1961 geometry and dynamics environment, Smale applied Morse theory to high-dimensional topology.
How it moved
Morse theory → cobordism and handles → Smale's dimensions-above-four proof → separate four- and three-dimensional routes
Do not overclaim
The 1961 result concerns dimensions above four and does not automatically include the four- or three-dimensional theorems.
Evidence sources
- UC Berkeley — Stephen Smale bibliography
  Supports: The 1961 generalized Poincaré paper and chronology of Smale's Berkeley appointment
Stable link to this scene
BerkeleyBerkeley
16 · 1984 CE
Berkeley · Discovery
A New Knot Fingerprint Emerging from Algebra outside Knot Theory — Jones
While studying subfactors of von Neumann algebras and braid representations, Vaughan Jones unexpectedly found a new knot polynomial. It separates many knots missed by the Alexander polynomial and can sometimes distinguish a knot from its mirror image. A calculation in one field became an invariant in another. It still does not distinguish every knot, and the quantum-field interpretation developed by Witten and others was a later step.
PAUSE AND ASK
Why did a new knot fingerprint emerge from calculations in operator algebras that seemed unrelated to knots?
How the idea changed
Translate subfactors and braid representations into knot invariants, exploiting structural correspondence between distant fields.
What this place made possible
Berkeley's operator-algebra and topology networks gave Jones a setting to compare the unexpected polynomial rapidly with knot theory.
How it moved
Von Neumann subfactors → braid representations → 1984 Jones invariant → 1985 publication → quantum topology and Khovanov homology
Do not overclaim
The Jones polynomial does not distinguish every knot, and Witten's quantum-field interpretation was a later development.
Evidence sources
- American Mathematical Society — A polynomial invariant for knots via von Neumann algebras
  Supports: The operator-algebra origin of the Jones polynomial and its 1984 discovery and 1985 publication
Stable link to this scene
BerkeleySaint Petersburg
17 · 2003 CE
Saint Petersburg · Main activity
Letting Space Flow until Its Shape Appears — Perelman's Three Preprints
In three arXiv preprints from 2002–2003, Grigori Perelman added entropy, non-collapsing control, and surgery arguments to Richard Hamilton's Ricci-flow program. Flowing an irregular metric like heat while controlling singularities reveals the geometric pieces of a three-manifold. Years of international verification led to acceptance of the geometrization and Poincaré results. A lone-genius story must not erase the prior program or the labor of checking the proof.
PAUSE AND ASK
How can flowing an irregular metric like heat reveal the underlying topology of a space?
How the idea changed
Reveal geometric pieces through the time evolution and singularity control of Ricci flow rather than static classification alone.
What this place made possible
The Steklov Institute geometry community and international correspondence and preprint networks supported Perelman's extension of Ricci flow.
How it moved
Hamilton's Ricci flow → Perelman's three 2002–2003 arXiv preprints → global seminars, elaboration, and verification → accepted resolution
Do not overclaim
The three preprints were the breakthrough but should not erase Hamilton's program or the community's later detailed verification.
Evidence sources
- arXiv — The entropy formula for the Ricci flow and its geometric applications
  Supports: Perelman's first 2002 preprint and its entropy and non-collapsing tools for Hamilton's Ricci flow
- arXiv — Ricci flow with surgery on three-manifolds
  Supports: The second 2003 preprint on surgery for three-manifolds and the geometrization program
- arXiv — Finite extinction time for the solutions to the Ricci flow
  Supports: The third preprint and the chronology of the three-paper resolution
Stable link to this scene
Saint PetersburgStanford
18 · 2005 CE
Stanford · Publication
Tracking Holes That Survive in a Cloud of Points — Persistent Homology
Afra Zomorodian and Gunnar Carlsson developed an algorithmic framework for persistent homology, following components and loops as the connection radius among data points grows. Barcodes compare short-lived features with those surviving across many scales. Topology became a way to read the shape of data. But the representation, metric, and filtration are choices that shape the result, and a persistent loop is not automatically a meaningful cause or truth.
PAUSE AND ASK
Given only scattered points, how can components and loops that outlast noise be compared across scales?
How the idea changed
Record births and deaths through a whole filtration as barcodes and persistence modules rather than fixing shape at one threshold.
What this place made possible
Stanford's intersection of computational geometry, algebraic topology, and data research made persistent homology algorithmic.
How it moved
Homology and Morse theory → complexes across scales → Zomorodian–Carlsson algorithms → TDA software and applications
Do not overclaim
TDA does not discard distance entirely: representation, metric, and filtration choices matter, and persistence does not guarantee meaning or causation.
Evidence sources
- Stanford Computational Geometry — Computing Persistent Homology
  Supports: The 2005 Zomorodian–Carlsson algorithm and computation of persistence modules and barcodes
- American Mathematical Society — Gunnar Carlsson, Topology and Data
  Supports: An overview of topological data analysis, including representation, scale choices, and application boundaries
Stable link to this scene

PAUSE THE FILM · BEND THE MAP

Change the drawing. Keep the connection.

Stretch the same bridge network, remodel one or two bridges, then climb from Euler trails to surfaces and a cloud of data. Each stage asks exactly what survives—and what the invariant still cannot tell you.

One small edit changes the answer

Yellow vertices have odd degree; green vertices have even degree. An open trail needs exactly zero or two odd vertices in this connected graph.

MAP-LIKE DRAWING

Degrees

3 · 3 · 5 · 3

Odd vertices

Verdict

No Euler trail

An invariant compresses a question

Degree decides the trail question, not walking distance or river width.

One number is never the whole shape

Equal χ can still leave orientation, boundary, and knot structure unresolved.

Choices return with data

TDA is interpretable only when metric and scale choices remain visible.

TOUCH THE MATHEMATICS