WHY MATHEMATICS?

Why Mathematics? | Persistent Homology for Data Shape: Vietoris–Rips Filtrations, Boundary Matrices, Barcodes and Stability

How persistent homology turns changing neighbourhood scale into algebraic evidence about components, loops and voids without treating every long bar as truth.

Why mathematics is important becomes easier to see when a real decision depends on it. This guide follows the mechanism from definitions and dimensions to a worked calculation, validation, practical learning and responsible use. It keeps factual evidence separate from interpretation and treats every illustrative scenario as an example rather than a promise of results.

Choose a reading route

Build a metric point cloud and nested filtration
Translate simplices into chains and boundary maps
Reduce the boundary matrix to pair births and deaths
Compare diagrams across scale, sampling and validation sets

Chapter map

1. Point cloud
2. Metric
3. Scale parameter
4. Proximity graph
5. Vietoris–Rips simplex
6. Filtration nesting
7. Chain group
8. Boundary operator
9. Cycles and boundaries
10. Betti numbers
11. Birth event
12. Death event
13. Boundary matrix reduction
14. Barcode and diagram
15. Stability distance
16. Statistical validation

Visit the How Mathematics Works hub for the wider series and its connections to education, reasoning, science, computing, design and careers.

1. Point cloud

The central mathematical idea is concrete. Rows are observations and columns are measured coordinates or features.

A concrete situation

Four square corners form a planar point cloud.

What the mathematics does not license

Feature scales distort distance. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Standardize only with scientific justification. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

2. Metric

Here the geometry becomes operational. A distance function defines closeness.

A concrete situation

Euclidean distance compares the square corners.

What the mathematics does not license

Euclidean geometry may ignore meaningful invariance. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Compare domain-relevant metrics. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

3. Scale parameter

This step deserves a careful pause. Epsilon controls which relationships are visible.

A concrete situation

Edges of length at most one enter at ε=1.

What the mathematics does not license

One selected scale can cherry-pick shape. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Use the full filtration. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

4. Proximity graph

The useful question is not only how to calculate, but what the calculation assumes. Vertices connect when distance is within threshold.

A concrete situation

The four unit square edges appear before diagonals.

What the mathematics does not license

Noise adds short accidental edges. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Study perturbations and subsamples. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

Did You Know? In this chapter, the most important number is useful only together with its construction rule. Changing scaling, sample membership or validation design can change the numerical answer without changing the arithmetic.

5. Vietoris–Rips simplex

A small example makes the mechanism visible. Every clique becomes a simplex at its diameter.

A concrete situation

A triangle enters when all three edges are present.

What the mathematics does not license

Rips complexes grow combinatorially. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Limit dimension and use reductions. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

6. Filtration nesting

This is where modelling choices begin to affect evidence. Every earlier complex is contained in every later one.

A concrete situation

Kε is a subset of Kδ when ε≤δ.

What the mathematics does not license

Inconsistent preprocessing can break nesting. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Verify filtration values. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

7. Chain group

The central mathematical idea is concrete. Formal sums of k-simplices are vectors over a field.

A concrete situation

Over F2, coefficients are zero or one.

What the mathematics does not license

Field choice can change torsion-sensitive results. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Report coefficients. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

8. Boundary operator

Here the geometry becomes operational. The boundary of a simplex is the alternating sum of faces.

A concrete situation

A triangle boundary contains its three edges.

What the mathematics does not license

Orientation signs matter outside F2. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Check ∂ composed with ∂ equals zero. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

9. Cycles and boundaries

This step deserves a careful pause. Cycles have zero boundary; boundaries come from higher simplices.

A concrete situation

A square edge loop is a cycle.

What the mathematics does not license

Not every cycle represents a hole. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Quotient cycles by boundaries. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

10. Betti numbers

The useful question is not only how to calculate, but what the calculation assumes. Dimension of homology counts independent classes.

A concrete situation

Before merging, β0 equals four.

What the mathematics does not license

Counts at one scale ignore lifetime. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Use persistent classes. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

11. Birth event

A small example makes the mechanism visible. A simplex can create a new homology class.

A concrete situation

Square edges create one H1 loop at ε=1.

What the mathematics does not license

Birth depends on tie ordering conventions. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Use consistent filtration order. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

12. Death event

This is where modelling choices begin to affect evidence. A later simplex can turn a cycle into a boundary.

A concrete situation

At ε=sqrt(2), diagonals and triangles fill the square loop.

What the mathematics does not license

Simultaneous events can yield zero bars. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Interpret exact ties carefully. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

13. Boundary matrix reduction

The central mathematical idea is concrete. Column additions expose low pivots and simplex pairings.

A concrete situation

A pivot pairs a birth column with a death column.

What the mathematics does not license

Naive matrices consume memory. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Use sparse and clearing algorithms. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

14. Barcode and diagram

Here the geometry becomes operational. Intervals or birth-death points summarize persistence.

A concrete situation

The square loop has interval [1,sqrt(2)).

What the mathematics does not license

Infinite classes need a convention. A trustworthy workflow records the choice before seeing the final test result.

Try it, check it, transfer it

Report truncation and essential bars. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

15. Stability distance

This step deserves a careful pause. Bottleneck or Wasserstein distances compare diagrams.

A concrete situation

Small perturbations move matched points modestly under theorem conditions.

What the mathematics does not license

Stability does not validate a bad metric. Evidence becomes stronger when the limitation is tested instead of hidden.

Try it, check it, transfer it

Perturb data and metric. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

16. Statistical validation

The useful question is not only how to calculate, but what the calculation assumes. Subsampling and null simulations assess recurring features.

A concrete situation

A loop reappears across patient-level resamples.

What the mathematics does not license

Observations within one sample are dependent. Keeping this boundary clear makes the result easier to audit and transfer.

Try it, check it, transfer it

Resample independent units and use domain evidence. Then explain the result to someone who has not seen the formula. If the explanation needs a hidden assumption, add that assumption to the model record.

A seven-day practice lab

Day 1: define the decision, observation unit, X variables and Y or reference condition.
Day 2: reproduce the smallest worked example and check every dimension, unit and sign.
Day 3: build a transparent baseline and write what it assumes.
Day 4: fit the main method inside a training-only pipeline; record every parameter.
Day 5: challenge one assumption with a deliberately difficult dataset or operating regime.
Day 6: compare validation metrics with the practical workload created by errors.
Day 7: present one supported conclusion, one unresolved uncertainty and one safe next experiment.

Questions for a student, parent or mentor

Which part of the model is a mathematical definition, and which part is a choice?
What small calculation would catch a sign, scale or dimension error?
What evidence comes from training, validation and the untouched test?
What benign condition could look like a fault or prediction failure?
How would the answer change for another response, asset, school project or time period?
What should a human review before acting on the score?

A method-specific audit before deployment

Persistent homology studies shape across scale. Begin with points and a distance. At threshold epsilon, join nearby points by edges and fill higher-dimensional simplices according to a chosen complex. As epsilon grows, connected components merge, loops appear and fill, and higher-dimensional voids can form and disappear. A filtration records this nested history instead of forcing one arbitrary scale.

In a Vietoris–Rips complex, a finite set of vertices becomes a simplex when every pair is close enough; equivalently, its filtration value is its diameter. Homology converts the complex into algebra. Boundary maps describe how each k-simplex is attached to its faces, cycles lie in the kernel, boundaries lie in the image, and the quotient distinguishes holes that are not merely boundaries of higher-dimensional pieces.

Matrix reduction over a field pairs births and deaths. A feature born at b and destroyed at d has persistence d minus b, represented as an interval or a point in a persistence diagram. Long lifetime can indicate structure that survives many scales, but persistence is not a universal importance score. Sampling density, metric, noise, filtration and scientific question determine what a long bar means.

Topological persistence is used in shape analysis, materials, neuroscience, sensor networks and biological data. Stability theorems show controlled changes under appropriate perturbations, not immunity to bad data or a wrong metric. Responsible use checks units and scaling, compares filtrations, uses subsampling or bootstrap summaries, reports zero-length and essential classes correctly, and validates topological features with domain evidence.

Frequently asked questions

What is persistent homology?

Persistent homology records when connected components, loops and higher-dimensional holes appear and disappear as a filtered complex grows.

What is a Vietoris–Rips complex?

At scale ε it is the clique complex of the proximity graph; a simplex enters when its diameter is at most ε.

What do H0, H1 and H2 represent?

In common geometric settings they summarize connected components, one-dimensional loops and enclosed voids.

What is a barcode?

It is a collection of intervals [birth, death), one for each homology class, showing the scales over which a feature exists.

Why reduce a boundary matrix?

Column reduction pairs the simplex that creates a class with the simplex that destroys it.

Does a long bar always mean a real object?

No. Persistence is relative to the metric, sampling, filtration and domain question; validation remains necessary.

Primary source and fact-check note

Herbert Edelsbrunner, David Letscher and Afra Zomorodian, “Topological Persistence and Simplification,” Discrete & Computational Geometry 28, 511–533 (2002), DOI 10.1007/s00454-002-2885-2. Definitions and core claims in this article were checked against this primary publication. Calculations shown here were independently recomputed. Examples are educational illustrations; they are not reported industrial performance results.

Useful next reading

Keep the options open. Mathematics education supports clearer questions, transferable problem-solving skills and informed choices across science, engineering, computing, AI, statistics, design and many other careers. It does not guarantee intelligence, income, admission or employment; those outcomes depend on far more than one subject or model.

Extended laboratory: topic-specific validation and responsible use

Metric audit

Record units, scaling and invariances.

Filtration audit

Verify simplex values and nesting.

Algebra audit

State coefficient field, dimension and boundary convention.

Complexity audit

Track simplex count, memory and approximation.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Pairing audit

Inspect births, deaths, ties and essential classes.

Stability audit

Perturb points, metrics and subsamples.

Null audit

Compare with domain-specific simulations and independent units.

Final claim

A persistent class is a scale-stable algebraic feature, not automatic truth.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Mechanism workshop: calculate, challenge and transfer

Workshop 1: Point cloud

Begin with this mechanism: Rows are observations and columns are measured coordinates or features. Recreate the concrete case: Four square corners form a planar point cloud. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Feature scales distort distance. Complete the transfer step: Standardize only with scientific justification. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 2: Metric

Begin with this mechanism: A distance function defines closeness. Recreate the concrete case: Euclidean distance compares the square corners. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Euclidean geometry may ignore meaningful invariance. Complete the transfer step: Compare domain-relevant metrics. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 3: Scale parameter

Begin with this mechanism: Epsilon controls which relationships are visible. Recreate the concrete case: Edges of length at most one enter at ε=1. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: One selected scale can cherry-pick shape. Complete the transfer step: Use the full filtration. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 4: Proximity graph

Begin with this mechanism: Vertices connect when distance is within threshold. Recreate the concrete case: The four unit square edges appear before diagonals. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Noise adds short accidental edges. Complete the transfer step: Study perturbations and subsamples. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 5: Vietoris–Rips simplex

Begin with this mechanism: Every clique becomes a simplex at its diameter. Recreate the concrete case: A triangle enters when all three edges are present. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Rips complexes grow combinatorially. Complete the transfer step: Limit dimension and use reductions. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 6: Filtration nesting

Begin with this mechanism: Every earlier complex is contained in every later one. Recreate the concrete case: Kε is a subset of Kδ when ε≤δ. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Inconsistent preprocessing can break nesting. Complete the transfer step: Verify filtration values. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 7: Chain group

Begin with this mechanism: Formal sums of k-simplices are vectors over a field. Recreate the concrete case: Over F2, coefficients are zero or one. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Field choice can change torsion-sensitive results. Complete the transfer step: Report coefficients. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 8: Boundary operator

Begin with this mechanism: The boundary of a simplex is the alternating sum of faces. Recreate the concrete case: A triangle boundary contains its three edges. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Orientation signs matter outside F2. Complete the transfer step: Check ∂ composed with ∂ equals zero. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 9: Cycles and boundaries

Begin with this mechanism: Cycles have zero boundary; boundaries come from higher simplices. Recreate the concrete case: A square edge loop is a cycle. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Not every cycle represents a hole. Complete the transfer step: Quotient cycles by boundaries. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 10: Betti numbers

Begin with this mechanism: Dimension of homology counts independent classes. Recreate the concrete case: Before merging, β0 equals four. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Counts at one scale ignore lifetime. Complete the transfer step: Use persistent classes. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 11: Birth event

Begin with this mechanism: A simplex can create a new homology class. Recreate the concrete case: Square edges create one H1 loop at ε=1. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Birth depends on tie ordering conventions. Complete the transfer step: Use consistent filtration order. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 12: Death event

Begin with this mechanism: A later simplex can turn a cycle into a boundary. Recreate the concrete case: At ε=sqrt(2), diagonals and triangles fill the square loop. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Simultaneous events can yield zero bars. Complete the transfer step: Interpret exact ties carefully. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Did You Know? A sophisticated adjustment can be algebraically correct and still answer the wrong scientific question when samples, labels, batches or validation boundaries are wrong.

Workshop 13: Boundary matrix reduction

Begin with this mechanism: Column additions expose low pivots and simplex pairings. Recreate the concrete case: A pivot pairs a birth column with a death column. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Naive matrices consume memory. Complete the transfer step: Use sparse and clearing algorithms. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 14: Barcode and diagram

Begin with this mechanism: Intervals or birth-death points summarize persistence. Recreate the concrete case: The square loop has interval [1,sqrt(2)). Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Infinite classes need a convention. Complete the transfer step: Report truncation and essential bars. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 15: Stability distance

Begin with this mechanism: Bottleneck or Wasserstein distances compare diagrams. Recreate the concrete case: Small perturbations move matched points modestly under theorem conditions. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Stability does not validate a bad metric. Complete the transfer step: Perturb data and metric. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.

Workshop 16: Statistical validation

Begin with this mechanism: Subsampling and null simulations assess recurring features. Recreate the concrete case: A loop reappears across patient-level resamples. Write coordinates, distances, thresholds, edges, simplices, filtration values, chain dimensions, boundary columns, pivots, Betti numbers, birth-death pairs, intervals and diagram distances with shapes, units, denominators and assumptions. Calculate the smallest non-trivial example by hand and reproduce it in a spreadsheet or short program. Challenge this limitation: Observations within one sample are dependent. Complete the transfer step: Resample independent units and use domain evidence. Change exactly one value definition, coordinate unit, weights rule, radius, neighbour count, row transformation, self-weight, mean, outlier, permutation restriction, alternative hypothesis, window boundary, edge correction, intensity model, radius grid, simulation count, metric, feature scaling, filtration, coefficient field, simplex dimension, tie order, resampling unit or random seed. Explain which mathematical object changes, which conclusion survives and which claim must weaken. Keep observation, preprocessing, spatial or metric structure, statistic, null model, computation, uncertainty, interpretation and independent validation on separate notebook lines. Reserve complete regions, time periods, specimens or patients outside tuning when testing transfer.