A cofinal filtration often gives the same limiting homotopy type as another, but the comparison between them is rarely a single strict formula. One chooses stage functions, cellular approximations and homotopies that make diagrams commute up to homotopy. Different competent choices can lead to different point-set maps while representing the same mathematical transport.
This guide continues the study of cofinal filtration comparisons, telescope equivalences and universal-map transport. It focuses on the choice itself: when two comparison maps should be regarded as equivalent, why homotopies between homotopies appear, and how the universal property of a phantom map gives a clean filtration-independent test for nullity.
The aim is precision without mystique. Higher coherence is not decoration, but neither should it be invoked vaguely. We will state what is chosen, what is canonical only up to homotopy, and which conclusion needs only the universal property rather than a strict comparison.
A practical orientation
Keep three levels separate. A filtration is a sequence of finite stages. A stagewise comparison uses an increasing index function and compatible maps. A telescope or universal map packages the sequence into a global object. Equality, homotopy and coherent homotopy are different claims at these levels.
Readers who need only filtration-independent nullity can go directly to the universal-property argument. Readers building functorial systems should follow the comparison and coherence chapters first.
- Set the finite-type context — for hypotheses and notation.
- Construct comparison maps — for stage functions and telescopes.
- Understand higher homotopies — for coherent choices.
- Follow the worked example — for even and tripled subsequences.
- Use the nullity test — for a filtration-independent conclusion.
Chapter contents
Open a route that matches the decision in front of you, or read straight through. Each route returns to the same evidence chain.
Start with the operating question
Build the evidence chain
Construct comparison maps through a common refinement
Work the example
1. Fix the setting and the claim
Let X be a pointed connected CW complex of finite type, with finite stages X₁ ⊂ X₂ ⊂ ··· whose union is X. A second filtration Y₁ ⊂ Y₂ ⊂ ··· is cofinal when every finite stage of one lies, up to the chosen comparison framework, inside a sufficiently large stage of the other.
The desired conclusion must be named. It may be a homotopy equivalence of telescopes, naturality of a universal construction, or merely independence of whether the universal phantom map is null. The last claim is weaker and admits a particularly robust proof.
- Pointed and connected hypotheses keep the mapping statements aligned.
- Finite type ensures the stages used in phantom tests are finite.
- Cofinality compares eventual information, not identical indexing.
- A strict point-set equality is not assumed.
2. Identify the choices hidden in a comparison
To map one filtration into another, choose an increasing function α so that Xₙ is represented inside Yα(n), then choose stage maps compatible with the inclusions up to homotopy. A second builder may choose a larger function α′ or different cellular approximations.
Neither choice is automatically wrong. The mathematical question is whether they determine homotopic maps after passing to the telescope or derived construction, and whether that homotopy is compatible with the structure used later.
| Choice | Why it appears | Appropriate sameness |
|---|---|---|
| Stage function n ↦ α(n) | Indices do not align | Common cofinal refinement |
| Stage map | Cellular or point-set models differ | Pointed homotopy |
| Compatibility homotopy | Squares need not commute strictly | Homotopy coherent commutation |
| Telescope map | Packages the stage data | Homotopy class, with stated naturality |
3. Construct comparison maps through a common refinement
Given two stage functions, pass to a pointwise larger cofinal function γ. Inclusion into the γ-stages provides a place where both choices can be compared. Stagewise homotopies and compatibility data then induce maps between mapping telescopes.
This common-refinement move is the practical engine behind “independence of cofinal subsequence.” It does not erase the construction; it shows how two choices enter a larger diagram where their induced maps can be connected.
4. Distinguish homotopy from canonical identity
If two induced telescope maps are homotopic, they represent the same morphism in the homotopy category. That is often exactly what the theorem needs. Calling the map “canonical” without qualification can overstate the result if no preferred point-set representative was chosen.
A safer formulation is “well defined up to pointed homotopy,” or “canonical in the homotopy category.” If later arguments compare several composable choices, record the chosen homotopies as part of the data rather than assuming they disappear.
5. See why homotopies between homotopies appear
Suppose three comparison choices are available. One can homotope the first to the second and the second to the third, or compare the first directly to the third. A higher homotopy is the coherence witnessing compatibility between those homotopies. It matters when one wants a functorial or multi-stage construction rather than a single isolated equivalence.
Higher coherence should be asserted only when the construction supplies it. In many ordinary nullity arguments it is unnecessary: the universal property bypasses the choice. In structured settings, however, ignoring coherence can make a diagram commute pairwise while failing to assemble globally.
| Level | Data | Typical purpose |
|---|---|---|
| Map | A → B | Compare two constructions |
| Homotopy | Two maps A → B | Show equality in the homotopy category |
| Higher homotopy | Two homotopies between maps | Make multiple comparisons coherent |
| Contractible choice space | All compatible choices connected coherently | Support a stronger canonical or functorial claim |
6. Worked example: even and tripled skeletal subsequences
Start with the skeletal filtration X₁ ⊂ X₂ ⊂ ···. Compare the even subsequence Aₙ = X₂ₙ with the tripled subsequence Bₙ = X₃ₙ. One stage function sends Aₙ into Bₙ because 2n ≤ 3n. Another sends Aₙ into B₂ₙ. Both are cofinal, but they give different stage bookkeeping.
Use Cₙ = X₆ₙ as a common refinement. Both chosen routes include into Cₙ, and the cellular inclusions give compatible homotopies at the telescope level. The resulting telescope comparisons need not be literally the same map, but they represent the same eventual transport under the stated hypotheses.
7. Transport universal phantom maps carefully
A phantom map from X is a map whose restriction to every finite subcomplex is null. A universal phantom map Θ has the property that every phantom map out of X factors through Θ, in the appropriate homotopy sense. A filtration construction may produce a model for Θ, but the universal property is the invariant content.
Comparison maps between filtrations can transport the constructed models. State whether the diagram commutes strictly or up to homotopy. For many conclusions, homotopy commutation is enough because factorisation and nullity live in homotopy classes.
8. Prove filtration-independent nullity by universality
Let ΘA and ΘB be universal phantom maps produced from two admissible cofinal filtrations. If ΘA is null, every phantom map f out of X factors as f ≃ g ∘ ΘA and is therefore null. In particular ΘB, itself a phantom map out of X, is null.
The converse is symmetric. Thus nullity of the universal phantom map is independent of which admissible filtration produced the model. Notice what this proof avoids: no strict equality of telescopes, no preferred stage function and no higher-coherence theorem are needed for the binary nullity conclusion.
- Assume each constructed map really has the stated universal property.
- Use factorisation in the pointed homotopy category.
- Apply the conclusion to the other universal phantom map.
- Repeat symmetrically for equivalence.
9. Know when comparison-map coherence is still needed
Universality settles nullity, but it does not automatically produce a chosen natural equivalence between entire constructions. If the goal is functoriality in X, compatibility with another operation, or a coherent diagram across many filtrations, comparison maps and higher homotopies return to centre stage.
Match the proof to the claim. Use the shortest invariant argument for nullity; build the stronger coherent system only when the theorem or application actually requires it.
10. Respect the scope and common failure modes
Cofinality alone is not a magic word. The stages must be appropriate for the phantom-map test, and the telescope model must satisfy the needed universal property. Pointedness, connectedness and finite-type assumptions may be essential to the particular theorem in use.
Do not transfer the argument unchanged to pro-categories, spectra, non-CW filtrations or unpointed settings without checking definitions. Also avoid inferring that a contractible choice space exists merely because two choices are homotopic.
11. Practical questions about choices and nullity
When reading a proof, mark every use of “choose,” “after passing to a subsequence,” and “canonical.” Ask what category the equality belongs to and whether subsequent composition needs coherence. These small annotations often reveal whether the stated conclusion is stronger than the construction supports.
If two comparison maps are homotopic, are all higher choices automatically coherent?
No. A homotopy identifies one pair of maps in the homotopy category. Coherence among several homotopies is additional data or a further theorem.
Why use universality for nullity instead of comparing telescopes directly?
Because the universal property makes the conclusion invariant and short. If one universal phantom map is null, every phantom map is null, including the universal map built from any other admissible filtration.
12. Changed-context transfer, evidence check and sources
Transfer the distinction to two cofinal Postnikov subsequences. Stage functions and comparison maps again require choices, and stronger naturality may need coherent data. But if two constructions represent an object with the same universal factorisation property, test an invariant such as nullity through that property before building an unnecessarily strict comparison.
Gray and McGibbon’s paper Universal Phantom Maps develops universal phantom-map constructions and their factorisation role. Hatcher’s Algebraic Topology supplies standard CW, homotopy and telescope background. Read the hypotheses of the exact theorem being used; the argument here explains the proof pattern and its limits, not a licence to drop those hypotheses.
Continue through the Gold Standard ecosystem
Return to the Topology subject root, the Gold Standard of Education parent, or the How X Works hub.
Immediate predecessor: cofinal filtration comparisons and universal-map transport · Prerequisite: universal phantom maps and finite skeleta · Topology root
These routes are grouped by learning purpose: prerequisite model, immediate predecessor, and the wider transfer framework. They are not substitutes for the evidence and limits stated in this guide.