Coverage contract
This project formalizes selected results from Hatcher's Algebraic Topology; it does not claim to formalize the whole book. The canonical artifact is a project-authored, citation-only execution ledger rather than redistributed book text. Its thirty-two lines record seventeen decomposed source areas, thirteen explicitly deferred areas, and two areas outside scope. Every decomposed row links directly to its formalizable roadmap leaves.
In scope
Hatcher §1.1, Basic constructions,
pages 25–38, decomposed into twenty-two explicit nodes. The main target is Theorem 1.7,
π₁(S¹) ≅ ℤ. The section's other in-scope results are:
Three qualifications on what these nodes do and do not prove.
The opening results are explicit nodes. Propositions 1.2, 1.5, and 1.6
track the exact pinned declarations Path.Homotopic.equivalence,
FundamentalGroup.fundamentalGroupMulEquivOfPath, and
simply_connected_iff_unique_homotopic. Example 1.4 tracks the exact stronger
input Convex.contractibleSpace; SimplyConnectedSpace.ofContractible and
Mathlib's subsingleton fundamental-group instance give Hatcher's conclusion.
Example 1.1 and Proposition 1.3 are local formalizations because they record
Hatcher's explicit affine homotopy and left-then-right concatenation
convention.
Theorem 1.8 is already upstream. Its statement is Mathlib's
Complex.exists_root, packaged by Complex.isAlgClosed, in
Analysis/Complex/Polynomial/Basic.lean and proved by Liouville's theorem.
The local formalization gives a second proof of a known result. It
is in scope as a source target and must not be counted as new Mathlib coverage.
Lemma 1.15 is reused by §1.2. It was formalized locally because Proposition 1.14 needed it, and it now supplies the surjectivity clause of Theorem 1.20 exactly as in Hatcher. The proof is adapted from the matching implementation in open Mathlib PR #28246. The overlap is recorded on the node.
§1.2 selected spine
Van Kampen's theorem is decomposed from its free-product presentation through the kernel calculation, wedge sums, cell attachments, and Corollary 1.28. Proposition 1.26 is split by part, and the Appendix compactness lemma needed for its 2-skeleton clause is a separate cross-chapter chain. The cover presentation, threefold-incidence homotopy decomposition, quotient form of van Kampen, pointed-wedge theorem, point-set cell-attachment model, strip cover, overlap analysis, and Proposition 1.26(a)–(b) are formalized. The classical successor-skeleton bridge, its 2-skeleton connectivity and finite-stage induction, and the independent presentation-complex route are formalized as well. All 67 leaves linked from the selected §1.2 unit are complete.
§1.3 selected classification and deck spine
Covering spaces is decomposed from homotopy lifting through the subgroup classification of connected covers. Exact pinned lifting results are separated from local source-facing wrappers. The path-class universal-cover construction, the fixed-universe quotient of pointed connected covers, and its cross-universe rigidity theorem are formalized. The deck-transformation branch continues through the normalizer quotient and orbit-action calculation in Propositions 1.39–1.40. All 32 leaves linked from the selected §1.3 unit are complete.
§2.1 selected slices
Simplicial and singular homology
is decomposed through the singular-chain and singular-homology definitions,
the exact H₀ calculation, the point calculation, induced maps, chain-homotopy
invariance, topological homotopy invariance, and Corollary 2.11. Eight nodes
are exact declarations in the pinned Mathlib; the point calculation and
homotopy-equivalence corollary are formalized locally, so all ten linked leaves
are complete. The project does not claim Hatcher's explicit prism formula is
formalized.
Relative homology and exact sequences adds a fifteen-leaf second slice: the augmented definition of reduced homology, relative chains and homology, Theorem 2.16 and the long exact sequence of a pair, pair- and reduced-pair-sequence naturality, Example 2.18, and Proposition 2.19. All fifteen leaves are complete: the four-node reduced-homology branch, singular-pair functor, relative-homology functor, pair long exact sequence with its connecting-map formula and naturality, relative homotopy branch through Proposition 2.19, reduced pair sequence with its naturality, and pointed comparison are formalized locally, while the generic algebraic exact-sequence theorem and simplicial-pair foundation are present in the pinned Mathlib.
The same milestone now has a four-leaf triple-sequence slice: a project-authored topological-triple interface, the short exact sequence of relative chain complexes, its long exact homology sequence including the degree-zero endpoint, and naturality. All four leaves are formalized locally using the pinned pair and generic homology-sequence APIs; open Mathlib PR #41318 is prior art rather than an upstream dependency.
All ten leaves in the small-chains/excision slice are complete. Its subdivision, small-chain equivalence, relative dévissage, binary-cover chain-level comparison, and both homology-level excision endpoints are formalized locally.
§2.1–§2.2 selected Mayer–Vietoris, sphere, Brouwer, and degree slices
Mayer–Vietoris sequences are decomposed into six formalizable leaves. The selected slice constructs the binary-cover chain complex, proves its short exact sequence, transports its homology to ordinary singular homology, and packages the ordinary and reduced long exact sequences through their degree-zero endpoints. All six leaves are formalized locally and complete.
The sphere-homology milestone
adds thirteen formalizable leaves. They cover reduced homotopy invariance and
contractible vanishing, ordinary and reduced neighborhood-retract transport,
the sphere-specific radial and stereographic geometry, Example 2.46's
recurrence, the S⁰ base case, and the full reduced sphere calculation of
Corollary 2.14. All thirteen leaves are formalized locally and complete.
Corollary 2.15 adds three leaves: the homological disk-boundary no-retraction theorem, the fixed-point-free boundary-ray retraction, and Brouwer's fixed-point theorem for positive-dimensional disks. All three are formalized locally and complete. The general §2.2 CW-subcomplex neighborhood extension beyond the selected Appendix A.5 support, and Examples 2.47–2.48, remain deferred with the rest of §2.2.
The good-pair quotient milestone decomposes Proposition 2.22, Theorem 2.13, and naturality into fifteen formalization leaves, all formalized locally and complete. Its shared strong-deformation-retract background leaf is also complete.
The post-excision applications milestone decomposes the arbitrary-pair mapping-cone comparison, Example 2.23, Corollaries 2.24–2.25, and Theorem 2.26 into twenty-seven formalization leaves. Its Appendix A.5 branch adds five regular-neighborhood leaves. All thirty-two leaves are formalized locally and complete.
The degree-foundations milestone decomposes Hatcher's degree properties (a)–(g), Theorem 2.28, and Proposition 2.29 into sixteen formalizable leaves. All sixteen are formalized locally and complete.
The local-degree milestone continues through Proposition 2.30 in twenty formalizable leaves. It plans the oriented local-homology class, neighborhood-independent local degree, the finite-puncture coproduct and projection calculation, and the sum of local degrees. These leaves are planned rather than asserted complete.
Deferred roadmap expansion
The remaining high-level section maps and retained source-unit remainders are explicitly deferred to future roadmap runs. Wholly deferred pages carry exposition and prior-art notes rather than formalization nodes:
- Chapter 0, Some underlying geometric notions
- Chapter 2, Homology, §2.3 and the remainder of §2.1–§2.2 outside the selected slices
- Chapter 3, Cohomology, all three sections
- Chapter 4, Homotopy theory, all three sections
- The remainder of the Appendix
Decomposing any of these is a future roadmap run, not proof work. Within
Chapter 1, Example 1.35, permutation reconstruction, and Examples 1.41–1.48
remain deferred. Within §2.1, Theorem 2.20 and Proposition 2.21 are already
decomposed into ten formalizable leaves,
and Corollary 2.14 is now part of the sphere milestone. Corollary 2.15 is
decomposed into three complete formalizable leaves. Theorem 2.13 and
Proposition 2.22 are decomposed and complete in the good-pair milestone. Lemma
2.1 and Examples 2.2–2.5, Proposition 2.6 beyond the selected S⁰ support, Example
2.17, and Theorem 2.27 remain deferred. The mapping-cone comparison, Example
2.23, Corollaries 2.24–2.25, and invariance of dimension are decomposed in the
post-excision milestone. Within
§2.2, Examples 2.31–2.32 and their degree-realization calculations, suspension
invariance, cellular homology, Euler characteristic, homology of groups, the
general CW-neighborhood construction, Examples 2.47–2.48, mapping-torus and
relative Mayer–Vietoris sequences, and homology with coefficients remain
deferred.
Out of scope
- Lettered Additional Topics. Sections 1.A, 1.B, 2.A–2.C, 3.A–3.H, and 4.A–4.L are excluded and have no roadmap pages. They are supplementary in the book and would double the mapping effort for material the main line does not depend on.
- Exercises. Hatcher's exercise sets are not source targets.
- Reproducing the text. The book is copyrighted and this repository is
public, so
/sources/is gitignored. Source notes carry coordinates and one-line paraphrases only.
Done means
A local theorem-like node is complete when its named Lean declarations compile
and its main result carries both statement: formalized and
proof: formalized. A data-valued declaration has no separate proof phase and
is complete when its main result compiles with statement: formalized. In both
cases, the implementation uses no sorry and no axioms beyond Lean's three. A
Mathlib-backed node is complete when its exact pinned declaration, kind, and
declaring file pass provenance checks. CI performs these checks.
A section counts as finished only when it is decomposed and every node beneath it is complete, and nothing numbered in it has been deferred. By that rule §1.1 is finished. A section that is merely mapped is never finished, whatever its prose says.
The project as a whole makes no completion claim. Decomposing additional sections does not by itself imply that their nodes are formalized.