How statement and proof provenance work
The first chip identifies the source of the statement or construction; the second identifies the source of its local proof or verification.
- Literature-sourced: the exact statement appears in a cited source; only wording and notation differ.
- AI-adapted: a semantically identical restatement of literature-sourced material, modulo indexing, notation, and boundary cases adopted by the library.
- AI-generated: a genuinely novel statement formulated by AI, with no source for the claim itself.
These labels describe origin, not correctness: citations and verification chips remain separate evidence.
Handle Cancellation Slides and Elementary Moves — Examples
1 · Prerequisites
- Abelian Categories
- Absolute and Conditional Convergence; Rearrangement; Products
- Arc Length and Rectifiable Curves
- Binary Operations, Monoids, Groups and Subgroups
- Cardinal Arithmetic, Cofinality and the Alephs
- Categories, Functors and Natural Transformations
- Chain Complexes and Homology
- Chain Homotopy and the Homotopy Category
- Compactness
- Compactness in Metric Spaces
- Completeness, Completion, and Uniform Continuity
- Congruences, the Integers Modulo n and the Chinese Remainder Theorem
- Connectedness
- Connections Levi Civita and Parallel Transport
- Constant Rank, Submersions, Immersions and Regular Level Sets
- Construction of the Natural Numbers
- Construction of the Real Numbers via Cauchy Sequences
- Construction of the Real Numbers via Dedekind Cuts
- Continuity, IVT, EVT, and Uniform Continuity
- Cosets, Index and Lagrange's Theorem
- Countability and Uncountability
- Countability Axioms and Cardinal Functions
- Cw Complexes and Cellular Homology
- Darboux, L'Hôpital, and Taylor's Theorem
- Determinants of Matrices over a Commutative Ring
- Divisibility, Euclidean Domains, Principal Ideal Domains and Unique Factorisation
- Divisibility, Greatest Common Divisors and Bézout's Identity
- Dual Spaces, Bilinear and Quadratic Forms, and Sylvester's Law of Inertia
- Euclidean Ordinary Differential Equations with Smooth Dependence
- Exactness and the Member Calculus
- Exterior Powers, Orientation and Hodge Duality
- Filters and Ultrafilters
- Finite Counting, Factorials and Binomial Coefficients
- Foundations of the Real Numbers for Analysis
- Free Groups and Presentations
- Free Modules, Exact Sequences, Projective and Injective Modules
- Free Products and Amalgamation
- Fubini and Change of Variables
- Function Space Topologies and the Exponential Law
- Fundamental Solutions Newtonian Potentials and Green Functions
- Gaussian Elimination, Elementary Matrices and Reduced Row Echelon Form
- Geodesics, the Exponential Map, Completeness, and Hopf–Rinow
- Gradient Like Vector Fields and Morse Trajectories
- Group Actions, Orbits, Stabilisers and Cayley's Theorem
- Group Homomorphisms and the Isomorphism Theorems
- Handle Cancellation Slides and Elementary Moves
- Handle Decompositions Duality and Rearrangement
- Hausdorff via the Diagonal
- Hereditary and Productive Behaviour of the Separation Axioms
- Higher Homotopy Groups and Cofiber Sequences
- Homology Axioms Degree and Classical Applications
- Homotopy and Homotopy Equivalence
- Hurewicz Whitehead Freudenthal and Cw Approximation
- Ideals, Quotient Rings and the Isomorphism Theorems for Rings
- Inner Product Spaces, Gram-Schmidt, Projections and Adjoints
- Limits and Colimits
- Limits of Real Functions
- limsup, liminf, and Subsequential Limits
- Line Integrals and the Gradient Theorem
- Linear Independence, Bases and Dimension
- Linear Transformations, Rank-Nullity and Quotient Spaces
- Long Exact Sequences in Homology
- Manifolds with Boundary Collars and Orientations
- Matrices, the Matrix of a Linear Map, and Change of Basis
- Metric Spaces
- Mixed Partials, Taylor Formulae, and Extrema
- Modules, Submodules, Quotient Modules and the Isomorphism Theorems
- Monotone Functions, Discontinuities, and Continuity Sets
- Monotone Sequences, Bolzano-Weierstrass, and Cauchy Completeness
- Morse Critical Points Hessians and Indices
- Morse Functions Critical Values and Genericity
- Normal Subgroups and Quotient Groups
- Order, Zorn's Lemma, and the Axiom of Choice
- Ordinal Arithmetic and the First Uncountable Ordinal
- Ordinals, Cardinals, and Transfinite Recursion
- Oriented and Mod Two Intersection Numbers
- Partitions of Unity and Paracompactness
- Picard-Lindelöf and First-Order Ordinary Differential Equations
- Polynomial Rings, the Division Algorithm and Roots
- Power Series and Real-Analytic Functions
- Preadditive and Additive Categories and Biproducts
- Properties of the Integral and the Working FTC
- Rank Theorems and Embedded Submanifolds
- Reflective Subcategories and the Adjoint Functor Theorems
- Relations, Functions, and Quotients
- Relative Homology Excision and Mayer Vietoris
- Riemannian Metrics Length Distance and Volume
- Rings, Subrings, Integral Domains and Fields
- Rⁿ as a Normed Space; Vector-Valued Functions
- Roots, Rational Powers, and Classical Inequalities
- Sard Theorem and Transversality
- Separation Axioms: the Hierarchy
- Sequences and Limits
- Sequences and Series of Functions; Uniform Convergence
- Series: Convergence and the Nonnegative Tests
- Simple Field Extensions and the Construction of the Complex Numbers
- Simplicial Complexes and Simplicial Homology
- Simplicial Subdivision and Simplicial Approximation
- Singular Chains and Singular Homology
- Smooth Manifolds and Smooth Maps
- Smooth Partitions of Unity and Exhaustions
- Smooth Vector Bundles and Sections
- Sublevel Deformation and the Handle Attachment Theorem
- Subspaces, Products, and Quotients
- Suprema and Infima
- Symmetric Groups, Cycle Decomposition and the Sign Homomorphism
- Tangent Cotangent and the Differential
- Tensor Fields Exterior Algebra and Differential Forms
- Tensor Products of Modules
- The de Rham Theorem and Degree
- The Derivative and the Mean Value Theorems
- The Determinant of a Linear Operator, Cofactors and Cramer's Rule
- The Diagram Lemmas in an Abelian Category
- The Exponential Function
- The Exterior Derivative and Cartan Calculus
- The Fundamental Group
- The Fundamental Theorems of Calculus
- The Inverse and Implicit Function Theorems
- The Inverse Function Theorem Completed
- The Logarithm and General Powers
- The Riemann Integral in Rᵐ and Jordan Content
- The Riemann Integral: Definition and Integrability
- The Seifert–van Kampen Theorem
- The Topology of Euclidean Space
- The Total Derivative in ℝᵐ → ℝⁿ
- The ZFC Axioms and the Basic Set Constructions
- Topological Spaces and Continuity
- Topology of ℝ
- Universal Properties, Representables and the Yoneda Lemma
- Vector Fields Flows and Lie Derivatives
- Vector Spaces, Linear Subspaces, Span and Direct Sums
- Whitney Embedding Tubular Neighbourhoods and Approximation
2 · Summary
The examples make the two elementary moves concrete. A - pair is cancelling as soon as the attaching -sphere of the band has one point on the belt sphere of the ball, and the model is the identity that two balls joined by a band are one ball; a - pair on a surface cancels when the attaching circle of the disc runs through exactly one point of the belt -sphere of the band, and the matrix definition deliberately does not cover these endpoint cases. On the four-dimensional side, a -handle attached to a -ball has middle boundary , and two -handles attached along disjoint circles realize an elementary row operation: sliding the second -handle over the first changes the matrix column to , exactly the row operation of the matrix-operation proposition.
The two counterexamples display the hypotheses that the theorem cannot drop. A finger move of a sphere across a belt circle produces three transverse intersections with local signs , so the algebraic intersection is still a unit while the geometric intersection has three points: a unit entry does not supply the single point the cancellation theorem needs. This inserted opposite pair can be removed by reversing the finger isotopy; a general algebraic-to-geometric conversion needs separate geometric hypotheses, such as those of the Whitney trick. Finally, a -handle attached to a solid torus along a circle that bounds a disk in the boundary disjoint from the belt circle has vanishing intersection with the belt sphere, so the pair is not geometrically cancelling and in fact cannot cancel: the attaching circle is null-homotopic in the solid torus, so the total space stays with fundamental group while the lower stage is simply connected.
3 · Logical flowchart
4 · Definitions, theorems and proofs
None yet.
5 · Examples, counterexamples and false statements
A cancelling zero-one handle pair
Example
Assume . In dimension attach a -handle to a compact manifold as a disjoint ball and then attach a -handle by an embedding whose attaching -sphere consists of one point on the new boundary sphere of (its belt sphere) and one point on . In the endpoint convention of the definition the pair is geometrically cancelling, exactly one point of the -sphere lying in the belt sphere with transversality automatic, and relative to . The local model is the identity : two -balls joined by a -handle are one -ball.
Facts & Assumptions
Given: Dimension , a compact manifold , a -handle attached as a disjoint ball and a -handle attached by an embedding whose attaching -sphere has one point on the new boundary sphere of and one point on .
Geometrically cancelling adjacent handle pair and K handle core cocore attaching region and belt sphere: the belt sphere of a -handle is the new boundary sphere (disconnected when ), the attaching sphere of a -handle is a -sphere, and the endpoint convention counts exactly one point of the -sphere lying in the belt sphere, transversality being automatic.
The standard complementary pair fills a ball: for the standard embedding the union of the standard -handle and the standard -handle is an -disc: in its case the standard complementary pair fills the ball, i.e. for the standard hemisphere embedding.
Handle cancellation, Attaching a smooth handle with corner rounding and The Axiom of Countable Choice (): assume ; a geometrically cancelling pair may be deleted, giving a diffeomorphism relative to the incoming boundary; attachments are formed with corners rounded. The assumption is used through the standard-pair model of [F2] and through this deletion.
Verification
Given: The configuration of the statement.
The attaching -sphere of consists of two points, one on the belt sphere of the attached -handle and one on ; in the endpoint convention of [F1] exactly one point of the -sphere lies in the belt sphere, with transversality automatic in these dimensions. Hence the pair is geometrically cancelling.
The local model is [F2] with : two -balls joined by a -handle form one -ball, , and the total is the boundary connected sum of with a disc along the point of at which the second foot lands.
By [F3] the cancelling pair may be deleted from ; equivalently the boundary connected sum with a disc is again, so relative to . The attaching-belt matrix is not defined for (the definition requires ), so this endpoint case is handled directly by the geometric criterion.
A cancelling one-two handle pair on a surface
Example
Assume . In dimension start with a closed disc (a -handle) and attach a -handle (a band) along two disjoint intervals of , with attaching orientations chosen so the disc orientation extends over the band; the result is an annulus whose two boundary circles each contain exactly one point of the belt sphere of the band. Attach a -handle (a disc) along one boundary circle. Then the attaching sphere of the -handle meets the belt sphere of the -handle in exactly one point, the pair is geometrically cancelling, and . The transverse count of the unique intersection point is mod and has local sign ; the matrix definition of this page is stated for and does not cover the endpoint case , which is handled here directly by the geometric criterion.
Facts & Assumptions
Given: Dimension : a closed disc as a -handle, a -handle (band) attached along two disjoint intervals of , with attaching orientations chosen so the disc orientation extends over the band, and then a -handle (disc) attached along one boundary circle of the resulting annulus.
Attaching a smooth handle with corner rounding and K handle core cocore attaching region and belt sphere: the -handle has attaching region and outgoing region , so its belt sphere in the outgoing boundary is a -sphere ; the -handle has attaching sphere .
Geometrically cancelling adjacent handle pair: in dimension and the attaching sphere of the -handle is a circle and the belt sphere of the -handle is a -sphere in the middle boundary (the two boundary circles of the annulus); they meet transversely in isolated points, and the pair is geometrically cancelling when exactly one point of the belt sphere lies on the attaching circle.
Handle cancellation, The local oriented intersection sign, The mod 2 intersection number and The Axiom of Countable Choice (): assume ; a geometrically cancelling pair may be deleted, so the total is diffeomorphic to the initial disc; the unique transverse intersection has local sign and mod-2 count .
Verification
Given: The configuration of the statement.
Attaching a band to along two disjoint boundary intervals gives the annulus , whose two boundary circles each contain exactly one of the two points of the belt sphere of the band: the two points are the end points of the cocore arc , one on each boundary circle.
Attaching the -handle along one boundary circle makes the attaching circle meet the belt sphere in exactly one point, and the intersection is isolated and automatic in these dimensions; by [F2] the pair is geometrically cancelling, with the transverse count equal to modulo and local sign by [F3].
By [F3] the pair cancels: relative to the incoming boundary. The attaching-belt matrix is not defined here: its definition requires , and with and we have , so the endpoint case is handled directly by the geometric criterion.
A handle slide realizes an elementary row operation
Example
Assume . In dimension start from the -handle and attach the standard -handle along the equatorial embedding, so that and the middle boundary is with belt sphere of a -sphere. Attach two -handles to along embedded circles with disjoint images, for instance and a small circle in a coordinate ball disjoint from , with product/local framings. They are chosen so that meets transversely in exactly one point and is disjoint from . Indexing rows by the -handles and the single column by the -handle, the attaching-belt matrix is the column over for suitable orientations, respectively over . Slide the -handle over . The total -manifold is unchanged by the slide, while the same column becomes over , respectively over : exactly the elementary row operation .
Facts & Assumptions
Given: Dimension ; the -handle with the standard -handle attached, middle boundary and belt sphere of ; two embedded circles with disjoint images, with meeting transversely in exactly one point and ; the -handles attached to along with chosen framings.
Attaching a smooth handle with corner rounding and K handle core cocore attaching region and belt sphere: in dimension a -handle has attaching region , outgoing region and belt sphere ; a -handle has attaching sphere and attaching region .
The standard complementary pair fills a ball: attaching the standard -handle to along the equatorial embedding gives with outgoing boundary , and up to this diffeomorphism the belt sphere of is the fiber sphere .
Attaching-belt intersection matrix of adjacent-index handles: for and the matrix is defined (), its rows are the -handles and its columns the -handles , and is the intersection number of the attaching sphere of with the belt sphere of in the middle boundary, over or according to the orientations.
The oriented intersection number, The local oriented intersection sign and The mod 2 intersection number: transverse complementary-dimensional intersections are finite with local signs ; a single transverse point of a circle with a -sphere has entry over and over , and disjoint spheres have entry .
Handle slide of one k handle over another and Handle slides preserve the relative diffeomorphism type: assume ; a slide of over is defined here (index , middle boundary of dimension , so ), replaces the attaching circle of by the band sum with a framed parallel copy of the attaching circle of , and leaves the total -manifold unchanged.
Handle slides act by elementary basis change on handle chains and Elementary matrix operations are realized by handle slides: assume ; under the disk-push diffeomorphism together with its specified lower-stage homotopy, the slid core satisfies in the handle chain group, so intersecting the fixed belt sphere with both sides gives , the elementary row operation ; this is case (ii) of the matrix-operation proposition because .
The Axiom of Countable Choice (): is assumed; it is used through [F5] and [F6].
Verification
Given: The configuration of the statement.
By [F1] and [F2] the middle boundary is and the belt sphere of is , a -sphere; the -handles are attached along the circles , so the matrix is the column with entries . By hypothesis meets in exactly one transverse point and is disjoint from , so by [F4] the column is over for suitable orientations and over .
Slide the -handle over ; the move is legitimate in the range of [F5], the total -manifold is unchanged, and the slid attaching circle has core class by [F6]. Countable Choice enters only through the slide and basis-change suppliers [F5] and [F6].
Recomputing the same column with the slid handle, over , respectively over : the column becomes , respectively , which is exactly the elementary row operation on the matrix.
The construction is legitimate in the range of both the matrix definition and the slide: with and one has and , so the example realizes case (ii) of Elementary matrix operations are realized by handle slides in the lowest dimension in which the row operation is available.
Algebraic intersection one with three geometric points
Statement refuted
Assume . In the standard -dimensional model whose middle boundary is with belt sphere of the -handle, let be the -sphere obtained from by a finger move across . Then meets transversely in exactly three points with local signs ; the oriented intersection number is and the mod-2 number is , while the attaching sphere and the belt sphere do not meet in one point. Hence the attached -handle and the -handle form a pair whose matrix entry is a unit but which is not geometrically cancelling: the cancellation theorem does not apply, although reversing this specific finger isotopy removes the extra pair. A general algebraic-to-geometric conversion is a separate Whitney-trick issue with additional hypotheses.
Facts & Assumptions
Given: The standard -dimensional model with middle boundary , belt sphere , and a -sphere obtained from by a finger move across .
Attaching-belt intersection matrix of adjacent-index handles and K handle core cocore attaching region and belt sphere: the belt sphere of the -handle in the middle boundary is a circle, and the attaching sphere of a -handle attached to is a -sphere; the matrix entry is the oriented, respectively mod-2, intersection number of the attaching sphere with the belt sphere.
Algebraic cancellation does not yet give geometric cancellation: on there are an embedded -sphere and an embedded circle meeting transversely in exactly three points with local signs , so that the oriented intersection number is although the geometric intersection has three points; the configuration is realized with the sphere as the attaching sphere of a -handle and the circle as the belt sphere of the -handle in the standard model.
The oriented intersection number, The local oriented intersection sign and The mod 2 intersection number: the oriented number is the sum of local signs over the transverse intersection, and the mod-2 number is its cardinality modulo two.
Geometric cancellation is a unit entry in the handle matrix and Handle cancellation: a single transverse point gives a unit entry, and only then does the cancellation theorem apply; a unit entry does not by itself supply a single geometric intersection point.
The Axiom of Countable Choice (): is assumed; it is used through the intersection-number and cancellation suppliers.
Counterexample
Given: The configuration of the statement.
The model has middle boundary with belt sphere of the -handle; the sphere obtained by the finger move meets transversely in exactly three points with local signs .
By [F3] the oriented intersection number is the sum , a unit in , and the mod-2 number is the cardinality modulo , a unit in ; the geometric intersection set has three points and the two spheres do not meet in one point.
Reading the sphere as the attaching sphere of a -handle attached to and the circle as the belt sphere of the -handle, [F1] gives matrix entry : a unit entry, but not a geometrically cancelling configuration. Hence the attaching spheres do not satisfy the single-point hypothesis of [F4], the cancellation theorem does not apply, and no single-point conclusion follows from the matrix alone. This inserted finger pair can be removed by the inverse finger isotopy of [F2].
Adjacent-index handles with zero intersection do not cancel
Statement refuted
Assume . In dimension start from , attach the standard -handle along two disks of , so that is a solid torus, and attach a -handle along an embedded circle that bounds a closed disk in disjoint from the belt sphere of ; such a circle is disjoint from the belt sphere, so the matrix entry is over both and . Then the pair is not geometrically cancelling and does not cancel: has fundamental group , because the attaching circle is null-homotopic in the solid torus and the relation it adds is trivial, while is simply connected. Hence no diffeomorphism relative to the lower stage removes the pair.
Facts & Assumptions
Given: In dimension the manifold with a -handle attached along two disks of , giving , and a -handle attached along an embedded circle that bounds a closed disk in disjoint from the belt sphere of .
Attaching a smooth handle with corner rounding and K handle core cocore attaching region and belt sphere: attaching a -handle to along two disks of the boundary and rounding the corner gives the solid torus , whose boundary is a torus; the belt sphere of the -handle is the meridian of that torus, a -handle attaches along a circle, and a circle in the boundary that bounds a disk there is null-homotopic in the solid torus.
Attaching-belt intersection matrix of adjacent-index handles and Geometric cancellation is a unit entry in the handle matrix: the matrix entry is the intersection number of with the belt sphere; if the two are disjoint the entry is over both and , whereas a geometrically cancelling pair would have a unit entry.
Seifert–van Kampen identifies the fundamental group with a group pushout: for two open path-connected sets with path-connected overlap, the fundamental group is the pushout of the two groups over the overlap group. A collar thickening of an attached -handle gives such a cover with the old manifold and handle as deformation retracts, and overlap retracting onto the attaching annulus .
Handle cancellation and Geometrically cancelling adjacent handle pair: a cancelled pair may be deleted, so if cancelled then would be diffeomorphic to relative to the lower stage, in particular simply connected.
The Axiom of Countable Choice (): is assumed; it is used through [F3] and [F4].
Counterexample
Given: The configuration of the statement.
Take with its boundary torus ; the belt sphere of the -handle is the meridian . The attaching circle bounds a closed disk in that avoids , so is disjoint from the belt sphere and, bounding a disk in the boundary torus, is null-homotopic in .
Since the two circles are disjoint, the attaching-belt matrix entry is over both and ; in particular the pair is not geometrically cancelling, because a geometrically cancelling pair has a unit entry by [F2].
Thicken the old stage and handle slightly across their seam to obtain open path-connected sets , with , , and . By [F3], is the pushout of ; the map into is represented by , which is null-homotopic by step 1.1. The pushout is therefore . The product contraction gives , so . This is a direct handle-gluing argument and does not assume an unspecified one-critical-point Morse presentation.
By [F4] a cancellation of the pair would give a diffeomorphism relative to the lower stage, hence an isomorphism of fundamental groups , which is impossible. Therefore the pair does not cancel, even though both the algebraic and the geometric intersection counts are zero.
Consequently the single-point criterion of the cancellation theorem cannot be replaced by a count-only condition: an adjacent-index pair with vanishing (algebraic and geometric) intersection need not cancel, and no diffeomorphism relative to the lower stage removes the pair.
Sources
- C. T. C. Wall, Differential Topology (Cambridge Studies in Advanced Mathematics 156; complete PDF)
- Wolfgang Lück, A Basic Introduction to Surgery Theory (ICTP lecture notes; complete author PDF)
- John Milnor, Lectures on the h-Cobordism Theorem (notes by L. Siebenmann and J. Sondow; scanned edition with text layer)