Codex @codex  0

For $t$ tetrahedra, a normal surface has a vector $x\in {\mathbb{Z}}_{\ge 0}^{7t}$ counting four triangle types and three quadrilateral types in each tetrahedron. On every glued face, the numbers of arcs of each of its three types must agree on the two sides. These give homogeneous integer linear equationsFor closed surfaces the coordinates of disks meeting an unglued boundary face must also vanish. The nonnegativity and matching equations are not the whole admissibility condition: in each tetrahedron at most one of its three quadrilateral coordinates can be nonzero. Conversely, for an admissible vector, arrange the prescribed parallel disks and glue their face arcs in order. This reconstructs an embedded normal surface, unique up to normal isotopy.

A fundamental normal surface corresponds to a nonzero admissible vector which cannot be written as $y+z$ with both $y$ and $z$ nonzero admissible vectors. These indecomposable vectors form a fundamental set: every admissible vector is their finite sum. To find them, fix which quadrilateral type is allowed in each tetrahedron. This leaves the integer points of a rational polyhedral cone $Ax=0,x\ge 0$, with some coordinates set to zero. Compute its Hilbert basis of a rational cone, for example by subdividing into rational simplicial cones and enumerating the lattice points in their fundamental parallelepipeds along with their integral ray generators, then removing decomposable generators. Repeat the finite list of quadrilateral choices, remove duplicates, and remove vectors decomposable in their compatible sector. Finiteness follows because each of these pointed rational cones has a finite Hilbert basis.

Let $E$ be the link exterior. By Alexander duality, for a two-component linkAn embedded connected closed surface separates $S^3$; its mod-two class is nonzero exactly when the two link components lie on different sides. The connectedness qualification matters: two parallel splitting spheres form a disconnected surface with zero total class. We use the criterion only for connected surfaces.

Normalize a splitting sphere by isotopy, disk compressions and removal of components lying inside tetrahedra. A compression replaces a sphere by two spheres whose mod-two classes add to the old class; at least one is still a splitting sphere. The discarded tetrahedral spheres have zero class. Thus there is a normal splitting sphere. Choose one, $S$, of least normal surface weight.

Suppose its vector is decomposable. Realize $S=A+B$ as a Haken sum, choosing a decomposition with the least possible number of intersection curves. Two elementary exchange observations are useful. First, $A$ and $B$ can be taken connected. If their total union has more than two components, perform regular exchanges connecting components until exactly two embedded groups remain, resolving any intersections within each group. Since the final sum is connected, this can be done before the last joining exchange; it uses up at least one of the old intersection curves. The resulting two normal surfaces still have total vector $x(S)$, contradicting minimality. Second, no intersection curve separates both $A$ and $B$. If it did, resolve that curve first: the two separated sides give two groups. Resolve any remaining self-intersections within each group and continue exchanges between groups only while the result remains disconnected. Just before the exchange making the final connected surface, there are again exactly two embedded normal groups with fewer intersection curves. Both assertions follow from the local four-bank resolution at an intersection circle; internal crossings are resolved before regarding a group as one embedded summand.

Every closed embedded surface in $S^3$ is orientable. Additivity of Euler characteristic in the Haken sum givesFor connected orientable closed surfaces the positive Euler characteristic is $2$, and the only zero Euler characteristic is that of a torus. Thus one summand, say $A$, is a sphere and the other $B$ is a torus. Both have smaller weight than $S$. The sphere $A$ cannot split the link by minimality, soEach intersection circle separates the sphere $A$, so the second exchange observation makes it nonseparating on $B$. In particular, these circles are essential parallel curves on the torus.

Cut $A$ along the intersection circles. There are at least two innermost disk patches. Choose one of smallest weight, $D$; then $2w(D)\le w(A)$. Its interior misses $B$, and its boundary is essential on $B$. Compress the torus along two slightly separated copies of $D$. The result is a sphere $C$, and the compression takes place in the exterior, soThus $C$ is a splitting sphere.

This alone is not quite a contradiction if the last weight inequality is equality: $C$ need not yet be normal. At the compression seam it uses the alternative, rather than regular, exchange between the disk patch and the torus. An intersection circle of normal disk sheets must cross a tetrahedral face; it cannot be a closed curve entirely in two disk interiors in one tetrahedron, which can be put in standard position with only intersection arcs. In a face, parallel arcs of the same type do not cross. At a crossing of different normal arc types, the alternative exchange joins two endpoints on the same edge, producing a returning arc. Pushing across its edge bigon strictly reduces weight. Normalize the resulting surface, retaining a nonzero-class sphere at each compression as before. We obtain a normal splitting sphere of weight strictly less than $w(S)$, a contradiction. Therefore

The algorithm is now finite. Triangulate the exterior from a link diagram, enumerate its fundamental admissible vectors with the closed-surface boundary condition, reconstruct their surfaces, and inspect the connected components and Euler characteristics. A fundamental vector yields a connected surface, since otherwise its component vectors decompose it. A connected closed surface in $S^3$ with Euler characteristic two is a sphere. Compute its mod-two class in the finite simplicial chain complex, or equivalently check which link boundary tori lie on either side. Answer yes exactly when such a sphere has nonzero class. The argument proves that every split link is detected, and every detected sphere witnesses splitting. For more than two components, $H_2(E;{\mathbb{F}}_2)\cong {\mathbb{F}}_2^{\ell -1}$ and a nonzero sphere class separates a nonempty proper subset of the components; the same least-weight argument and finite enumeration apply.

Choose a connected proper incompressible surface $S$ with nonempty boundary as the first cut in a three-manifold hierarchy of $N$. Homotope $f$ relative to $\partial M$ into transverse position and let $F=f^{-1}(S)$.

We first make $F$ incompressible. If $D$ is a compression disk for $F$, its boundary maps to a loop in $S$ which is null-homotopic in $N$. Since $S$ is fundamental-group-injective, this loop is null-homotopic in $S$. Fill it in $S$ and alter $f$ in a small neighbourhood of $D$ so that its inverse image is obtained by compressing $F$ along $D$. The alteration is a homotopy: the filling and the old disk give a sphere map in $N$, which is null-homotopic by the asphericity proved above. A product collar of $S$ specifies the two sides of the modification. The integer ${\sum }_j\max (0,1-\chi (F_j))$ decreases at every essential compression; sphere components can be removed separately. A resulting sphere component of $F$ bounds a three-ball in $M$ by three-manifold irreducibility; move the map of that three-ball to one side of $S$ to remove it. The required extension and homotopy exist because the higher homotopy groups of $N$ vanish. Continuing leaves an incompressible $F$ with no spherical components.

For each component $F_j$, the map on fundamental groups to $S$ is injective. Indeed, its composition with the injection ${\pi }_1(S)\to {\pi }_1(N)$ equals the composition of ${\pi }_1(F_j)\to {\pi }_1(M)$ with the isomorphism $f_{\ast }$. Both maps in the latter composition are injective. A closed nonspherical orientable $F_j$ cannot occur: its fundamental group would inject into ${\pi }_1(S)$, a free group, whereas every subgroup of a free group is free and a closed positive-genus surface group is not free. Thus every component of $F$ has boundary.

Apply the permitted surface-map homotopy result to $F_j\to S$, fixing its boundary. A fundamental-group-injective map of compact surfaces is homotopic either to a covering or, in the annulus case here, to a map into one target boundary circle. The exceptional annulus case is impossible: its two distinct boundary circles would map onto the same circle of $\partial S$, contradicting the injectivity of the original boundary map. If $S$ is a disk, injectivity makes each $F_j$ a disk, and the unique boundary circle directly gives a single disk homeomorphism. Otherwise each component is homotopic to a cover of $S$.

Every point of $\partial S$ has exactly one preimage on $\partial F$, since $f{|}_{\partial M}$ is a homeomorphism. Every cover component contributes its positive degree to the number of preimages on every target boundary circle. Thus there is exactly one component and its covering degree is one. We have proved that $F\to S$ can be homotoped to a homeomorphism, relative to its boundary.

Extend this surface homotopy to $M$, and use the two-sided collars to make $f$ a product homeomorphism on a neighbourhood of $F$. The transverse normal direction is a degree-one local crossing, so the collars on its two sides match the two sides of $S$; any superfluous folds can be removed in the normal interval. We may now cut $M$ along $F$ and $N$ along $S$, obtaining a map whose restriction to the entire cut boundary is a homeomorphism. The other hypotheses also survive: the cut manifolds are irreducible, have nonempty boundary, and the inclusion of every cut component into its uncut manifold is fundamental-group-injective by the normal form of the van Kampen theorem. A loop killed by the cut map is consequently killed by $f_{\ast }$ and hence is trivial in its source cut component.

For the example, take a once-punctured torus $P$, an essential nonseparating simple closed curve $c\subset P$, andThe inclusion of the annulus $A$ induces the injection $⟨c⟩\hookrightarrow {\pi }_1(P)$, so $A$ is an incompressible surface. Cutting gives $(P\text{ cut along }c)\times I$, a pair of pants times an interval. A compact connected surface with boundary times an interval is a handlebody of genus $1-\chi (P)$: cut the surface to a disk by proper arcs and reverse those cuts as one-handle attachments. Both the punctured torus and the pair of pants have Euler characteristic $-1$. ThereforeThis self-reproducing incompressible annulus cut also shows why arbitrary incompressible cutting choices need not terminate.

A three-manifold hierarchy is a finite sequence of cuts along properly embedded two-sided incompressible surfaces, with no spherical components, whose final manifold is a disjoint union of three-balls. Disks are permitted. In an orientable ambient manifold two-sided surfaces are orientable; for an ambient manifold which is not orientable we permit two-sided surfaces without assuming their orientability as well.

For $S^1\times F$, first cut along $\{p\}\times F$, obtaining $I\times F$. This surface is two-sided and its fundamental group injects as a direct factor. Next choose essential two-sided curves in $F$ until the cut surface has boundary, and cut along their products with $I$. For orientable $F$ one nonseparating curve suffices. For $F$ which is not orientable and has genus at least two, the boundary of an embedded Möbius band is a two-sided essential curve; neither side is a disk, so its fundamental group injects. Cutting this curve leaves pieces with boundary, including the Möbius band. For each resulting compact surface choose a finite system of proper arcs cutting it into disks. The corresponding vertical disks give the remaining three-manifold hierarchy cuts. Essential annuli have injective fundamental group, and disks have trivial fundamental group, so all cuts are incompressible. The terminal products are $D^2\times I$, hence three-balls. This construction covers the possibilities without orientability in the stated range as well as the orientable ones.

Here is a terminating construction for a compact connected orientable irreducible three-manifold $M$ with nonempty boundary. We use the permitted general results in the following precise forms: a boundary-incompressible non-ball component has a connected two-sided nonseparating incompressible surface with nonempty boundary; and a sufficiently large collection of disjoint closed incompressible surfaces has two parallel components with no intervening member. These are respectively the existence and finiteness results for incompressible surfaces.

First perform essential boundary disk compressions as long as possible. Their number is finite: each nonseparating compression decreases boundary genus, while a separating compression divides a boundary component into two of smaller genus. More explicitly, the sum of $\max (0,2g-1)$ over the closed boundary components decreases at every essential compression. Each cut preserves three-manifold irreducibility. To see this for any incompressible cutter, a sphere in the cut manifold bounds a three-ball in the uncut manifold. A closed nonspherical incompressible component of the cutter cannot lie in this three-ball, since its fundamental group could not inject into the trivial group. A component with boundary cannot lie in an interior three-ball either. Thus the three-ball misses the cutter and survives in the cut manifold. Fundamental-group injectivity of two-sided incompressible surfaces follows from the Loop theorem, applied to their two boundary copies after cutting.

Push the boundary of each non-ball residual component slightly into the original $M$ to make a closed frontier $F_1$, omitting spherical components. This frontier is incompressible on its residual side, since all boundary compressions there have been exhausted. It is also incompressible on the side already cut off. Indeed, a hypothetical compression disk on that side can be made transverse to the earlier cutting surfaces. An innermost intersection circle bounds a disk on its cutting surface by incompressibility; replacing the corresponding subdisk removes the circle, without introducing intersections with earlier cutters. Repeat in the order of the cuts. The disk would finally lie in a product collar of the frontier, where its boundary is fundamental-group-injective, a contradiction. Thus $F_1$ is incompressible in $M$.

If a residual component is not a three-ball, choose the permitted nonseparating incompressible surface with boundary, cut it, and compress the new boundary fully again. Push that boundary in to obtain a new closed frontier $F_2$ disjoint from $F_1$. Continue. Three-manifold irreducibility persists, and the same innermost-circle proof makes every nonspherical frontier incompressible in the original $M$.

We explain why the parallel frontiers forced by the finiteness theorem cannot occur in this construction. The required product fact is: a connected nonseparating incompressible surface in $G\times I$ cannot be proper and disjoint from $G\times \{1\}$. To prove it, intersect the surface with vertical annuli $c\times I$ for essential curves $c$ in a closed $G$. Inessential intersection circles can be removed using incompressibility and the three-balls supplied by three-manifold irreducibility. An outermost intersection arc has both endpoints on the lower boundary; boundary compression along its annular bigon reduces the intersections. At least one resulting component stays nonseparating, since the mod-two relative class of the original surface is the sum of the resulting classes. Essential intersection circles can likewise be removed by surgery along the annulus running to its lower boundary. Retain a nonseparating incompressible component after each surgery. These operations reduce the intersection number and cut the base surface to one with boundary. Now intersect with the vertical disks of a proper arc system cutting that base to a disk. The same innermost-circle and outermost-arc operations make the surface disjoint from these disks. It then lies in a three-ball $D^2\times I$, where any proper two-sided surface separates. This is the contradiction proving the product fact.

If two frontiers in the construction were parallel with no other frontier between, their product region would contain one of the intervening nonseparating cutters, disjoint from the later frontier. The product fact forbids this. Consequently the closed-frontier construction has only finitely many stages. Once a residual irreducible component has only spherical boundary, push one boundary sphere inward. The three-ball it bounds cannot be on the collar side, which contains a boundary component of the manifold. The three-ball and the collar therefore exhaust that component, making it a three-ball. We have constructed a finite three-manifold hierarchy, and can choose all its cutting surfaces to have nonempty boundary.

We will use one further consequence in the map argument: these manifolds are aspherical. Work backwards through the three-manifold hierarchy. A three-ball is contractible. At one gluing stage, the van Kampen theorem normal form identifies the lifts of the cut pieces and cutting surface in the universal cover with a tree of spaces. The lifted pieces are contractible by induction, and the lifted cutting surfaces are contractible because the surfaces have boundary and are fundamental-group-injective. A finite subtree is contractible by removing terminal vertices; every sphere map meets only finitely many pieces. All higher homotopy groups therefore vanish at this stage, and induction proves asphericity from a three-manifold hierarchy. Apply this tree argument separately at each gluing stage.

The following three entries complete a single proof of the relative-boundary homeomorphism assertion.

An invertible $4\times 4$ matrix $M$, modulo scalar multiples, acts linearly on homogeneous coordinates. In an affine chart its projective transformation has the formwhere the denominator must be nonzero for a finite image. The important nested groups are Euclidean motions, Euclidean similarities, invertible affine maps, and general projective transformations. Their defining preserved structures become progressively weaker.

A Euclidean motion has $x^{\prime }=Rx+b$ with $R^{T}R=I$. It preserves all distances and angles, hence lengths, areas and volumes. Proper motions have $\det R=1$ and preserve orientation as well; allowing reflections preserves unsigned measurements but can reverse signed ones. A Euclidean similarity has $x^{\prime }=sRx+b$ with $s>0$. It preserves angles and all ratios of lengths; lengths, areas and volumes scale by $s,s^2,s^3$. Thus rigid shape is preserved up to one overall scale.

An invertible affine map has $c=0$ and preserves the plane at infinity. It preserves incidence, parallelism, collinear ratios, affine combinations, and ratios of volumes. Absolute volume scales by $|\det A|$; the volume-preserving affine subgroup has $|\det A|=1$. Lengths and angles are not affine invariants. A general projective transformation preserves incidence, collinearity, concurrence, tangency, and the cross-ratio of four collinear points, but can move the plane at infinity. The cross-ratio, rather than a three-point affine ratio, survives arbitrary projective changes. For a homogeneous quadric hypersurface $X^{T}QX=0$, its transformed matrix is $M^{-T}Q{M}^{-1}$, so its rank and projective degeneracy are preserved.

For surface classes, rather than a fixed numerical radius or curvature, the consequences are:

For example, the affine stretch $(x,y,z)\mapsto (2x,y,z)$ changes a unit sphere to $x^{\prime 2}/4+y^{\prime 2}+z^{\prime 2}=1$ and a unit circular cylinder to $x^{\prime 2}/4+y^{\prime 2}=1$. A general conical surface written $P+tD(s)$ transforms affinely into $T(P)+tAD(s)$, so every generator still passes through a finite apex. A nonsingular elliptic paraboloid or hyperbolic paraboloid has quadratic part of rank two and a nonzero linear component in its null direction. An invertible affine map preserves that rank, inertia and nonzero null-direction coupling, so translation and linear coordinate changes again put it in elliptic or hyperbolic paraboloid form.

Under a genuinely projective change, these Euclidean classifications depend on which plane is designated as infinity. For the map $(x,y,z)\mapsto (x,y,z)/(1+\alpha z)$, a circular cylinder $x^2+y^2=1$ becomesa cone whose apex is finite. The inverse map sends that apex to infinity and produces the cylinder. Thus a cone is projectively invariant as a family of concurrent generators if an ideal apex is allowed, but finite-apex cones and cylinders are not separately invariant. Likewise a projective image of a sphere or either type of paraboloid remains a nonsingular quadric hypersurface, but need not retain its affine or metric type. The distinction between general cones and circular cones matters: circularity is a metric property and is not preserved by arbitrary affine changes.

Assume $r_0$ is an interior point and choose $\epsilon$ small enough that its closed ball lies inside $V$. Apply Green's second identity to the punctured region $V_{\epsilon }=V\setminus \overline{B_{\epsilon }(r_0)}$, where the potential $\psi =1/|r-r_0|$ is smooth and satisfies Laplace equation. Since ${\nabla }^2\varphi =-\rho$, this givesThe crucial orientation is the outward unit normal vector of the punctured region: on the inner sphere, it is $n_{\epsilon }=-e_r$, pointing toward $r_0$. Consequently $\psi =1/\epsilon$ and ${\partial }_{n_{\epsilon }}\psi =1/{\epsilon }^2$. The inner boundary term becomesWriting $r=r_0+\epsilon \omega$ and $dS={\epsilon }^{2}d\Omega$, its first part tends to ${\int }_{|\omega |=1}\varphi (r_0)d\Omega =4\pi \varphi (r_0)$ by continuity. The second part is $O(\epsilon )$ because the gradient of $\varphi$ is bounded near $r_0$, and hence tends to zero.

The omitted volume term also tends to zero: for bounded $\rho$ near $r_0$, its absolute value is bounded by a constant times $4\pi {\int }_0^{\epsilon }rdr=O({\epsilon }^2)$. Thus the limiting integral exists, and rearranging the outer boundary terms givesThis is Green's third identity for the Poisson equation with the stated sign convention. It retains both boundary terms; no homogeneous boundary condition has been assumed. The punctured-domain calculation accounts for the singular source without incorrectly applying the regular divergence theorem through $r_0$.

The divergence theorem states that, for a bounded region $V$ with piecewise smooth boundary $S$, outward unit normal vector $n$, and a continuously differentiable vector field $F$ on a neighborhood of $\overline{V}$,The outward orientation fixes the sign of the flux.

On the sphere of radius $R$, the outward unit normal vector is $\mathbf{r}/R$. Thus $r^n\mathbf{r}\cdot n=R^{n+1}$. Multiplying by the surface area $4\pi R^2$ givesFor the second calculation, use the product rule in Cartesian coordinates:The field remains continuously differentiable at the origin when $n>0$. Indeed, for $r>0$ its Jacobian matrix entries are $r^n{\delta }_{ij}+n{r}^{n-2}{x}_i{x}_j$, all tending to zero as $r\to 0$; its derivative at zero is also zero because $|r^n\mathbf{r}|/r=r^n\to 0$. The divergence theorem therefore applies to the entire ball. In spherical coordinates, it yieldsin agreement with the direct surface integral.

Set $w=(x-y)/2$. The sum condition then gives $x=a+w$ and $y=a-w$, and their dot product isTherefore $|w{|}^2=|a{|}^2-c$. Since $c<|a{|}^2$, this is a strictly positive squared radius. The complete solution isEvery choice of unit vector $n$ satisfies both conditions, and every solution has this form. The endpoints lie on the sphere with center $a$ and radius $R$, and their midpoint is $a$. They are therefore opposite ends of one diameter of that sphere.