On Computability and Triviality of Well Groups
Discrete Comput Geom
On Computability and Triviality of Well Groups
Peter Franek 0
Marek Krcˇál 0
0 IST Austria , Am Campus 1, 3400 Klosterneuburg , Austria
The concept of well group in a special but important case captures homological properties of the zero set of a continuous map f : K → Rn on a compact space K that are invariant with respect to perturbations of f . The perturbations are arbitrary continuous maps within L ∞ distance r from f for a given r > 0. The main drawback of the approach is that the computability of well groups was shown only when dim K = n or n = 1. Our contribution to the theory of well groups is twofold: on the one hand we improve on the computability issue, but on the other hand we present a range of examples where the well groups are incomplete invariants, that is, fail to capture certain important robust properties of the zero set. For the first part, we identify a computable subgroup of the well group that is obtained by cap product with the pullback of the orientation of Rn by f . In other words, well groups can be algorithmically approximated from below. When f is smooth and dim K < 2n − 2, our approximation of the (dim K − n)th well group is exact. For the second part, we find examples of maps f, f : K → Rn with all well groups isomorphic but whose perturbations have different zero sets. We discuss on a possible replacement of the well groups of vector valued maps by an invariant of a better descriptive power and computability status. This is an extended version of a paper that is to appear that appeared in the proceedings of the Symposium on Discrete Geometry, 2015.
Mathematics Subject Classification 65H10 · 68U05 · 55S35 · 55Q55
1 Introduction
In many engineering and scientific solutions, a highly desired property is the resistance
against noise or perturbations. We can only name a fraction of the instances: stability
in data analysis [
7
], robust optimization [
3
], image processing [
17
], or stability of
numerical methods [
19
]. Some very important tools for robust design come from
topology, which can capture stable properties of spaces and maps.
In this paper, we take the robustness perspective on the study of the solution set of
systems of nonlinear equations, a fundamental problem in mathematics and computer
science. Equations arising in mathematical modeling of real problems are usually
inferred from observations, measurements or previous computations. We want to
extract maximal information about the solution set, if an estimate of the error in
the input data is given.
More formally, for a continuous map f : K → Rn on a compact Hausdorff space
K and r > 0 we want to study properties of the family of zero sets
Zr ( f ) := {g−1(0) : f − g
≤ r },
where · is the max-norm with respect to some fixed norm | · | in Rn. The functions
g with f − g ≤ r (or f − g < r ) will be referred to as r -perturbations of f
(or strict r -perturbations of f , respectively). Quite notably, we are not restricted to
parameterized perturbations but allow arbitrary continuous functions at most (or less
than) r far from f in the max-norm.
Well Groups. Recently, the concept of well groups was designed to quantify
transversality of a map K → Y with respect to a subspace Y of Y . Following [
11
], we assume1
a subspace P of C (K , Y ) with a metric d. Any h ∈ P for which d(h, f ) ≤ r is called
an r -perturbation of f and we define Xr ⊆ K to be the union of g−1(Y ) over all
r -perturbations h of f . The inclusion h−1(Y ) → Xr induces a map in homology
jh : H (h−1(Y )) → H (Xr ), where H is a suitable homology theory. For r > 0, the
well group U (r ) is defined to be the intersection ∩ Im( jh ) over all r -perturbations h of
f . Informally, well group capture the robustness of intersection of f (K ) with Y via
means of homology, namely “homological properties” that are common to all h−1(Y )
for all r -perturbations h of f .
We will deal with the special but important case when Y = Rn, Y = {0}, and P is
the space of all continuous functions C (K , Rn) with the max-norm with respect to a
fixed norm on Rn. Then well group is a property of Zr ( f ) that captures homological
1 Our spaces K , Y are denoted by “X, A” in the original references, but we changed this notation as we
will use the symbols X and A for different purposes.
properties common to all zero sets in Zr ( f ). We enhance the theory to include a relative
case2 that is especially convenient in the case when K is a manifold with boundary.
Let B ⊆ K be a pair of compact Hausdorff spaces and f : K → Rn continuous.
f −1[0, r ] where | f | denotes the function x → | f (x )|; this is the smallest
Let X := | |
space containing zero sets of all r -perturbations h of f and coincides with Xr from
the notation of [
11
].
In the rest of the paper, for any space Z ⊆ K we will abbreviate the pair (Z , Z ∩ B)
by (Z , B) and, similarly for homology, H∗(Z , Z ∩ B)) by H∗(Z , B). Everywhere
in the paper we use homology and cohomology groups with coefficients in Z unless
explicitly stated otherwise. For brevity we omit the coefficients from the notation.
We define the j th well group U j ( f, r ) of f at radius r as the subgroup of H j ( X, B)
U j ( f, r ) :=
i∗
Im H j (Z , B) −→
H j ( X, B) ,
Z ∈Zr ( f )
where i∗ is induced by the inclusion i : Z → X and H refers to a convenient homology
theory of compact metrizable spaces that we describe below.3 This reduces to the
original definition whenever B = ∅. For a simple example of a map f with nontrivial
first well group see Fig. 1 on page 5.
Significance of Well Groups. We only mention a few of many interesting things mostly
related to our setting. The well group in dimension zero characterizes robustness of
solutions of a system of equations f (x ) = 0. Namely, ∅ ∈ Zr ( f ) if and only if
U0( f, r ) ∼= 0. Higher well groups capture additional robust topological properties of
the zero set such as in Fig. 1. Perhaps the most important is their ability to form well
diagrams [
11
]—a kind of measure for robustness of the zero set (or more generally,
robustness of the intersection of f with other subspace Y ⊆ Y ). The well diagrams
are stable with respect to taking perturbations of f .4
Homology Theory. For the foundation of well groups we need a homology theory on
compact Hausdorff spaces that satisfies some additional properties that we specify later
in Sect. 3. Roughly speaking, we want that the homology theory behaves well with
respect to infinite intersections. Without these properties we would have to consider
only “well behaved” perturbations of a given f in order to be able to obtain some
nontrivial well groups in dimension greater than zero. We explain this in more detail
also in Sect. 3. For the moment it is enough to say that the Cˇ ech homology can be
used and that for any computational purposes it behaves like simplicial homology.
2 Authors of [
4
] develop a different notion of relativity that is based on considering a pair of spaces (Y , Y0)
instead of the single space Y . This direction is rather orthogonal to the matters of this paper.
3 In [
4,11
], well groups were defined by means of singular homology. But then, once we allow arbitrary
continuous perturbations, to the best of our knowledge, no f : K → Rn with nontrivial U j ( f, r ) for j > 0
would be known. In particular, the main result of [4] would not hold. The correction via means of Steenrod
homology was independently identified by the authors of [
4
].
4 Namely, so called bottleneck distance between a well diagrams of f and f is bounded by f − f .
The stability does not say how well the well diagrams describe the zero set. This question is also addressed
in this paper.
A
X
f = r
f = − r
In Sect. 3 we explain why using singular homology would make the notion of well
groups trivial.
A basic ingredient of our methods is the notion of cap product
: H n ( X, A) ⊗ Hk ( X, A ∪ B) →
Hk−n ( X, B)
between cohomology and homology. We refer the reader to [28, Sect. 2.2] and [18,
p. 239] for its properties and to Appendix 4 for its construction in Cˇ ech (co)homology.
Again, it behaves like the simplicial cap product when applied to simplicial complexes.
For an algorithmic implementation, one can use its simplicial definition from [28].
1.1 Computability Results
The main result introduced in this section is the computability of certain subgroup of
the well group that in many cases coincides with the full well group. A core ingredient
is certain cohomology class that we will call primary obstruction. Intuitively, it assigns
to each n-cell in a triangulation of X the intersection number with arbitrary zero set
in Zr ( f ). In the example illustrated in Fig. 1, it would be an element assigning ±1
to each vertical edge of a triangulation of the rectangle X that connects the “lower
face” with the “upper face”, indicating that such edge intersects the zero set at least
once. If X is a compact oriented connected manifold, there is a well-defined Poincaré
dual of the primary obstruction: this is a homology class of a cycle that intersects the
cells of X as prescribed by the primary obstruction. We will show that not only the
Poincaré dual but, even more generally, any cap product5 of the primary obstruction
is contained in the well group of f .
5 The Poincaré dual of a cohomology class ϕ can be in fact defined as the cap product [X ] ϕ, where [X ]
is so called fundamental homology class of the manifold X . One advantage of the cap product is that it is
defined for any simplicial complex X .
Computer Representation. To speak about computability, we need to fix some computer
representation of the input. Here we assume the simple but general setting of [
14
],
namely, K is a finite simplicial complex, B ⊆ K a subcomplex, f is simplexwise
linear with rational values on vertices6 and the norm | · | in Rn can be (but is not
restricted to) 1, 2 or ∞ norm.
Previous Results. The algorithm for the computation of well groups was developed
only in the particular cases of n = 1 [
4
] or dim K = n [
8
]. In [
14
] we settled the
computational complexity of the well group U0( f, r ). The complexity is essentially
identical to deciding whether the restriction f |A : A → Sn−1 can be extended to
X → Sn−1 for A = | f |−1(r ), or equivalently, A = f −1(Sn−1). The extendability
problem can be decided as long as dim K ≤ 2n − 3 or n = 1, 2 or n is even. On the
contrary, the extendability of maps into a sphere—as well as triviality of U0( f, r )—
cannot be decided for dim K ≥ 2n − 2 and n odd, see [
14
].7 In this paper we shift our
attention to higher well groups.
Higher Well Groups—Extendability Revisited. The main idea of our study of well
groups is based on the following. We try to find r -perturbations of f with as small
zero set as possible, that is, avoiding zero on X for X ⊆ X as large as possible. It
is shown in [13, Lem. 3.1] that for each strict r -perturbation g of f we can find an
extension e : X → Rn of f | A with g−1(0) = e−1(0) and vice versa. Thus equivalently,
we try to extend f |A to a map X → Sn−1 for X as large as possible. The higher
skeleton8 of X we cover, the more well groups we kill.
Observation 1 Let f : K → Rn be a map on a compact space. Assume that the pair
of spaces A ⊆ X defined as | f |−1(r ) ⊆ | |
f −1[0, r ], respectively, can be triangulated
and dim X = m. If the map f | A can be extended to a map A ∪ X (i−1) → Sn−1 then
U j ( f, r ) is trivial for j > m − i .
Assume, in addition, that there is no extension A∪ X (i) → Sn−1. By the connectivity
of the sphere Sn−1, we have i ≥ n. Does the lack of extendability to X (i) relate to higher
well groups, especially Um−i ( f, r )? The answer is yes when i = n as we show in our
computability results below. On the other hand, when i > n, the lack of extendability
is not necessarily reflected by Um−i ( f, r ). This leads to the incompleteness results we
show in the second part of the paper.
The Primary Obstruction. The lack of extendability of f |A to the n-skeleton is
measured by the so called primary obstruction that is defined in terms of cohomology
theory as follows. We can view f as a map of pairs ( X, A) → ( Bn , Sn−1) where Bn
is the ball bounded by the sphere Sn−1 := {x : |x | = r }. Then the primary obstruction
6 We emphasize that the considered r -perturbations of f need not be neither simplexwise linear nor have
rational values on the vertices.
7 We cannot even approximate the “robustness of roots”: it is undecidable, given a simplicial complex K
and a simplexwise linear map f : K → Rn , whether there exists ε > 0 such that U0( f, ε) is nontrivial or
whether U0( f, 1) is trivial. The extendability can always be decided for n even, however, the problem is
less likely tractable for dim K > 2n − 2.
8 The i -skeleton X (i) of a simplicial (cell) complex X is the subspace of X containing all simplices (cells)
of dimension at most i .
φ f is equal to the pullback f ∗(ξ ) ∈ H n ( X, A) of the fundamental cohomology class
ξ n ∈ H n ( Bn , Sn−1). 9
Theorem 1 Let B ⊆ K be compact spaces, f : K → Rn be continuous, r > 0,
and assume the choice of Cˇ ech (co)homology theory. Let | f |−1[0, r ] and | f |−1(r )
be denoted by X and A, respectively, and φ f be the primary obstruction. Then
φ f Hk ( X, A ∪ B) is a subgroup10 of Uk−n ( f, r ) for each k ≥ n.
After the proof of Theorem 1 we will show that the groups φ f Hk ( X, A ∪ B)
are not only subgroups of well groups but they also fit into a natural sequence of
homomorphisms for varying r . The resulting persistence module is then a submodule
of so called well module.
Our assumptions on computer representation allow for simplicial approximation
of X, A and f . The pullback of ξ n ∈ H n ( Bn , Sn−1) and the cap product can be
computed by standard formulas. This together with more details worked out in the
proof in Sect. 3 gives the following.
Theorem 2 Under the assumption on computer representation of K , B and f as
above, the subgroup φ f Hk ( X, A ∪ B) of Uk−n ( f, r ) (as in Theorem 1) can be
computed. The running time is polynomial when the dimension of K is fixed.
The Gap Between Uk−n and φ f Hk ( X, A ∪ B). There are maps f with φ f trivial but
nontrivial U0( f, r ).11 But this can be detected by the above mentioned extendability
criterion. We do not present an example where Uk−n ( f, r ) = φ f Hk ( X, A ∪ B)
for k − n > 0, although the inequality is possible in general. In the rest of the paper
we work in the other direction to show that there is no gap in various cases and various
dimensions.
An important instance of Theorem 1 is the case when X can be equipped with the
structure of a smooth orientable manifold.
Theorem 3 Let f : K → Rn and X, A be as above. Assume that X can be equipped
with a smooth orientable manifold structure, A = ∂ X , B = ∅ and n + 1 ≤ m ≤ 2n − 3
for m = dim X . Then
Um−n ( f, r ) = φ f
Hm ( X, ∂ X ).
When m = n, the well group U0( f, r ) can be strictly larger than φ f Hn ( X, ∂ X )
but it can be computed.
We believe that the same claim holds when X is an orientable PL manifold. It
remains open whether the last equation holds also for m > 2n − 3. Throughout the
9 This is the global description of the primary obstruction as presented in [32]. It can be shown that
φ f depends on the homotopy class of f |A only. Another way of defining the primary obstruction is the
following. It is represented by the so-called obstruction cocycle z f ∈ Z n (X, A) that assigns to each
nsimplex σ ∈ X the element [ f |∂σ ] ∈ πn−1(Sn−1) ∼= Z [28, Chap. 3]. Through this definition it is not
difficult to derive that the map f |A can be extended to X (n) → Sn−1 if and only if φ f = 0, see also [28,
Chap. 3].
10 We give a precise definition of cap product between Cˇ ech homology and cohomology in Appendix 4.
11 This is the case for f : R4 → R3 given by f (x) := |x|η(x/|x|) where η : S3 → S2 is the Hopf map.
proof of Theorem 3, we will show that if g : K → Rn is a smooth r -perturbation of f
transverse to 0, then the fundamental class of g−1(0) is mapped to the Poincaré dual
of the primary obstruction. This also holds if B = ∅ and in all dimensions.
1.2 Well Groups U∗( f, r) are Incomplete as an Invariant of Zr ( f )
A simple example illustrating Theorem 3 is the map f : S2 × B3 → R3 defined by
f (x , y) := y with B3 considered as the unit ball in R3. It is easy to show that
for every 1-perturbation g of f and every x ∈ S2
there is a root of g in {x } × B3.
(1)
This robust property is nicely captured by (and can be also derived from) the fact
U2( f, 1) ∼= Z.
The main question of Sect. 4 is what happens, when the primary obstruction φ f is
trivial—and thus f |A can be extended to X (n)—but the map f |A cannot be extended
to whole of X . The zero set of f can still have various robust properties such as
(η1:).SI3t →is thSe2 icsaasehoomfof t:opSi2ca×llyBn4on→trivRia3l mdeafipnseudchbyasft(hxe,Hyo)p:f=m|ayp|.ηT(hye/|zye|r)owsehteoref
each r -perturbation g of f intersects each section {x } × B4, but unlike in the example
before, well groups do not capture this property. All well groups U j ( f, r ) are trivial
for j > 0 and,12 consequently, they cannot distinguish f from another f having only
a single robust root in X . We will describe the construction of such f for a wider
range examples.
In the following, Bqi will denote the i -dimensional ball of radius q, that is, Bqi =
{y ∈ Ri : |y| ≤ q}. We also emphasize that this and the following theorem hold for
arbitrary coefficient group of the homology theory H∗.
Theorem 4 Let i, m, n ∈ N be such that m − i < n < i < (m + n + 1)/2 and both
πi−1(Sn−1) and πm−1(Sn−1) are nontrivial. Then on K = Sm−i × B1i we can define
two maps f, f : K → Rn such that for all r ∈ (0, 1]
– f , f induce the same X = Sm−i × Bri and A = ∂ X and have the same well groups
for any coefficient group of the homology theory H∗ defining the well groups,
– but Zr ( f ) = Zr ( f ).
In particular, the property
for each Z ∈ Zr (.) and x ∈ Sm−i there exists y ∈ Bri such that (x , y) ∈ Z
is satisfied for f but not for f . Namely, Zε( f ) contains a singleton for each ε > 0.
12 Namely U2( f, r) ∼= 0 as is shown by the r-perturbation g(x, y) = f (x, y) − r x with the zero set
homeomorphic to the 3-sphere.
In Sect. 4 we discuss that the maps f and f are no peculiar examples but rather
typical choices given that the underlying space K is the solid torus Sm−i × Bi and
that both Zr (·) are nontrivial. Further we indicate that the same result holds for even
more realistic choice of the underlying space K = Bm and B = ∂ K . For the sake of
exposition, we chose the case where f is large on the boundary of K and we do not
need to consider nonempty B.
The Lack of Extendability Not Reflected by Um−i ( f, r ). The key property of the
example of Theorem 4 is that the maps f |A and f |A can be extended to the (i − 1)-skeleton
X (i−1) of X , for i > n. The difference between the maps lies in the extendability to
X (i). Unlike in the case when i = n, the lack of extendability is not reflected by the
well groups. The crucial part is the triviality of the well groups in dimension m − i
and13 this triviality holds in greater generality:
Theorem 5 Let f : K → Rn, B ⊆ K , X := | f |−1[0, r ] and A := | f |−1{r }. Assume
that the pair (X, A) can be finitely triangulated.14 Further assume that f |A can be
extended to a map h : A ∪ X (i−1) → Sn−1 for some i such that m − i < n < i <
(m + n)/2 for m := dim X . Then Um−i ( f, r ) = 0 for any coefficient group of the
homology theory H∗.
The proof is all delegated to Appendix 3 as its core idea is already contained in the
proof of Theorem 4. Note that in the setting of the last Theorem, U j is trivial for
j > m − i by Observation 1. One could ask the question of triviality in dimensions
smaller than m − i as well. Our favorite problem is the following one.
Problem 1 Let f be as in Theorem 5 and let i = n +1, that is, the primary obstruction
is trivial. Is it true that all well groups U j ( f, r ) for j ≥ (m − n + 2)/2 are trivial?
The bound j ≥ (m − n + 2)/2 is not known to be necessary (we only know that
the statement is not true for j = 1). But passing the bound seems to bring various
technical difficulties such as inapplicability of the Freudenthal suspension theorem.
Our Subjective Conclusion on Well Groups of Rn-Valued Maps. We find the problem
of the computability of well groups interesting and challenging with connections
to homotopy theory (see also Proposition 1 below). Moreover, well groups may be
accessible for non-topologists: they are based on the language of homology theory
that is relatively intuitive and easy to understand. On the other hand, well groups
may not have sufficient descriptive power for various situations (Theorems 4 and 5).
Furthermore, despite all the effort, the computability of well groups seems far from
being solved. In the following paragraph, we propose an alternative based on homotopy
and obstruction theory that addresses these drawbacks.
13 This dimension is somewhat important as all higher well groups are trivial by Lemma 7 and all lower
homology groups of X may be trivial as is the case in Theorem 4. On the other hand, Hm−i X, πi−1(Sn−1)
has to be nontrivial in the case when X is a manifold for the reasons following from obstruction theory and
Poincaré duality.
14 That is, there exist finite simplicial complexes A ⊆ X and a homeomorphism (X , A ) → (X, A).
1.3 Related Work
A Replacement of Well Groups of Rn -Valued Maps. In a companion paper [
13
], we
find a complete invariant for an enriched version of Zr ( f ). The starting point is the
surprising claim that Zr ( f )—an object of a geometric nature—is determined by terms
of homotopy theory.
Proposition 1 ([13, Thm. A]) Let f : K → Rn be a continuous map on a compact
Hausdorff domain, r > 0, and let us denote the space | f |−1[r, ∞] by Ar . Then the
set Zr ( f ) := {g−1(0) : g − f ≤ r } is determined by the pair (K , Ar ) and the
homotopy class of f |Ar in [ Ar , {x ∈ Rn : x ≥ r }] =∼ [ Ar , Sn−1].15
Note that since the well groups is a property of Zr ( f ), they are determined by the
pair (K , Ar ) and the homotopy class [ f |Ar ]. Thus the homotopy class has a greater
descriptive power and the examples from the previous section show that this inequality
is strict. If K is a simplicial complex, f is simplexwise linear and dim Ar ≤ 2n − 4
then [ Ar , Sn−1] has a natural structure of an Abelian group denoted by π n−1( Ar ).
The restriction dim Ar ≤ 2n − 4 does not apply when n = 1, 2 and16 otherwise we
could replace [ Ar , Sn−1] with [ Ar(2n−4), Sn−1] which contains less information but
is computable. The isomorphism type of π n−1( Ar ) together with the distinguished
element [ f |Ar ] can be computed essentially by [6, Thm. 1.1]. Moreover, the inclusions
As ⊆ Ar for s ≥ r induce computable homomorphisms between the corresponding
pointed Abelian groups. Thus for a given f we obtain a sequence of pointed Abelian
groups π n−1( Ar ) for r > 0. On such sequences, there is a natural distance function
(called the interleaving distance17) and it can be easily shown that the distance between
the sequences corresponding to a given functions f and f is bounded by f − f .
Thus after tensoring the groups by an arbitrary field, we get persistence diagrams (with
a distinguished bar) that will be stable with respect to the bottleneck distance and the
L∞ norm. The construction is detailed in [
13
].
The computation of the cohomotopy group π n−1( A) is naturally segmented into
a hierarchy of approximations of growing computational complexity. Therefore our
proposal allows for a compromise between the running time and the descriptive power
of the outcome. The first level of this hierarchy is the primary obstruction φ f . One
could form similar modules of cohomology groups with a distinguished element as we
did with the cohomotopy groups above. However, in this paper we passed to homology
via cap product in order to relate it to the established well groups. In the “generic”
case when X is a manifold no information is lost as from the Poincaré dual φ f [X ]
we can reconstruct the primary obstruction φ f back.
15 Here [Ar , Sn−1] denotes the set of all homotopy classes of maps from Ar to Sn−1, that is, the cohomotopy
group πn−1(Ar ) when dim Ar ≤ 2n − 4.
16 Note that for n = 1 the structure of the set [A, Sn−1] is very simple and for n = 2 we have [A, Sn−1] ∼=
H 1(A; Z) no matter what the dimension of Ar is.
17 The interleaving distance between the sequences πn−1(Ar ) and πn−1(Ar ), r > 0 is at most α whenever
there are families of homomorphisms of pointed Abelian groups ur : πn−1(Ar ) → πn−1(Ar+α) and
vr : πn−1(Ar ) → πn−1(Ar+α) such that the compositions vr+αur and ur+αvr give the corresponding
inclusion induced maps.
The Cap-Image Groups. The groups φ f Hk (X, A) (with B = ∅) has been studied by
Amit K. Patel and R. MacPherson under the name cap-image groups. In fact, his setting
is slightly more complex with Rn replaced by arbitrary manifold Y . Instead of the zero
sets, he considers preimages of all points of Y simultaneously in some sense. Although
his ideas have not been published yet, they influenced our research; the application of
the cap product in the context of well groups should be attributed to Patel.18
The advantage of the primary obstructions and the cap image groups is their
computability and well understood mathematical structure. However, the incompleteness
results of this paper apply to these invariants as well.
Verification of Zeros. An important topic in the interval computation community is
the verification of the (non)existence of zeros of a given function [27]. While the
nonexistence can be often verified by interval arithmetic alone, a proof of existence
requires additional methods which often include topological considerations. In the
case of continuous maps f : Bn → Rn, Miranda’s or Borsuk’s theorem can be used
for zero verification [
2,16
], or the computation of the topological degree [
9,15,21
].
Fulfilled assumptions of these tests not only yield a zero in Bn but also a “robust” zero
and a nontrivial 0th well group U0( f, r ) for some r > 0. Recently, topological degree
has been used for simplification of vector fields [29].
The primary obstruction φ f is the analog of the degree for underdetermined systems,
that is, when dim K > n in our setting. To the best of our knowledge, this tool has not
been algorithmically utilized.
2 Topological Preliminaries
In this section we introduce some definitions from algebraic topology that we need
throughout the proofs. The details can be found in standard textbooks, such as [
18,30
].
Simplicial Complexes and Cell Complexes. A simplicial complex is a collection of
simplices such that the intersection of any two of them is again a (possibly empty)
simplex in the collection. It can easily be topologized and the underlying space is
compact whenever the simplicial complex consists of finitely many simplices. The i th
skeleton X (i) of a simplicial complex X is the subcomplex consisting of all simplices
of dimension at most i . A simplicial map X → Y is a function that maps simplices of
X to simplices of Y . It again has a simple representation as a continuous map between
the underlying topological spaces.
A cell complex is an inductive description of a topological space X that consists
of skeleta X (0) ⊆ X (1) ⊆ · · · ⊆ X . The 0-skeleton X (0) is a discrete set and the
n-skeleton X (n) is defined to be the quotient space X (n−1) α Bαn/ ∼ where {Bαn}
is a collection of closed n-balls and the equivalence ∼ identifies points in ∂ Bαn with
points in X (n−1) via continuous attaching maps ϕα : Sn−1 → X (n−1). The balls Bαn
are called cells and the composition Bαn → X (n−1) α Bαn → X (n−1) α Bαn/ ∼ is
18 We originally proved that when K is a triangulated orientable manifold, the Poincaré dual of φ f is
contained in Um−n( f, r). Expanding the proof was not difficult, but the preceding inspiration of replacing
the Poincaré duality by cap product came from Patel. The cap product provides a nice generalization to an
arbitrary simplicial complex K .
called the characteristic map of the cell Bαn. All cell complexes occurring in this paper
consists of finitely many cells in which case this uniquely describes a topology of
X = i X (i).
Basic Operation on Spaces. For pointed topological spaces (X, x ) and (Y, y), the
wedge sum (X, x ) ∨ (Y, y) is the space (X Y )/{x , y}, that is, x and y are identified to
a single point. If the choice of the base points x , y is not important, we will simply write
X ∨ Y . The smash product X ∧ Y is defined to be the quotient space (X × Y )/(X ∨ Y ).
For pointed spheres Si and S j , their smash product Si ∧ S j is homeomorphic to the
pointed sphere Si+ j .
For a space X we define the cone over X to be the space C X := (X × [
0, 1
])/
(X × {1}) and the suspension Σ X to be C X/(X × {0}). For spheres, the identity
Σ Si Si+1 holds. The suspension of f : X → Y is the map Σ f : Σ X → Σ Y defined
by (x , t ) → [( f (x ), t )] for t ∈ [
0, 1
], which converts Σ into a functor Top → Top.
Homotopy Groups of Spheres. For a pointed topological space (X, x ), the i th
homotopy group of X is a group which underlying space is the set of homotopy classes
of maps (Si , ∗) → (X, x ) and the product is defined as follows. We realize the i
sphere as (Si , ∗) = (I i /∂ I i , ∂ I i /∂ I i ), the quotient of the unit cube I i , and define
( f g)(s1, . . . , sn) to be f (2s1, s2, . . . , sn) for s1 ∈ [0, 1/2] and g(2s1 − 1, s2, . . . , sn)
for s1 ∈ [1/2, 1]. This boils down to a group operation on homotopy classes of maps
that is commutative whenever i > 1. This group is denoted by πi (X, x ). If the choice
of base point x is not important, we will use the simpler notation πi (X ).
Throughout this paper, we will use the facts that πi (Si ) Z, π3(S2) Z, and
that πi (S j ) is nontrivial for many i > j (although not in all cases). The suspension
operator induces a natural map Σ : πi (X ) → πi+1(Σ X ). An important ingredient
of some proofs in this paper is the Freudenthal suspension theorem which states that
the suspension map πi (Sn) → πi+1(Sn+1) is an isomorphism for i < 2n − 1 and a
surjection for i = 2n − 1.
(Co)Homology. We assume that the reader is familiar with simplicial (co)homology.
While we assume the input spaces K , X, B to be triangulable, a continuous
perturbation h of f can be “wild” and have non-triangulable zero set. In its full generality,
(co)homology is a (contravariant) functor from the category of topological pairs to the
category of Abelian groups which satisfies Eilenberg–Steenrod axioms. These axioms
are known as homotopy invariance, exactness, excision, and the dimension axiom,
see [
12
] for exact formulations in the original source. The most common (co)homology
theory is the singular (co)homology. We require a (co)homology theory satisfying
additional axioms specified in the next section of which the Cˇ ech (co)homology is
a good example. Cˇ ech homology can violate the exactness axiom in case of “wild”
spaces and is therefore, strictly speaking, not a homology theory. However, it
constitutes a good basis for the definition of well groups.
3 Computing Lower Bounds on Well Groups
Homology Theory Behind the Well Groups. For computing the approximation φ f
Hk (X, A ∪ B) of well group Uk−n( f ) we only have to work with simplicial complexes
and simplicial maps for which all homology theories satisfying the Eilenberg–Steenrod
axioms are naturally equivalent. Hence, regardless of the homology theory H∗ used,
we can do the computations in simplicial homology. Therefore the standard algorithms
of computational topology [
10
] and the formula for the cap product of a simplicial
cycle and cocycle [28, Sect. 2.2] will do the job.
The need for a carefully chosen homology theory stems from the courageous claim
that the zero set Z of arbitrary continuous perturbation supports φ f β for any
β ∈ H∗(X, A ∪ B), i.e. some element of H∗(Z , B) is mapped by the
inclusioninduced map to φ f β. Without more restrictions on the perturbations, the zero sets
can be “wild” non-triangulable topological spaces that can fool singular homology
and render this claim false and—to the best of our knowledge—make well groups
trivial. See an example after the proof of Theorem 1.
For the purpose of the work with the general zero sets, we will require that our
homology theory satisfies the Eilenberg–Steenrod axioms with a possible exception
of the exactness axiom, and these additional properties:
1. Weak Continuity Property: for an inverse sequence of compact pairs
(X0, B0) ⊃ (X1, B1) ⊃ · · ·
the homomorphism H∗ l←im−((XXii ,, BBii )) →→ l←(iXm−j H,B∗(j X)iis, Bsui r)jeincdtiuvcee.d by the family of
inclusion lim(Xi , Bi ) =
←−
2. Strong Excision: Let f : (X, X ) → (Y, Y ) be a map of compact pairs that maps
X \ X homeomorphically onto Y \ Y . Then f∗ : H∗(X, X ) → H∗(Y, Y ) is an
isomorphism.
An intuitive explanation of how are the properties will be used in the proof of Theorem 1
is the following: we work with a sequence of shrinking neighborhoods X0, X1, . . .
of the zero sets of our a given perturbation g of f . In these neighborhood we can
identify the “look-for homology classes” with the help of the excision (but in the
case g − f = r we have to use the strong excision above). Finally, by using the
weak continuity property we can identify the looked-for homology class in the infinite
intersection i Xi —the zero set g−1(0)—as well.
Cˇech homology theory satisfies these properties as well as the Eilenberg–Steenrod
axioms with the exception of the exactness axiom, and coincides with simplicial
homology for triangulable spaces [31, Chap. 6].
In addition, we need a cohomology theory H ∗ that satisfies the Eilenberg–Steenrod
axioms and is paired with H∗ via a cap product H n(X, A) ⊗ Hk (X, A ∪ B) −→
Hk−n(X, B) that is natural19 and coincides with the simplicial cap product when
applied to simplicial complexes. We have not found any reference for the definition of
cap product in Cˇ ech (co)homology, so we present our own construction in Appendix 4.
However, if (X, A) is a triangulable pair, then we may as well use simplicial cap product
and identify φ f H∗(X, A ∪ B) with the corresponding subgroup of our homology
theory. After slight changes in the proof of Theorem 1, all cap products could be only
19 Naturality of the cap product means that if f : (X, A ∪ B, A) → (X; A ∪ B , A ) is continuous, then
f∗( f ∗(α˜ ) β) = α˜ f∗(β) for any β ∈ H∗(X, A ∪ B) and α˜ ∈ H ∗(X , A ).
X 2
X 1
X 0 = X
applied to triangulable spaces. Thus Theorem 1 would still hold under the assumption
that the pair (X, A) can be triangulated, that is, the expression φ f Hk (X, A ∪ B)
makes sense there. At least for computability results, this is no severe restriction.
With this in mind, we might as well use the Steenrod homology theory of compact
metrizable spaces [24] with cap product defined simplicially on triangulable spaces.
The advantage of Steenrod homology is that it satisfies the exactness axiom. We also
believe that it is possible to pair it with a suitable cohomology theory by a cap product
but we do not know how.
Proof of Theorem 1 We will not explicitly deal with the definition of Cˇ ech (co)
homology and will only use the facts that it satisfies the weak continuity property,
strong excision, axiom of homotopy invariance, and naturality of the cap product.
We need to show that for any map g with g − f ≤ r , the image of the
inclusioninduced map
H∗(g−1(0), B) → H∗(X, B)
contains the cap product of the primary obstruction φ f := f ∗(ξ ) with all relative
homology classes of (X, A ∪ B). Let us first restrict to the less technical case of g
being a strict r -perturbation, that is, g − f < r .
f −1[0, r ] and A0 := A = | f |−1(r ). Next we choose
Let us denote X0 := X = | |
a decreasing positive sequence ε1 > ε2 > . . . with limi→∞ εi = 0 and with ε1 <
r − f − g . Then we for each i > 0 we define
– Xi := |g|−1[0, εi ],
– Ai := Xi \ Xi+1,
– Ai−1 := Xi−1 \ Xi+1, and,
– Bi−1 := B ∩ Xi−1.
See Fig. 2 for an illustration of the spaces defined above. Note that i Xi = g−1(0),
and consequently, i Bi = g−1(0)∩ B. For any given β ∈ Hk (X, A ∪ B), our strategy
is to find homology classes αi ∈ Hk−n(Xi , Bi ), with α0 = φ f β, that fit into the
sequence of maps Hk−n(X0, B0) ← Hk−n(X1, B1) ← . . . induced by inclusions.
This gives an element in lim Hk−n(Xi , Bi ), and consequently, by the weak continuity
←−
property (requirement 1 above), we get the desired element α ∈ Hk−n(g−1(0), B).
The elements αi will be constructed as cap products. To that end, we need to obtain
“analogs” of β and for that we will need a more complicated sequence of maps. It is
the zig–zag sequence
that restricts to the zig–zags
and
The pair (Xi+1, Ai+1 ∪ Bi+1) is obtained from (Xi , Ai ∪ Bi ) by excision of
|g|−1(εi+1, εi ], that is, Xi+1 = Xi \ |g|−1(εi+1, εi ] and Ai+1 ∪ Bi+1 = ( Ai ∪ Bi ) \
|g|−1(εi+1, εi ]. Hence by excision,20 each inclusion of the pairs (Xi , Ai ∪ Bi ) →
(Xi+1, Ai+1 ∪ Bi+1) induces isomorphism on relative homology groups. Therefore
the zig–zag sequences (2) and (4) induce a sequence
Hk (X0, A0 ∪ B0) → Hk (X0, A0 ∪ B0) =∼ Hk (X1, A1 ∪ B1) → Hk (X1, A1 ∪ B1) =∼ · · ·
∈ ∈ ∈
∈
β0 := β
β0
β1
β1
· · ·
that can be made pointed by choosing the distinguished homology classes βi ∈
Hk (Xi , Ai ∪ Bi ) and βi ∈ Hk (Xi , Ai ∪ Bi ) that are the images of β0 := β ∈
Hk (X, A ∪ B) in this sequence.
Similarly, we want to construct a pointed zig–zag sequence in cohomology induced
by (2) and (3). The distinguished elements φi ∈ H n(Xi , Ai ) and φi ∈ H n(Xi , Ai ) are
defined as the pullbacks of the fundamental cohomology class ξ ∈ H n(Rn, Rn \ {0})
by the restrictions of g. Because of the functoriality of cohomology, φi and φi fit into
the sequence induced by (2) and (3):
H n(X0, A0) ← H n(X0, A0) → H n(X1, A1) ← H n(X1, A1) → · · ·
∈
∈
∈
∈
φ0
φ0
φ1
φ1
· · ·
Since g is an r -perturbation of f and thus g|(X,A) is homotopic to f |(X,A) via the
straight line homotopy, we have that φ0 = φ f ∈ H n(X, A).
20 Because of our careful choice of the spaces Ai and Ai we do not need the strong excision here. However,
we do not know how to avoid it in the case when g − f = r.
(2)
(3)
(4)
From the naturality of the cap product we get that the elements φi
fit into the sequence
βi and φi
βi
where each component ιi is induced by the inclusion i Xi → Xi . Let α ∈
Hk−n(g−1(0), B) be arbitrary preimage of α˜ under the surjection (5). By construction,
α is mapped to α0 = φ f β by the map ι0.
It remains to prove the theorem in the case when g − f = r . The proof goes
along the same lines with only the following differences:
– For arbitrary decreasing sequence 1 = ε0 > ε1 > ε2 > · · · with lim εi = 0 we
define hi := εi f + (1 − εi )g for i ≥ 0. We will furthermore need that 2εi+1 > εi
for every i ≥ 0. Let
⊆
⊆
Xi := |hi |−1[0, εi r ],
Ai := |hi |−1(εi r ).
Ai := {x ∈ X : |hi (x )| ≤ εi r and |hi+1(x )| ≥ εi+1r } and
We have Ai ⊆ Ai because by definition hi − hi+1 ≤ (εi − εi+1)r and thus
|hi (x )| = εi r implies |hi+1(x )| ≥ εi+1r . Similarly Ai+1 ⊆ Ai and Xi+1 ⊆ Xi .
Therefore as before, the zig–zag sequence (2) restricts to (3) and (4).
– The homology classes βi and βi are defined as above. We only need to use the
strong excision for the inclusion (Xi , Ai ∪ Bi ) ← (Xi+1, Ai+1 ∪ Bi+1).
– We define the cohomology classes φi := hi∗(ξ ) and φi := h∗
i+1(ξ ). We only need
to check that hi is homotopic to hi+1 as a map of pairs (Xi , Ai ) → (Rn, Rn \ {0}).
Indeed, they are homotopic via the straight-line homotopy since |hi+1(x )| ≥ εi+1r
implies |hi (x )| ≥ εi+1r −(εi −εi+1)r = (2εi+1−εi )r > 0. We used the inequality
2εi+1 > εi which was our requirement on the sequence (εi )i>0. We also have
φ0 = φ f as h0 = f and (X0, A0) = (X, A).
– We continue by defining cap products αi , their limit α˜ and its preimage α under the
surjection Hk−n( i Xi , B) → l←im−i Hk−n(Xi , B). To finish the proof we claim
that i Xi = g−1(0). Indeed, g(x ) = 0 implies hi (x ) ≤ hi − g = εi r for each
i and g(x ) > 0 implies hi (x ) > 0 for i such that 2εi r < |g(x )|.
Importance of the Choice of Homology Theory. The surjectivity of (5) and the strong
excision is not only a crucial step for Theorem 1 but implicitly also for the results
stated in [4, p. 16]. If we defined well groups by means of singular homology, then
even in a basic example f (x , y) = x 2 + y2 − 2 and r = 1, the first well group
U1( f, r ) would be trivial. The zero set of any 1-perturbation g is contained in the
annulus X := {(x , y) : 1 ≤ x 2 + y2 ≤ 3} and the two components of ∂ X are
not in the same connected components of {x ∈ X : g(x ) = 0}. However, we could
construct a “wild” 1-perturbation g of f such that g−1(0) is a Warsaw circle [23] which
is, roughly speaking, a circle with infinite length, trivial first singular homology, but
nontrivial Cˇech homology. Thus Cˇ ech homology serves as a better theoretical basis for
the well groups. Another solution to avoid problems with wild zero sets would be to
restrict ourselves to “nice” perturbations, for example piecewise linear or smooth and
transverse to 0. Such approach would lead, to the best of our knowledge, to identical
results.
Proof of Theorem 2 Assume that the dimension of K is fixed. Under the
assumption on computer representation of K and f , the pair (X, A) is homeomorphic to a
polynomial-time computable simplicial pair (X , A ) such that X is a subcomplex
of a subdivision K of K [14, Lem. 3.4]. Therefore, the induced triangulation B of
B ∩ X is a subcomplex of X . Furthermore, a simplicial approximation f : A → S
of f |A : A → Sn−1 can be computed in polynomial time. The computation is implicit
in the proof of Theorem 1.2 in [
14
] where the sphere Sn−1 is approximated by the
boundary S of the n-dimensional cross polytope B . The simplicial approximation
(X , A ) → (B , S ) of f |X can be constructed consequently by sending each vertex
of X \ A to an arbitrary point in the interior of the cross polytope, say 0 ∈ Rn. The
pullback of a cohomology class can be computed by the definition of the induced
map in simplicial cohomology. Namely, if a generator ξ of H n(B , S ) is represented
by a cocycle that assigns 1 to a distinguished n-simplex σ n ∈ B and 0 to any other
simplex, then f ∗(ξ ) is represented by a cocycle that assigns to a simplex τ n ∈ X the
number ±1 iff f (τ n) = ±σ n and 0 otherwise.21 Therefore φ f and H∗(X, B) can
be computed and the explicit formula for the cap product in [28, Sect. 2.1] yields the
computation of φ f H∗(X, B).
Modules and Diagrams Associated with φ H∗(X, A ∪ B). If the parameter r varies,
well groups U ( f, r ) naturally fit into a zig–zag sequence of homomorphisms called
21 We assume that, for simplicial (co)homology computations, an orientation of all simplices had been
chosen. The notation f (τ n) = ±σ n means that f maps vertices of τ n onto vertices of σ n in an orientation
preserving/reversing way.
well module that can be converted into a well diagram, a multiset of real numbers
indicating the death of homology classes of Xr supported by all Z ∈ Zr ( f ) as r
increases [
11
]. In this section, we show that the subgroups V := φ H∗(X, A ∪ B)
are not only subgroups of well groups for fixed r but they naturally form a persistence
module which is a sub-module of the well module in some sense. It can further be
converted to a sub-diagram of the well diagram. Unlike for well groups, the persistence
module {Vr → Vs }0<s<r is computable. It contains incomplete but often nontrivial
information about homology classes of zero sets and their robustness.
Let r1 > r2 > 0 and let X1, X2, A1, A2 be | f |−1[0, r1], | f |−1[0, r2], | f |−1{r1},
| f |−1{r2} respectively, φ1, φ2 be the respective obstructions. Further, let A1 :=
| f |−1[r2, r1] and φ1 = f ∗(ξ ) ∈ H n(X1, A1) be the pullback of the
fundamental class ξ ∈ H n(Rn, Rn \ {0}). The inclusions (X1, A1) ⊆ (X1, A1) ⊇ (X2, A2)
induce cohomology maps that take φ1 to φ1 resp. φ2. Let us denote, for
simplicity, by V1 the group φ1 H∗(X1, A1 ∪ B), V2 := φ2 H∗(X2, A2 ∪ B) and
V1 := φ1 H∗(X1, A1 ∪ B). Further, let U1 resp. U2 be the well groups U ( f, r1)
resp. U ( f, r2).
V2
∩
U2
∩
ι12
a
U2/(U2 ∩ ker i21) b
V1
∩
U1.
∩
H∗(X2, B) i21
H∗(X1, B)
⊃ i21(U2)
In this section, we analyze the relation between V1 and V2. First let i1 be a map
from V1 to V1 that maps φ1 β1 to φ1 i∗(β1). By the naturality of cap product,
φ1 β1 = φ1 i∗(β1), so i1 is an inclusion. By excision, there is an
inclusioninduced isomorphisms i1 : H∗(X2, A2 ∪ B) →∼ H∗(X1, A1 ∪ B) and its inverse
induces an isomorphism i2 : V1 →∼ V2 by mapping φ1 β1 to φ2 (i1)−1(β1). The
composition i2 ◦ i1 =: ι12 is a homomorphism from V1 to V2. Being the composition
of an inclusion and an isomorphism, ι12 is an injection and one easily verifies that
the inclusion-induced map i21 : H∗(X2, B)→H∗(X1, B) satisfies i21 ◦ ι12 = id|V1 .
It follows that {V (ri ), ιi,i+1}ri >ri+1 is a persistence module consisting of shrinking
abelian groups and injections Vi → Vi+1 for ri > ri+1. The relation between ι and
well diagrams described in [
11
] is reflected by the commutative diagram above.
The rank of U (r ) resp. V (r ) can only decrease with increasing r . In [
11
], authors
encode the properties of well groups to a well diagram that consists of pairs {(r j , μ j )}
where r j is a number in which the rank of U decreases by μ j ∈ N. Using computable
information about {V (r )}, we may define a diagram consisting of pairs (r j , μ j ) where
the rank of V (r ) decreases in r j by μ j . This is a subdiagram of the well diagram in
the following sense: each rk is then contained in {r j } j and μk ≤ μk .
The Idea Behind the Proof of Theorem 3. In the special case when X is a smooth
m-manifold with A = ∂ X , the zero set of any smooth r -perturbation g transverse to 0
is an (m − n)-dimensional smooth submanifold of X . It is not so difficult to show that
its fundamental class [g−1(0)] is mapped by the inclusion-induced map to φ f [X ],
where [X ] ∈ Hm (X, ∂ X ) is the fundamental class of X . If g−1(0) is connected, then
Hm−n(g−1(0)) is generated by its fundamental class and we immediately obtain the
reverse inclusion φ f Hm (X, A) ⊇ Um−n( f, r ). The nontrivial part in the proof of
Theorem 3 is to show that in the indicated dimension range, we can find a perturbation
g so that g−1(0) is connected. The full proof is in Appendix 2.
4 Incompleteness of Well Groups
In this section, we study the case when the primary obstruction φ f is trivial and thus
the map f |A can be extended to a map f (n) : X (n) → Sn−1 on the n-skeleton X (n)
of X . Observation 1 (proved in Appendix 3) implies that the only possibly nontrivial
well groups are U j ( f, r ) for j ≤ m − n − 1.
The following lemma summarizes the necessary tools for the constructions of this
section. They directly follow from [13, Lem. 3.1] and from [14, Lem. 3.3].
Lemma 1 Let f : K → Rn be a map on a compact Hausdorff space, r > 0, and let
us denote the pair of spaces | f |−1[0, r ] and | f |−1{r } by X and A, respectively. Then
1. for each extension e : X → Rn of f |A we can find a strict r -perturbation g of f
with g−1(0) = e−1(0);
2. for each r -perturbation g of f without a root there is an extension e : X → Rn \{0}
of f |A (without a root).
In the following we want to show that well groups can fail to distinguish between
maps with intrinsically different families of zero sets. Namely, in the following
examples we present maps f and f with U0( f, r ) = U0( f , r ) = Z for each r ≤ 1
and Ui ( f, r ) = Ui ( f, r ) = 0 for each r ≤ 1 and i > 0. However, Zr ( f ) will be
significantly different from Zr ( f ).
Proof of Theorem 4 We have that B = ∅ and K = S j × Bi , where Bi is represented
by the unit ball in Ri and j = m − i . Let the maps f, f : K → Rn be defined by
f (x , y) := |y| ϕ(x , y/|y|) and
f (x , y) := |y|ϕ (x , y/|y|),
where ϕ, ϕ : S j × Si−1 → Sn−1 ⊆ Rn are defined by
– ϕ(x , y) := μ(y) where μ : Si−1 → Sn−1 is an arbitrary nontrivial map.
– ϕ is defined as the composition S j × Si−1 → Sm−1 →ν Sn−1 where the first
map is the quotient map S j × Si−1 → S j ∧ Si−1 ∼= Sm−1 and ν is an arbitrary
nontrivial map. In other words, we require that the composition ϕ Φ—where Φ
denotes the characteristic map of the (m − 1)-cell of S j × Si−1—is equal to the
composition νq, where q is the quotient map Bm−1 → Bm−1/(∂ Bm−1) ∼= Sm−1.
Well Groups Computation. Next we prove that the well groups of U∗( f, r ) and
U∗( f , r ) are the same for r ∈ (0, 1], namely, nonzero only in dimension 0, where
they are isomorphic to Z. We obviously have X = S j × {y ∈ Ri : |y| ≤ r } S j × Bi
and A = ∂ X for both maps. The restriction f |A and f |A are equal to ϕ and ϕ (after
normalization). We first prove that U0( f, 1) ∼= U0( f , 1) ∼= Z. This fact follows from
H0(X ) ∼= Z, from non-extendability of ϕ and ϕ and from Lemma 1 part 2 (or [14,
Lem. 3.3]).
Lemma 2 The map ϕ cannot be extended to a map X → Sn−1.
We postpone the proof to Appendix 1. Since the map μ : Si−1 → Sn−1 cannot be
extended to Bi ⊃ Si−1, also ϕ cannot be extended to X .
Since then only the j th homology group of X is nontrivial, the remaining task is
to show that U j ( f, 1) ∼= U j ( f , 1) ∼= 0. We do so by presenting two r -perturbations
g and g of f and f , respectively:
– g(x , y) := f (x , y) − r x = |y|μ(y/|y|) − r x where we consider S j ⊆ R j+1 as a
subset of Rn naturally embedded in the first j + 1 coordinates (here we need that
j = m − i < n).
– We first construct an extension e : X → Rn of ϕ = f |A and then the r
perturbation g is obtained by Lemma 1 part 1. The extension e is defined as
constant on the single i -cell of X , that is, e (x0, y) is put equal to the basepoint of
Sn−1 ⊆ Rn. On the remaining m-cell Bm ∼= {z ∈ Rm : |z| ≤ 1} of X we define
e (z) := |z|e (z/|z|), where each point z is identified with a point of X via the
characteristic map Ψ1 : Bm → X of the m-cell Bm .22
By definition the only root of g is the single point Ψ1(0) of the interior of X . Therefore
U j ( f, 1) ∼= 0. Note that the role of Ψ1(0) could be played by an arbitrary point in the
interior of X .23
The zero set g−1(0) = {(x , y) : |y| = r and μ(y/|y|) = x } is by definition
homeomorphic to the pullback (i.e., a limit) of the diagram
S j
ι
Si−1
μ
Sn−1
(6)
where ι is the equatorial embedding, i.e., sends each element x to (x , 0, 0, . . .). In
plain words, the zero set is the μ-preimage of the equatorial j -subsphere of Sn−1. We
will prove that under our assumptions on dimensions, this is the (m − n)-sphere Sm−n.
Then from m − n > m − i = j it will follow that H j (g−1(0)) ∼= 0 which proves
Theorem 4.
22 Thus the formal definition is e (Ψ1(z)) := |z|e Ψ1(z/|z|) .
23 With more effort we could show that for any point z of X there is an r-perturbation of f with z being
its only zero point.
The topology of the pullback is particularly easy to see in the case when j = n − 1
and ι is the identity. There it is simply the domain of μ, that is, Si−1 where i − 1 =
m − j − 1 = m − n.
In the general case, the only additional tool we use to identify the pullback is the
Freudenthal suspension theorem. The pullback is homeomorphic to the μ-preimage
of the equatorial subsphere Sm−i ⊆ Sn−1. By Freudenthal suspension theorem μ is
homotopic to an iterated suspension Σ a η for some η : Si−1−a → Sn−1−a assuming
i − 1 − a ≤ 2(n − 1 − a) − 1. We want to choose a so that n − 1 − a = m − i and
thus images Im(η) = Sn−1−a and Im(ι) = S j ⊆ Sn−1 coincide (since j = m − i by
definition). The last inequality with the choice a = n − 1 − m + i is equivalent to
the bound i ≤ (m + n − 1)/2 from the hypotheses of the theorem. In our example,
we may have chosen f in such a way that μ = Σ a η. But even for the choices of
μ only homotopic to Σ a η we could have changed f on a neighborhood of ∂ K by
a suitable homotopy. To finish the proof we use the fact that, by the definition of
suspension, the μ-preimage of Sm−i ⊆ Sn−1 is identical to the η-preimage of Sm−i ,
that is Si−1− j = Sm−n .
Difference Between Zr ( f ) and Zr ( f ). Because the map μ is homotopically nontrivial,
the zero set of each extension e : X → Rn of f |A intersects each “section” {x } × Bi
of X . By Lemma 1 part 2 (or [14, Lem. 3.3]) applied to each restriction f |{x}×Bi , the
same holds for r -perturbations g of f as well. In other words, the formula “for each
x ∈ S j there is y ∈ Bi such that f (x , y) = 0” is satisfied robustly, that is
∀Z ∈ Zr ( f ) : ∀x ∈ S j : ∃y ∈ Bi : (x , y) ∈ Z
is satisfied. The above formula is obviously not true for f as can be seen on the
r -perturbations g . In particular, for every r ∈ (0, 1] the family Zr ( f ) contains a
singleton. This completes the proof of Theorem 5.
Robust Optimization. As an example of another relevant property of Zr ( f ) not
captured by the well groups, we mention the following. For any given u : K → R,
we may want to know what is the r -robust maximum of u over the zero set of f ,
i.e., inf Z∈Zr ( f ) maxz∈Z u(z). Let, for instance, u(x , y) = u(x ) depend on the first
coordinate only. Then the r -robust maximum for f is equal to maxx∈S j u(x ) as
follows from the discussion in the previous paragraph. On the other hand, the r -robust
maximum for f is equal to minx u(x ) and is attained in g when we set the value
Ψ1(0) := (arg minx∈S j u(x ), 0) from the proof above. This holds for r arbitrarily
small. The robust optima constitutes another and, in our opinion, practically relevant
quantity whose approximation cannot be derived from well groups.
Further Remarks on Theorem 4. We first want to indicate that in some sense the maps f
and f are no peculiar examples but rather typical choices. More precisely, we assume
that r > 0 is fixed and that X = S j × Bi and A = ∂ X . (Note that in the natural
cell structure of X there is only one i -dimensional and one (i + j )-dimensional cell
outside of A.) It can be easily proved that under these assumptions the maps f and f
can be chosen arbitrarily in such a way that
– f | A cannot be extended to X (i) (it extends to X (i−1) trivially as A = X (i−1)) and
– f | A extends to X (i) but not to X .24
The only addition needed to prove this more general version is in the computation of
Um−i ( f, r ). For that we can either use Theorem 5 when i < (m + n)/2 or enhance
the proof of Theorem 4 when i = (m + n)/2 which we omit here. Note that the
nonextendability properties of f and f require nontriviality of the homotopy groups
of spheres as in the hypothesis of Theorem 4. Then only for the requirement i > n we
know that is strict. The other two inequalities are used to find the map ι such that the
pullback (6) is connected enough. The inequality i < (m + n + 1)/2 can be relaxed
to requiring the existence of [μ] ∈ πi−1(Sn−1) such that [μ] = Σ a η for a sufficiently
large as stated in the proof.
Finally, we remark that the same incompleteness results could be achieved for even
more realistic domain K = B j × Bi ∼= Bm . We only need to choose f and f with
X = B j × Bri and A = B j ×(∂ Bri ) and with the same properties as above. Then for the
natural choice B = ∂ K and under the same hypotheses, both well groups will be equal.
Sketch of the Proof of Lemma 2. The ultimate claim is that ϕ cannot be extended to
the m-cell of X no matter how the extension on the i -cell was chosen. To this end,
we need two properties of the obstruction to extendability on the m-cell (which is an
element of πm−1(Sn−1)):
1. First, that the obstruction is independent of the choice of the extension on the i -cell.
This essentially follows from the bilinearity of the Whitehead product πi (Sn−1) ⊗
πm−i (Sn−1) → πm−1(Sn−1), namely, that the Whitehead product of a trivial
element with an arbitrary element is again a trivial element.
2. Second, that the obstruction depends linearly on the choice of the element [ν] ∈
πm−1(Sn−1) in the definition of the map ϕ . This amounts to the basic obstruction
theory and the cell structure of the solid torus.
The full proof is presented in Appendix 1.
Acknowledgments Open access funding provided by Institute of Science and Technology (IST Austria).
We are grateful to Ryan Budnay, Martin Cˇ adek, Marek Filakovský, Tom Goodwillie, Amit Patel, Martin
Tancer, Lukáš Vokˇrínek and Uli Wagner for useful discussions. This research was supported by Austrian
Science Fund (FWF): M 1980, by the Czech Science Foundation (GACR) Grant Number 15-14484S with
institutional support RVO:67985807, and by the People Programme (Marie Curie Actions) of the European
Union’s Seventh Framework Programme (FP7/2007–2013) under REA grant agreement n [291734].
Appendix 1: The Nonextendability Proof (Proof of Lemma 2)
Proof of Lemma 2 As our ultimate claim will be that ϕ cannot be extended to the
mcell of X = S j × Bi , we will need to analyse the gluing map of that cell. In particular,
we need to establish its relation to the gluing map of the m-cell of T = S j × Si . In
the first row of the following commutative diagram we have the factorization of the
24 The only remaining category consists of those f where f |A extends to whole of X , i.e., U∗( f, r ) ∼= 0,
or equivalently, ∅ ∈ Zr ( f ).
characteristic map Ψ of the m-cell of T into the characteristic map Ψ1 of the m-cell
of X and another quotient map Ψ2.
B j × Bi
∂(B j × Bi )
Ψ1
Ψ2
ψ2
X = (B j /∂ B j ) × Bi
T = (B j /∂ B j ) × (Bi /∂ Bi )
ψ1
X (m−1)
T (m−1) = S j ∨ Si
Note that above we identify spheres with the quotients of balls by their boundary. The
restriction of each characteristic map to the boundary ∂(B j × Bi ) gives the respective
gluing maps as is shown in the second row. We have that X (m−1) = S j × ∂ Bi ∪
{∗} × Bi and25 indeed the quotient map Ψ2 (which identifies each {x } × ∂ Bi into a
point (x , ∗)) maps X (m−1) to S j ∨ Si . The crucial consequence is that the gluing map
ψ of the m-cell in T is the composition of the gluing map ψ1 of the m-cell in X and
the restriction ψ2 of the quotient map Ψ2 as above.
Another tool that we need is the Whitehead product [18, Chap. X, 7.2] for which
we need the explicit construction. Again, ψ : Sm−1 → S j ∨ Si denotes the gluing
map of the m-cell in T = S j × Si . Then the homotopy class of the composition
Sm−1 →ψ S j ∨ Si ω−1∨→ω2 Sn−1 defines the Whitehead product of arbitrary elements
[ω1] ∈ π j (Sn−1) and [ω2] ∈ πi (Sn−1). We will use the bilinearity of the Whitehead
product, especially, that the product of the trivial element [const] and arbitrary [ω] is
trivial.
Let us assume that
h : ei ∪ A → Sn−1
X(i)∪A
is an extension of ϕ on the unique i -cell ei of X . The map h can be extended to
X if and only if there is a nullhomotopy for the composition hψ1 : Sm−1 → Sn−1
where again ψ1 : Sm−1 → X (m−1) = X (i) ∪ A is the gluing map of the m-cell of
X . Roughly speaking, the key difficulty is that ϕ can be extended on ei in various
essentially different ways (whenever πi (Sn−1) is nontrivial). The key observation is
that this choice does not influence the homotopy class of hψ and that it is always equal
to the above chosen nontrivial element [ν] ∈ πm−1(Sn−1) up to a sign.
Towards that end, let us form an auxiliary map h : ei ∪ A → Sn−1 that is constant
on ∂ X and equal to h on the unique i -cell of X . We want to show first, that h ψ1 is
homotopically trivial, and second, that [hψ1] = [h ψ1] ± ν.
1. Since h is constant on {x } × (∂ Bi ) for each x ∈ S j , it factors through the
corresponding quotient S j ∨ Si of X (i) as follows (in the following the factorization is
25 Having arbitrary space in mind, the sign ∗ will denote its basepoint. In our case, it will always be the
single 0-cell of the space.
preceded by ψ1):
Sm−1
ψ1
X (m−1) = X (i)
ψ2
S j ∨ Si
const∨ω
Sn−1
Here const denotes the constant map and ω is the map determined by h (or
equivalently by h ) on {∗} × Bi . Since by the above considerations ψ2ψ1 = ψ , the
composition h ψ1 is equal to (const∨ω)ψ —the representative of the Whitehead
product of [const] and [ω]. By the bilinearity of the Whitehead product, h ψ1 is
trivial.
2. The second claim—[hψ1] = [h ψ1] ± ν—follows from basic obstruction
theory. This claim follows from the fact, that for any pair of maps h and h
that agree on X (m−2), the difference of their obstruction cocycles zhm − zhm ∈
Z m X, πm−1(Sn−1) equals the coboundary of the difference cochain dh,h ∈
C m−1 X, πm−1(Sn−1) . To get the conclusion, we need two ingredients: that the
coboundary map is an isomorphism and that the difference cochain is nontrivial,
namely, that it assigns ν to the (m − 1)-cell of X . The first ingredient was already
shown in the first paragraph of this proof. Since the cellular chain structure of
X is rather simple—having one generator in both dimensions m and m − 1—we
rephrase everything in an elementary language below.
The first ingredient is that the degree d of the composition
Sm−1 →ψ1 X (m−1) → X (m−1)/(S j ∨ Bi ) ∼= Sm−1
is equal to ±1. The second ingredient is that, once we denote the characteristic map
of the (m − 1)-cell of X by Φ, the difference map of h Φ and hΦ equals ±ν. The
difference map of any given maps f : Bm−1 → Sn−1 and g : Bm−1 → Sn−1 with
f |∂ Bm−1 = g|∂ Bm−1 is defined as δ f,g := f ∪∂ Bm−1 g : Sm−1 → Sn−1. In words,
f defines δ f,g on the northern hemisphere and g defines δ f,g on the southern
hemisphere. Because there are factorizations hΦ = ϕ Φ = νq and26 h Φ =
const = ν const through Sm−1, we have that δhΦ,h Φ = νδq,const. Obviously, the
map δq,const has degree ±1 and thus the second ingredient holds.
By the definition of the addition in πm−1(Sn−1), we have that [hψ1] = [h ψ ] ±
d[δhΦ,h Φ ].
This concludes the proof of Lemma 2 and thus the proof of Theorem 4 as well.
Appendix 2: Proof of Theorem 3
Overview. The proof of Theorem 3 will be divided into several steps. Theorem 1
implies one inclusion and for the other, it is sufficient to find a smooth r -perturbation
g of f such that 0 is a regular value of g and the homology image of g−1(0) in X
generates the cap product φ f Hm (X, ∂ X ). First we show that the general case
2q6uoWtieenhtamvaepthBamth−Φ1 →= hB|mA−Φ1/a(n∂dBhm|−A1=), wϕasanredqwuiererdemininthdetdheafitnthiteioenquoaflϕity ϕ Φ = νq, where q is the
easily reduces to the case where X is connected. In the next step, we describe the
(m − n)th homology of zero sets of perturbations transverse to zero: we prove that
if 0 is a regular value of a strict r -perturbation g of f , then the Poincaré dual of φ f
equals the image of the fundamental class of the submanifold g−1(0) (Lemma 3). If
g−1(0) is connected, then the fundamental class of g−1(0) generate its top homology
Hm−n(g−1(0)). In this way, we reduce the proof of Theorem 3 to the statement that
if n + 1 ≤ dim X ≤ 2n − 3 and X is connected, then there exists some smooth strict
r -perturbation g of f such that 0 is a regular value of g and g−1(0) is connected. To
prove this, we introduce the notion of framed submanifolds and show that if a framed
(m −n+1)-submanifold W has framed boundary consisting of S f −1(0), then S is the
zero set of some smooth strict r -perturbation g having 0 as regular value (Lemma 4).
Finally, we show that there exists a framed submanifold W and a connected framed
(m − n)-submanifold S of X s.t. ∂ W = f −1(0) S (Lemma 5).
Reduction to the Case of Connected X . Assume that Theorem 1 holds for X
connected. The compact space X can only contain finitely many connected components,
say X1, . . . , Xk . Then Hm (X, ∂ X ) i Hm (Xi , ∂ Xi ) and φ f Hm (X, ∂ X )
j ι∗j φ f Hm (X j , ∂ X j ) where ι j : X j → X is the inclusion. If we assume that
Theorem 3 holds for X connected, we may use it for f |Xi : Xi → Rn and get that
g|Xi : g|Xi − f |Xi ≤r
(7)
is contained in ιi∗φ f Hm (Xi , ∂ Xi ) for all i . However, each r -perturbation g of f
induces r -perturbations g|Xi of f |Xi and each set of r -perturbation gi of f |Xi induces
an r -perturbation g of f ; therefore
g: g− f ≤r
i
Im Hm−n(g−1(0)) −→∗ Hm−n(X )
is isomorphic to the direct sum of (7) over i and is therefore a subset of
Hm (X j , ∂ X j ) φ f Hm (X, ∂ X ).
In the rest of the proof, we will assume that X is connected.
j ι∗j φ f
Poincaré Dual of the Fundamental Class. Now we will show that the Poincaré dual
of the first obstruction equals the image of the fundamental class of the zero set of a
smooth strict r -perturbation transverse to 0.
Lemma 3 Let X be a smooth oriented m-manifold with boundary, A and B be (m
−1)submanifolds of ∂ X , ∂ X = A ∪ B, ∂ A = ∂ B, f : (X, A) → (Rn, Rn \ {0}) be smooth
with 0 a regular value of f and f |∂ X , [X ] ∈ Hm (X, ∂ X ) the fundamental class of X ,
φ f = f ∗(ξ ) ∈ H n(X, A) the first obstruction and φ f [X ] its Poincaré dual.
Then the smooth submanifold f −1(0) of X can be endowed with an orientation
such that its fundamental class [ f −1(0)] ∈ Hm−n( f −1(0), B) satisfies
i∗([ f −1(0)]) = φ f
where i : f −1(0) → X is the inclusion.
It follows immediately that φ f [X ] equals to the image of the fundamental class
of any smooth g such that φg = φ f : this happens, in particular, if A = | f |−1(r ) and
g is a smooth strict r -perturbation of f transverse to 0.
Proof If 0 is a regular value of f and f |∂ X , then f −1(0) is a smooth (m −
n)dimensional submanifold of X and ∂ f −1(0) = f −1(0) ∩ B is a smooth submanifold
of B; it also follows that the inclusion i : f −1(0) → X induces a homology map that
maps the fundamental class [ f −1(0)] ∈ Hm−n( f −1(0), ∂ f −1(0)) to Hm−n(X, B).
Smooth manifolds can be triangulated and a triangulation of f −1(0) ∩ B can be
extended to a triangulation of all B and subsequently to f −1(0), ∂ X and X [26,
Thm. 10.6.]. In the rest, we will work with simplicial (co)homology and simplicial
cap product, which for simplicial complexes coincides with our homology theory. We
will show that the Poincaré dual φ f [X ] of the obstruction is the image of the
fundamental class [ f −1(0)] by induction on n.
Let n = 1. For the use of simplicial homology, choose an ordering of all
vertices such that the vertices in f −1[−r, 0) have lower rank then vertices in f −1[0, r ].
The obstruction φ f can be represented by a simplicial cocycle z f that assigns 1 to
each oriented 1-simplex x y with f (x ) < 0 and f (y) ≥ 0, and 0 to other oriented
1-simplices. The fundamental class [X ] ∈ Hm (X, ∂ X ) consists of all m-simplices
in X and the cap product z f ( m ∈X(m) m ) consists of all (m − 1)-simplices
m−1 = [y0, y1, . . . , ym−1] in f −1(0) such that [x , y0, . . . , ym−1] is an m-simplex
in X , f (x ) < 0 and f (y j ) = 0 for all j . The set of all such m−1 yields a
triangulation of f −1(0) with the orientation induced from the chosen orientation of X .
It follows that φ f [X ] equals the image of the fundamental class of [ f −1(0)] in
Hm−1(X, B).
Let n > 1 and f = ( f1, f2) with f1 scalar valued and f2 : X → Rn−1. Each
x ∈ f −1(0) is a regular point of f and f |∂ X , hence it is a regular point of both
f1, f1|∂ X and f2, f2|∂ X . It follows that there exists a neighborhood U of f −1(0)
s.t. 0 is a regular value of both f1|U , f1|U ∩∂ X and f2|U , f2|U ∩∂ X . Possibly
changing f1 and f2 outside U without changing f −1(0) = f1−1(0) ∩ f2−1(0), we may
assume that 0 is a regular value of both f1, f1|∂ X and f2, f2|∂ X , so that f1−1(0)
and f2−1(0) are smooth manifolds of dimensions m − 1 and m − n + 1,
respectively. Choose a compact (m − 1)-submanifold A1 of A such that 0 ∈/ f1( A1) and
so that A2 := A \ A1 satisfies 0 ∈/ f2( A2). Both A2 and B ∪ A1 = ∂ X \ A2
are smooth m − 1-dimensional submanifolds of ∂ X , A2 ∪ (B ∪ A1) = ∂ X and
∂ A2 = ∂(B ∪ A1).
The maps f, f1 and f2 can be considered as maps of pairs f : (X, A) →
(Rn, Rn \{0}), f1 : (X, A1) → (R, R\{0}) and f2 : (X, A2) : (Rn−1, Rn−1 \{0}). Let
ξ n, ξ 1, resp. ξ n−1 be fundamental classes of H j (R j , R j \{0}) where j equals n, 1, resp.
n − 1; here we assume a canonical orientation on R j . Let φ1 := f1∗(ξ 1) ∈ H 1(X, A1)
and φ2 := f2∗(ξ n−1) ∈ H n−1(X, A2) be the corresponding obstructions. The cross
product in cohomology [18, p. 214]
H 1(R, R \ {0}) × H n−1(Rn−1, Rn−1 \ {0}) −×→ H n(Rn, Rn \ {0})
takes (ξ 1, ξ n−1) to ξ n. Using this we obtain
( p2 f )∗(ξ n−1) = f1∗(ξ 1)
f2∗(ξ n−1)
for p1 and p2 the projections in Bn to the first, resp. the remaining components.
Comparing the left and right hand side of the last equation yields φ f = f ∗(ξ n) =
f1∗(ξ1) f2∗(ξ n−1) = φ1 φ2.
Now we use the induction hypothesis for the Rn−1-valued map f2 and the
subcomplexes A2 and B ∪ A1 of X . It says that φ2 [X ] is the image of
the fundamental class [ f2−1(0)] ∈ Hm−n+1( f2−1(0), B ∪ A1) in Hm−n+1(X, B ∪
A1).
The naturality of the cap product yields the following scheme:
Hm−n+1(X, B ∪ A1)
Hm−n(X, B)
H 1(X, A1)
i∗
i∗
H 1( f2−1(0), A1) × Hm−n+1( f2−1(0), B ∪ A1)
Hm−n( f2−1(0), B).
i∗
i∗
Hm−n( f −1(0), B)
The pullback φ˜1 := i ∗φ1 ∈ H 1( f2−1(0), A1) is the obstruction associated to the
restriction of f1 to f2−1(0). The restrictions f1| f2−1(0) as well as f1|∂ f2−1(0) have 0 as
regular value, so using again the induction hypothesis, we get that φ˜1 [ f2−1(0)] is
the inclusion-induced image of [ f −1(0)] ∈ Hm−n( f −1(0), B) in Hm−n( f2−1(0), B).
Using the commutativity of diagram (8), we get that the inclusion-induced image of
[ f −1(0)] in Hm−n (X, B) equals
φ1
[X ] = φ f
which completes the proof.
In the rest of this section, we will only need the last lemma for the case where A = ∂ X
and B = ∅.
Reduction to the Existence of a Perturbation g Such That g−1(0) is Connected. Assume
f −1[0, r ], | f |−1(r ), X is a smooth connected manifold, A = ∂ X ,
that (X, A) = | |
g is smooth, 0 is a regular value of g, g − f < r , and g−1(0) is connected. The
constraint g − f < r immediately implies that f and g are homotopic as maps
A → Rn \ {0} and φ f = φg. As X is connected, the group Hm (X, ∂ X ) is generated
by the fundamental class of X and we already know by Lemma 3 that φ f [X ] =
φg [X ] equals the image of the fundamental class of g−1(0) in Hm−n(X ). But if the
manifold g−1(0) is connected, then Hm−n(g−1(0)) is generated by the fundamental
class of g−1(0), so its image in Hm−n (X ) is generated by φ f [X ]. It follows that
Um−n( f, r ), being the intersection over all r -perturbations, cannot contain anything
else than multiples of φ f [X ], and we obtain the desired inclusion
φ f
Hm−n(X, ∂ X ) ⊇ Um−n( f, r ).
So, it remains to prove that if n +1 ≤ dim X ≤ 2n −3, X is connected and A = ∂ X ,
then there exists a smooth strict r -perturbation g of f transverse to 0 such that g−1(0)
connected. To show this, we need to introduce additional concepts from differential
topology.
Framed Submanifolds. Assume that X is a smooth m-manifold endowed with a
Riemannian metric; the results will be independent on the choice of the metric. Let S ⊆ X
be a smooth (m − n)-submanifold contained in the interior of X ; for each x ∈ S, the
tangent space Tx X decomposes as a direct sum of the tangent space Tx S and the
normal space Nx S. A framing on S is a trivialization of the normal bundle N S, in other
words, a smooth mapping T such that for each x ∈ S, T (x ) = (T1(x ), . . . , Tn(x )) is
a basis of the normal space Nx S.
If f : X → Rn has 0 a regular value, then f −1(0) is naturally a framed
(m − n)-submanifold, Ti (x ) being the unique vector in Nx f −1(0) mapped by d f
to ei ∈ Rn. We will denote these vectors by f ∗(ei ). Assume that W is a framed
(m − n + 1)-submanifold of X with framing T2(x ), . . . , Tn(x ) and that ∂ W is the
boundary of W . The existence of collars implies that some neighborhood of ∂ W in W
is diffeomorphic to ∂ W × [0, 1) with coordinates (w, t ). The framing of W induces a
framing of its boundary, given by (T1(x ), . . . , Tn(x )) where T1(x ) is the vector ∂t in
the “inwards” direction and (T2(x ), . . . , Tn(x )) the framing of W in x ∈ ∂ W .
Lemma 4 Let X be a smooth m-manifold, r > 0, f : X → Rn be smooth with 0 a
f −1(r ), and dim X ≤ 2n − 3. Let S ⊆ X be a framed
regular value of f , A = ∂ X = | |
boundary-free (m − n)-submanifold of X disjoint from A and assume that there exists
a framed (m − n + 1)-submanifold W ⊆ X disjoint from A so that ∂ W = f −1(0) S
and W induces the framing of f −1(0) S.
Then there exists a smooth g so that g − f < r , 0 is a regular value of g and
g−1(0) = S.
We will see that g can be even chosen so that the S-framing (T1(x ), . . . , Tn(x )) satisfies
T1(x ) = −g∗(e1) and Ti (x ) = g∗(ei ) for i > 1.
Proof Step 1: Reduction to the existence of h homotopic to f , h−1(0) = S.
We will construct a smooth map h s.t. h−1(0) = S and h/|h| will be homotopic to
f /| f | as maps from A → Sn−1. This is sufficient, because then we might easily change
h in a collar of A diffeomorphic to A × [
0, 1
] that is disjoint from h−1(0) to obtain a
smooth extension e : X → Rn of f |A that coincides with h outside this neighborhood.
According to [13, Lem. 3.1] some positive scalar multiple χ (x ) e(x ) =: g(x ) satisfies
g − f < r . The map χ can be chosen to be smooth: then 0 is a regular value of g
and g−1(0) = h−1(0) = S. In the rest of the proof, we will show how to construct h.
Step 2: Constructing a perturbation f˜ of f .
Let (T2, . . . , Tn) be the framing on W , inducing the framing (T1, . . . , Tn) on f −1(0)
S = ∂ W . On f −1(0), Ti coincides with f ∗(ei ). Let Bεn be a closed neighborhood of
Fig. 3 Illustration of the framed
manifold W that consists of W
and f˜−1R−e1. The boundary
0
∂W consists of S and some
submanifold of A = ∂ X
W
2
S
X
f˜− 1(R 0− e1)
A
νW
∂(νW )
0 ∈ Rn consisting of regular values of f and let L be the closed straight line segment
connecting 0 and −εe1 ∈ Rn. Then the f -preimage of L is an (m −n +1)-submanifold
of X with boundary f −1({0, −ε} × {0}), with a framing ( f ∗(e2), . . . , f ∗(en)), where
e2, . . . , en are the normal vectors to L for each x ∈ L. By making ε possibly smaller,
we can assume that f −1(L) ∩ W = f −1(0), because in a small enough neighborhood
U of f −1(0), f1 is positive on (W \ f −1(0)) ∩ U (by definition, (d f1)(T1(x )) > 0
for x ∈ f −1(0)) and negative on ( f −1(L) \ f −1(0)) ∩ U . The vector field T1(x ) for
x ∈ f −1(0) is in the tangent space of both W and f −1(L); it has the inwards direction
wrt. W and outward wrt. f −1(L).
Let V := f −1(Bεn). The restriction f |V is transverse to the closed set
R0−e1 := (−∞, 0] × {0} ⊆ Rn
by construction. Using a relative version of transversality theorem, the space of smooth
functions that coincide with f on V and are transverse to R−e1 is dense and open
0
in {g ∈ C ∞(X, Rn ) : g|V = f |V } in Whitney C 1-topology (this follows from [1,
Thm. 19.1]) so there exists an arbitrary small perturbation f˜ of f that is smooth,
transverse to R0−e1 and f˜|V = f |V . Then f˜−1(R0−e1) is a smooth (m − n +
1)submanifold of X with boundary f −1(0) = f˜−1(0).
The assumption m ≤ 2n−3 implies that dim W +dim f˜−1(R0−e1) = 2(m−n+1) <
m, so both W and f˜−1(R0−e1) have dimension less than one half of m = dim X .
Therefore, we can replace f˜ by another arbitrary small perturbation, without changing
it on V , assume that it is transverse to R0−e1 and moreover, f˜(R0−e1) intersects W only
in f −1(0). Assume that f˜ is close enough to f so that f |A is homotopic to f˜|A as
maps from A to Rn \ {0}. Without loss of generality, we may assume that f˜−1(R0−e1)
intersects A transversally (otherwise we replaced f˜ by another perturbation that differs
from f˜ in a neighborhood of A) and hence f˜−1(R0−e1) ∩ A is an (m − n) dimensional
submanifold of A.
The submanifold f˜−1(R0−e1) is endowed with a framing ( f˜∗(e2), . . . f˜∗(en))
where e2, . . . , en are vectors of the canonical basis in T(y,0)Rn for y ≤ 0. This framed
manifold intersects W in f −1(0), the tangent spaces of both manifolds coincide in
f −1(0), T1 directs inwards wrt. W and outwards wrt. f˜−1(R0−e1), and the framing on
both submanifolds coincide in f −1(0) (Fig. 3).
Step 3: Gluing W and f˜−1(R0− e1) to one smooth submanifold.
Both submanifolds W and f˜−1(R0−e1) of X intersect in their common boundary
f −1(0) and both the tangent spaces and framings coincide in f −1(0). We would like
to smoothly “glue” them to one framed manifold W ∪ f˜−1(R0−e1) but unfortunately,
such union does not need to yield a smooth submanifold in general.
We claim that there exists a smooth framed manifold W that coincides with W ∪
f˜−1(R0−e1) everywhere except on a neighborhood of f −1(0) in X that can be chosen
to be arbitrary small. Choose a continuous tangent vector fields v in W ∪ f˜−1(R0−e1)
that is smooth on W and smooth on f˜−1(R0−e1), such that v| f −1(0) is nonzero and
points inwards wrt. W and outwards wrt. f˜−1(R0−e1) (it may coincide with f ∗(e1)).
The flow of this vector fields induces collar neighborhoods C1 resp. C2 of f −1(0) in
W resp. f˜−1(R0−e1) contained in X diffeomorphic to f −1(0) × [0, ε), resp. f −1(0) ×
(−ε, 0] for some ε > 0. Let as denote the embeddings f −1(0) × [0, ε) → X and
f −1(0) × (−ε, 0] → X by w1 and w2, respectively. Let w : f −1(0) × (−ε, ε) → X
be defined by w1(x , t ) for t ≥ 0 and w2(x , t ) for t < 0: this map is a C 1-embedding
(the differentials dw1 and dw2 coincide on f −1(0)) and is C ∞ whenever t = 0.
Let ψα : Uα → Rm be a collection of X -charts and let {Vβ }β be an open covering
of f −1(0) × [−ε/2, ε/2] such that for each Vβ , w(Vβ ) is contained in some Uα(β).
Further, let Vβ ⊆ Vβ be so that Vβ ⊆ Vβ and {Vβ }β is still an open covering of
f −1(0) × [−ε/2, ε/2]. Let β1, . . . , βk be such that ∪ j Vβ j is an open neighborhood of
f −1(0) × {0}. The space C ∞(Vβ1 , Rm ) is dense in C 1(Vβ1 , Rm ) with the Whitney C
1topology [20, Thm. 2.4] so we may replace ψα(β1) ◦ w|Vβ1 : Vβ1 → Rm by an arbitrary
close map (in the Whitney topology) Vβ1 → Rm that is smooth on Vβ1 and unchanged
in a neighborhood of ∂ Vβ1 . This defines a map w1 : f −1(0) × [−ε/2, ε/2] → X that
is smooth on Vβ1 and coincides with w on f −1(0) × {±ε/2}. On w1(Vβ1 ), we can also
use the ψα(β1)-chart to define a framing that is smooth on w1(Vβ1 ) and coincides with
the original framing on a neighborhood of w1(∂ Vβ1 ). In the same way, we smoothen
the function on Vβ2 , . . . , Vβk and obtain a smooth map w : f −1(0) × (−ε, ε) → X
arbitrary close to w that coincides with w on a neighborhood of f −1(0) × {±ε}. If we
chose w close enough to w, it is an embedding by [20, Thm. 1.4]. The manifold W
can now be defined as
W := Im(w ) ∪ (W ∪ f˜−1(R0−e1)) \ Im(w ) ,
which is a smooth embedded framed submanifold that coincides with W ∪ f˜−1(R0−e1)
except on a neighborhood of f −1(0) that can be chosen to be arbitrary small. The
framing coincides with the framing on W ∪ f˜−1(R0−e1) outside Im(w ). By construction,
the boundary of W consists of S and a submanifold W ∩ A of A.
Step 4: Choice of the metric.
Let us choose a vector field v on A = ∂ X in the inwards-direction so that for x ∈
W ∩ A, v(x ) ∈ Tx W . This can be extended to a nowhere zero vector field in a
neighborhood of A and used to define a collar neighborhood of A diffeomorphic to
A × [0, ε) for some ε > 0, the diffeomorphism induced by the flow of v. We endow X
with a new smooth Riemannian metric that is a product metric on this neighborhood.
Due to this choice of the metric, the geodesics in A coincide with the geodesics in
X . In what follow, we will assume that such metric has been defined and we identify
the given framing of W and S with normal vectors wrt. this metric. In particular, W
intersects A orthogonally and the W -framing vectors in W ∩ A are all in the tangent
space of A.
Step 5: Construction of h.
The first framing vector T1 of the S-framing can be extended to a smooth tangent (wrt.
W ) vector field in a neighborhood of S in W and further to a neighborhood of S in X .
The flow of −T1 then generates an external collar C of W diffeomorphic to S × [0, ε]
for some ε > 0 such that W := C ∪ W is a smooth submanifold of X . Using charts
and partition of unity, we may easily extend the W -framing to a framing on W .
Without loss of generality, we can assume that the external collar C of W is disjoint
from A. The flow of −T1 induces a neighborhood ν(S) of S in W diffeomorphic to
S ×[−ε, ε] with S ×(−ε, 0] corresponding to a neighborhood of S in W and S ×[0, ε)
to the open external collar of S contained in W . The projection on [−ε, ε] defines a
smooth scalar valued map h1 : ν(S) → R s.t. h1−1(0) = S, h1 ≥ 0 on C and h1 ≤ 0
on ν(S) ∩ W : (h1)∗ maps T1 (which directs inwards to W ) to (−∂x ) ∈ T0R.
By construction, f˜1 is negative in W ∩ A. Let U be a closed neighborhood of A in
X such that f˜1 is still negative on U ∩ W and extend h1 to a smooth map W → R
such that (h1)|U ∩W = ( f˜1)|U ∩W and h1 < 0 on W \ S.
The geodesic flow of the W -framing of W induces a diffeomorphism ϕ of W ×
Bn−1 and some set ν(W ) ⊆ X , where Bn−1 is the closed ball in Rn−1 of small enough
diameter (due to our choice of the metric, framing vectors in W ∩ A induce geodesics
in A). This set ν(W ) is not open in X , but it contains an open neighborhood of W .
The projection on Bn−1 defines a smooth function h : ν(W ) → Rn−1 transverse to 0
such that h −1(0) = W and h induces the given framing on W . Let us extend h1 to a
smooth scalar valued map ν(W ) → R arbitrarily and finally define h : ν(W ) → Rn
by h = (h1, h ). It is easy to see that h−1(0) = S and 0 is a regular value of h.
Summarizing the construction, we have a (closed) neighborhood ν(W ) of W in X
and a smooth map h : ν(W ) → Rn such that
∩ h−1(R0−e1) = W ,
– h−1(0) = S, the original framing of S equals (−h∗(e1), h∗(e2), . . . h∗(en )),
– (h2, . . . , hn)−1(0) = W and (h2, . . . , hn) induces the original framing on W ⊆
W ,
– W
– h|U ∩W
= f˜|U ∩W .
By construction, h restricted to ∂(ν(W )) has values in Rn \ {R0− e1}: this is because
h(x ) ∈ R−e1 implies x ∈ W and W is in the interior of ν(W ). The topological space
Rn \ (R0−0e1) is homotopically trivial as it deformation retracts to a point, so h|∂ν(W )
can be extended to a continuous map X \ ν(W ) → Rn \ {R0− e1} (for this, we need
the homotopy extension property of X \ ν(W ) and its closed subset ∂ν(W )) which
in turn defines an extension h : X → Rn of the map h|ν(W ) that we have already
defined. Possibly perturbating h slightly outside of some neighborhood of W , we
may assume that it is smooth [22, Thm. 2.5]. By construction, 0 is a regular value of
h.
Step 6: The restriction h|A is homotopic to f |A.
We will show that H := h/|h| and F := f /| f | are homotopic as maps from A →
Sn−1. Let F˜ be the restriction of ( f˜/| f˜|) to the open neighborhood U of A. We
assumed that f |A is homotopic to f˜|A as maps A → Rn \ {0}, so the sphere-valued
maps F |A is homotopic to F˜ |A and it remains to show that F˜ |A is homotopic to H |A.
By construction, ( f˜|U )−1(R−e1) = (h|U )−1(R0−e1) = U ∩ W , both maps are
0
nowhere zero on U , they coincide on U ∩ W and both maps induce the same framing
on U ∩ W . It follows that F˜ −1(−e1) = (H |U )−1(−e1) = U ∩ W and both F˜ and
H induce the same framing on U ∩ W (it coincides with the original framing of
U ∩ W up to scalar multiples of framing vectors): this framing restricts on W ∩ A
to a framing of the normal space to W ∩ A in A. Therefore, for x ∈ F˜ −1(−e1) ∩ A,
F˜∗(x ) = H∗(x ) and consequently (F˜ |A)∗(x ) = (H |A)∗(x ). It follows that F˜ |A and
H |A induce the same framing of the normal bundle N ((F˜ |A)−1(−e1) ∩ A) in A and
by [25, Lem. 4, p. 48], F˜ |A ∼ H |A are homotopic.
Connecting Disconnected Components. In this section, we show that if S1 is a framed
submanifold of X with dimension at least 1 and codimension at least 3, then there exists
a framed submanifold W ⊆ X such that ∂ W = S1 S2 where S2 is connected. This
will finish the proof of Theorem 3, because it follows that for the framed submanifold
S1 := f −1(0), we can construct a strict r -perturbation g of f s.t. g−1(0) = S2 is
connected by Lemma 4. The constraint n + 1 ≤ m ≤ 2n − 3 that we assume in
Theorem 3 implies that the dimension of f −1(0) is at least 1 and that n ≥ 4, so all the
dimensional assumptions of Lemmas 5 and 4 are satisfied.
Lemma 5 Let X be a smooth connected manifold, S1 a framed closed submanifold of
X 27 and assume that 1 ≤ dim S1 ≤ dim X −3. Then there exists a framed submanifold
W in the interior of X such that ∂ W = S1 S2, W induces the framing on S1 S2
and S2 is connected.
The main idea of the proof is to construct a manifold W1 S1 × [
0, 1
], cut out two
holes in S1 ×{1} around x and y that are in different components of S1 ×{1} and connect
them with a tubular (dim S1 + 1)-dimensional neighborhood of a curve connecting x
and y. While there is a well-known construction called “boundary connected sum” for
abstract differential manifolds [22], we could not find any reference that this can be
done all inside the ambient space X , so here we present the sketch of our construction.
Proof Let m := dim X and n := dim X − dim S1. By the product neighborhood
theorem, the framing of S1 determines a diffeomorphism d from S1 × B2n to a (closed)
neighborhood U of S1 in X , where B2n ⊆ Rn is the closed ball of diameter 2. Let
as choose a smooth metric on S1, extend it to a product metric on U via the
diffeomorphism d and smoothly extend it to a metric on the whole of X . Let v1, . . . , vn
be the vector fields on U defined as the d∗-image of the euclidean coordinate vector
fields e1, . . . , en ∈ T B2n identified with vectors of T (S1 × B2n) T S1 × T B2n
orthogonal to T S1 × {0} ⊆ T(s,b)(S1 × B2n). The image d(S1 × [
0, 1
] × {0}) =: W1 is a
27 That is, S1 is compact and without boundary.
Fig. 4 The m − n-dimensional
neighborhoods Dx of x and Dy
of y in S1 are connected via an
m − n + 1 dimensional tubular
neighborhood T of a curve ϕ
connecting x and y
Dx
v1
x
ϕ
y
smooth submanifold of X of dimension m − n + 1 contained in U , with boundary
∂ W1 = S1 S1 where S1 = d(S1 × {1} × {0}). The vectors v2, . . . , vn form a framing
of W1 and the vector field (v1)|W1 is tangent to W1 such that on the boundary, v1|S1
goes in the “inwards” and v1|S1 in the “outwards” direction wrt. W1. Further, S1 is
a diffeomorphic copy of S1, so it has the same number of connected components. Let
x , y ∈ S1 be two points in different components of S1 (Fig. 4).
Let ϕ : [
0, 1
] → X be a smooth embedded curve such that ϕ(0) = x , ϕ(1) = y,
ϕ(t ) ∈/ W1 for t = 0, 1 and there exists an δ > 0 such that ϕ˙(t ) = v1(ϕ(t )) for
t ∈ [0, δ] and ϕ˙(t ) = −v1(ϕ(t )) for t ∈ [1 − δ, 1]. Such curve exists, because
X is connected and the codimension of W1 in X is at least two, so X \ W1 is still
connected. Without loss of generality, we may assume that δ is small enough so that
ϕ([0, δ] ∪ [1 − δ, 1]) is in the image of d : S1 × Bn
2 → U .
The curve ϕ[
0, 1
] is contractible, so any fibre bundle over it is trivial and admits
a global section; in particular, there exists a global section T of the principal bundle
of all framings of its m − 1-dimensional normal bundle. That means, for x = ϕ(t ),
T (x ) := (u1, . . . , um−1) is a framing of Nx ϕ[
0, 1
] and T is smooth. Any other
framing of ϕ[
0, 1
] is determined by a smooth map [
0, 1
] → Gl(m − 1) which linearly
transforms the framing vectors in each ϕ(t ). Let w1, . . . , wm−n be a basis of Tx S1,
resp. Ty S1, oriented so that (w1(x ), . . . , wm−n(x ), v2(x ), . . . , vn(x )) has the same
orientation of Nx ϕ[
0, 1
] as T (x ) and (w1(y), . . . , wm−n(y), v2(y), . . . , vn(y)) has
the same orientation of Ny ϕ[
0, 1
] as T (y). Let as naturally extend the vector fields
w1, . . . , wm−n to ϕ([0, δ] ∪ [1 − δ, 1]) by means of parallel transport along ϕ. The
connectedness of Gl+(m −1) implies that there exists a framing T of N (ϕ[
0, 1
]) such
that T (ϕ(t )) coincides with (w1, . . . , wm−n, v2, . . . , vn) for t ∈ [0, δ] ∪ [1 − δ, 1].
By a slight abuse of notation, we again denote the first m − n framing vectors of T by
w1, . . . , wm−n: these are normal vector fields on ϕ[
0, 1
] extending the already defined
{wi }i in ϕ([0, δ] ∪ [1 − δ, 1]).
For any a ∈ ϕ[
0, 1
] and u ∈ Rm−n, let F (a, u) be equal to γ (1) where γ is a
geodesic, γ (0) = a and γ˙ (0) = u1w1 + u2w2 + . . . + um−nwm−n whenever the
geodesic is defined on [
0, 1
]. If ε > 0 is small enough, then F : ϕ[
0, 1
] × Bεm−n →
X is a smooth embedding and its image is an embedded (m − n + 1)-dimensional
submanifold of X (with corners in {x , y} × ∂ Bεm−n). Using the properties of our
metric, the S1-geodesics in x , resp. y coincide with the geodesics in X , so F maps
{x , y} × Bεm−n to a closed neighborhood Dx Dy of {x , y} in S1, where Dx resp. Dy
is a geodesic ε-ball in S1.
If t < δ or t > 1 − δ, then F (ϕ(t ), Bεm−n) is contained in U and disjoint from W1
due to the choice of the product metric. If t ∈ [δ, 1−δ], then there exists some ε(t ) < ε
and a neighborhood U (t ) of t such that F (ϕ(U (t )), Bεm(t−)n) is disjoint from W1. By
compactness of [
0, 1
], we can make ε smaller and assume that F (ϕ(0, 1) × Bεm−n)
is disjoint from W1. Let T := F (ϕ[
0, 1
] × Bεm−n). This is a smooth contractible
(m − n + 1)-manifold (with corners), so its normal bundle admits a global framing.
By an argument completely analogous to the one above, we may extend the framing
v2, . . . , vn defined on F (ϕ([0, δ] ∪ [1 − δ, 1]) × Bεm−n) to a smooth framing of T .
By construction, T ∩ W1 consists of two (m − n)-discs Dx = F ({x } × Bεm−n) and
Dy in S1. At this point, W1 ∪ T is a framed manifold, but we still need to “smooth the
corners” ∂ Dx and ∂ Dy .
Let ψ : [
0, 1
] → [ 21 ε, ε] be a smooth function such that ψ (0) = ψ (1) = ε, ψ (0) =
−∞, ψ (1) = ∞ and further, for some β > 0, the inverse function (ψ |[0,β])−1 :
[ψ (β), ε] → [0, β] can be extended to a smooth function [ψ (β), ∞) → [0, β] by
sending each x > ε to 0, and similarly (ψ |[1−β,1])−1 : [ψ (1 − β), ε] → [1 − β, 1] can
be extended to a smooth function defined on [ψ (1 − β), ∞) by mapping each x > ε
to 1.28 Finally, define T ⊆ T by
T := {F (ϕ(t ), u) : |u| ≤ ψ (t )}.
We claim that W := W1 ∪ T is a smooth manifold with boundary. By construction,
W1 and T \ W1 are smooth manifolds with W1 ∩ T = Dx ∪ Dy , so we just need to
analyze their intersection (Fig. 5).
Let u be in the interior of Dx , resp. Dy . Let V be an open neighborhood of v with
positive distance from ∂ Dx (resp. ∂ Dy ) and define an S1-chart φu : V → Rm−n that
maps a neighborhood V ⊆ Dx (resp. Dy ) of u to Rm−n. Let N ⊆ X be a neighborhood
of v in X that is disjoint from the topological boundary of W in X and N ∩ S1 ⊆ V .
Then the diffeomorphism (φu , id) ◦ d−1 takes N to an open subset of Rm such that
N ∩W is the preimage of Rm−n+1 ×{0} (the projection to the (m −n +1)’th component
of the image of N is a neighborhood of 1 ∈ R).
It remains to show that any v ∈ ∂ Dx resp. ∂ Dy is in the boundary of W = T ∪ W1,
that is, some neighborhood of v in W is mapped by an X -chart to Rm−n ×(−∞, 1]×{0}.
Let φv be an S1-chart mapping a neighborhood V of x in S1 to Rm−n such that
uv(v) = 0 and let N := d−1(V × (0, 1 + δ)) ⊆ X be a neighborhood of v in X . Let
(φv, id) ◦ d−1 : N → Rm be an X -chart: it maps
– S1 ∩ N to Rm−n × {1} × {0},
– W1 ∩ N to Rm−n × (0, 1] × {0},
– T ∩ N to {(z1, . . . , zm−n, y, 0, . . . , 0) : 1 ≤ y ≤ ω(z)}, and
– W ∩ N to {(z1, . . . , zm−n, y, 0, . . . , 0) : 0 ≤ y ≤ ω(z)}
where ω(x ) = 1 whenever x ∈/ Dx and ω(z) = sup {t + 1 : d(z, t, 0) ∈ T }.
If e(u) is the geodesic distance of u from x , then ω(z) = sup {t + 1 : e(z) ≤
ψ (t )}. The function ψ maps a neighborhood of 0 to some (ε − α, ε] and the inverse
function ψ −1 to this restriction has derivative 0 in ε. If we extend ψ −1(a) to be 0
for a > ε, we get a smooth function from a neighborhood of ε in R to nonnegative
numbers with ψ −1(ε) = (ψ −1) (ε) = 0. We can rewrite ω(z) to ψ −1(e(z)) + 1
28 Equivalently, the graph of ψ united with {0, 1} × [1, ∞) is a smooth submanifold of R2.
Fig. 5 Illustration of corner
smoothing. T ⊆ T is chosen so
that, in a neighborhood of
v ∈ ∂ Dx , T consists of points
in W1 ∪ T whose distance from
W1 is bounded by a function
ω(z) that has zero gradient in v
T
u
T
ω(z )
z
S 1
to see that it is a smooth function defined on a neighborhood of v in S1 with zero
gradient in v (note that e(v) = ε). It follows that W ∩ N is diffeomorphic to some
neighborhood of (0, . . . , 0, 1, 0, . . . , 0) in {(z1, . . . , zm−n , y, 0, . . . , 0) : y ≤ ω(z)}
which is diffeomorphic to Rm−n × (−∞, 0] × {0}.
Hence W is a smooth framed submanifold of X of dimension m −n +1, the framing
of the normal bundle being a restriction of the framing of W1 and T . If m − n ≥ 2,
then dim ∂ Dx Sm−n−1 is a sphere of dimension at least one and hence is connected:
by construction, ∂ Dx and ∂ Dy are in the same component of the boundary ∂ W . If
m − n = 1, then S1 is a finite disjoint union of circles and yields ∂ W to be a connected
sum of two circles containing x and y united with S1 and other components of S1.
In both cases, the number of connected components of ∂ W \ S1 is smaller then the
number of connected components of S1. The compactness of X and S1 implies that S1
has only a finite number of components. In the same way as above, we may continue
attaching tubular neighborhoods of curves connecting different components of S1 to
obtain a framed manifold W such that ∂ W \ S1 =: S2 is already connected.
Appendix 3: Proof of Theorem 5
We need the following observation, cf. [5, p. 9].
Lemma 6 Let (X, A) be an m-dimensional pair of simplicial complexes such that
A contains the (i − 1)-skeleton of X . Then there is an (m − i )-subcomplex Y of a
subdivision X of X such that Y is disjoint with A and X ⊆ A ∗ Y .
Moreover, for each (m − i )-simplex τ in Y there is an i -simplex σ of X \ A such
that the following holds: σ ∗ τ is a simplex of X if and only if σ is a face of σ .
The subcomplex Y from the previous lemma is called the dual complex to A in X .
Also the simplex σ is called the dual simplex to τ .
Proof We proceed by induction on the dimension of X . When m = i −1, the statement
holds for the empty subcomplex Y of X .
When m ≥ i we can use the induction hypothesis on X (m−1) to obtain the
dual complex Y (m−i−1) ⊆ X (m−1) where X (m−1) is a subdivision of X (m−1). For
each m-simplex σ of X \ A we have that σ ∼= ∂σ ∗ bσ where bσ is the
barycenter of σ . Since ∂σ can be considered as a subcomplex of X (m−i−1), we have
that X (m−1) ∪ σ is a subspace of X (m−1) ∗ bσ . Similarly whole X is a subspace
of X (m−i) ∗ {bσ : σ is an m-simplex of X }. Therefore the desired Y can be set to
Y (m−i−1) ∗ {bσ : σ is an m-simplex of X }.
To prove the second statement, we need to distinguish two cases. First, when m = i ,
then the dual simplex to each [bσ ] is σ . Second, when m > i , each (m − i )-simplex
τ has the form b ∗ τ for some (m − i − 1)-simplex τ of Y (m−i−1). The dual simplex
σ to τ is equal to the dual simplex of τ from the induction since σ ∗ τ ∈ X if and
only if σ ∗ τ ∈ X (m−1).
Observation 1 directly follows from the following lemma:
Lemma 7 Let f : K → Rn be a map such that (X, A) := | f |−1([0, r ], {r }) is pair of
simplicial complexes. If the map f |A can be extended to a map f (i−1) : X (i−1) ∪ A →
Sn−1, then there is an r -perturbation g of f such that g−1(0) is an (m −i )-dimensional
subcomplex of some subdivision of X .
Proof To prove the lemma we need to find an extension g : X → Rn of a given map
f (i−1) : X (i−1) ∪ A → Sn−1 such that g−1(0) is a simplicial complex of dimension
at most m − i . Let Y be the dual complex to X (i−1) ∪ A in X . We define the extension
g : X → Rn by g := f ∗ 0Y , that is, for each point (x , t, y) of each simplex σ ∗ τ
in X ⊆ (X (i−1) ∪ A) ∗ Y we define g(x , t, y) := t f (x ) + (1 − t )0. Clearly, g−1(0)
Proof of Theorem 5 We are given an m-dimensional simplicial pair (X, A), natural
numbers i and n such that n < i ≤ (m + n)/2 − 1 and a map h : X (i−1) ∪ A → Sn−1.
To prove the theorem, we will find an extension g : X → Rn of h such that g−1(0)
is a cell complex obtained from an (m − i − 1)-dimensional simplicial complex by
attaching cells of dimension m − n.
We will use the dual complex Y to X (i−1) ∪ A in X as in Lemma 6. Part of the
map g is easy to define, namely, for every (x , t, y) ∈ (X (i−1) ∪ A) ∗ Y (m−i−1) we set
g(x , t, y) := (1 − t )h(x ). (Note that for t = 1, we have (x , t, y) ∈ Y (m−i−1).) For
the rest, we need to define g on each := (∂σ ) ∗ τ where τ is an arbitrary (m − i
)simplex of Y and σ is its dual i -simplex in X \ A. We have that = ∂σ ∗ (bτ ∗ ∂τ ) =
(∂σ ∗ bτ ) ∗ ∂τ where bτ is the barycenter of τ . Therefore we can write each point p
of the join (∂σ ∗ bτ ) ∗ ∂τ as p = (x , s, t, y) where x ∈ ∂σ , y ∈ ∂τ and s, t ∈ [
0, 1
].
(We have p ∈ ∂σ for s = 0 and p ∈ τ for t = 1.)
1. In the case where h|∂σ is homotopically nontrivial, we define
g(x , s, t, y) := (1 − t ) (1 − s)h(x ) − t sι(y) ,
where ι : ∂τ ∼= Sm−i−1 → Sn−1 is an embedding of ∂τ to the equatorial (m−i
−1)subsphere. Here we need that m − i ≤ n.
2. Otherwise, we choose an arbitrary extension h : (∂σ ) ∗ bτ → Sn−1 ⊆ Rn of h|∂σ
and define
g(x , s, t, y) := (1 − t )h (x , s).
We can see that (g| )−1(0) is equal to ∂τ in this case.
To finish the proof, it suffices to show that in the case 1 above, (g| )−1(0) ∼= Cone(η)
for some η : Sm−n−1 → ∂τ ∼= Sm−i−1. Roughly speaking, we need to solve the
equation h|∂σ (x ) = ι(y), that is, to identify the (h|∂σ )-preimage of equatorial
(m − i − 1)-subsphere of Sn−1. Informally, our strategy is to employ the fact that
the elements [h|∂σ ] of the stable homotopy group πi−1(Sn−1) are iterated suspensions
and thus, without loss of generality, the (h|∂σ )-preimage of the equatorial (m − i −
1)subsphere is the equatorial subsphere of the same codimension (and the map η above
is the restriction of h onto this subsphere).
– Formally, by the Freudenthal suspension theorem we know that [h|∂σ ] equals a
j -fold suspension Σ j [η] for some η : Si−1− j → Sn−1− j assuming the condition
i − 1 − j ≤ 2(n − 1 − j ) − 1. Given the requirement n − 1 − j = m − i − 1, the
condition is equivalent to i ≤ (m + n)/2 − 1—the assumption of the theorem.
– Without loss of generality, we can assume that h|∂σ = Σ j η. Indeed, in general,
there is a homotopy H : h|∂σ ∼ Σ j η. We can parameterize a regular neighborhood
N of ∂σ in (∂σ ∗ bτ ) by N ∼= ∂σ × [0, 1). Since \ (N ∗ ∂τ ) ∼= , the map g
can be defined on the domain \ (N ∗ ∂τ ) via the same formula as above. For
each point (x , s, t, y) of N ∗ ∂τ we define g(x , s, t, y) := α H (s, t ).
– Now it is easy to see that
(g| )−1(0) =
, η(x ) ∈
: x ∈ Sm−n−1 and s ∈ [0.5, 1] ,
where Sm−n−1 denotes the equatorial subsphere of ∂σ ∼= Si−1. Because of the
identifications (x , s, 1, y) ∼ (x , s , 1, y) and (x , 1, 0, y) ∼ (x , 1, 0, y ) in the
join = (∂σ ∗ bτ ) ∗ ∂τ , we get that the zero set is homeomorphic to {(x , s) ∈
Sm−n−1 × [0.5, 1]}/∼ where the equivalence ∼ is defined by
(x , 1) ∼ (x , 1) for each x , x and (x , 0.5) ∼ (x , 0.5) when η(x ) = η(x ).
But this space is homeomorphic to Cone(η) by definition.
Appendix 4: Cap Product in Cˇ ech (Co)homology
Let (X, A) be a pair of topological spaces and denote H∗(X, A) (H ∗(X, A)) the Cˇ ech
(co)homology of (X, A); we assume a fixed coefficient group G for the homology and
the constant sheaf G for the cohomology. If U is a covering of X and V a refinement,
then UA := {U ∩ A : U ∈ U } is a covering of A and VA is its refinement. We
associate to U the nerv N (U ) of the covering and further define the nerv N (UA) to
be a simplicial complex whose q-simplices are all sets {U0, . . . , Uq } ⊆ U containing
q + 1 elements such that (∩qj=0U j ) ∩ A = ∅: this is a subcomplex of N (U ). A map
p : V → U that maps each set in V to a superset in U , induces a map pA : VA → UA
and maps on the nerves N (V) → N (U ), N (VA) → N (UA), which are simplicial maps
between simplicial complexes. Any other choice p : V → U yields a homotopic map
N (V) → N (U ), so a subcovering induces well-defined maps p∗ resp. p∗ between
simplicial (co)homologies of the nerves p∗ : H∗(X, A, V) → H∗(X, A, U ) and p∗ :
H∗(X, A, U ) → H∗(X, A, V). The Cˇ ech (co)homology is defined by H∗(X, A) :=
lim −→U H∗(X, A, U ).
←−LUeHtφ∗(∈XH,A∗(,XU,) Aan)danHd ∗β(X∈,HA∗)(:X=, lAim∪ B) for some A, B ⊆ X . The direct limit can
be defined as a disjoint union of all H∗(X, A, U ) factored by the relation λ − p∗(λ)
and the inverse limit can be defined as {μ ∈ H∗(X, A ∪ B, U ) : p∗(μV ) =
μU }. Therefore, β can be represented as a net (βU )U so that βU = p∗βV for any
refinement V of any covering U , and φ can be represented by an element φ ˜ contained
U
in H∗(X, A, U˜ ) for some fine enough U˜ .
For each such U˜ , we define αU˜ := φU˜ βU˜ ∈ H∗(X, B, U˜ ) by means of simplicial
cap product. If V˜ is a refinement of U˜ , then the naturality of simplicial cap product—
nTahmereelfyo,rteh,ewreelmataioyncφonU˜sistenpt∗l(yβdV˜e)fi=ne pα∗U( pfo∗r(φaUn˜y) U toβVb˜ e) —eqiumaplltioespt∗hαaVt αwU˜he=repV∗ αisV˜a.
fine enough refinement of U such that αV is already defined. This construction yields
an element α ∈ H∗(X, B) and we define the Cˇ ech cap product φ β to be equal to
α.
Let f : X → X , A, B ⊆ X , A , B ⊆ X , f ( A) ⊆ A and f (B) ⊆ B . We will
show the naturality of , that is, the relation φ f∗(β) = f∗( f ∗(φ ) β) for
φ ∈ H ∗(X , A ) and β ∈ H∗(X, A ∪ B).
Let U be a covering of X and U a covering of X so that for any U ∈ U there
exists U ∈ U such that f (U ) ⊆ U . Given U and f , then any covering of X admits
a refinement U with this property. Such map f and a choice of the set U for any U
induces a simplicial map f : N (U ) → N (U ); other assignment of the supersets U
of f (U ) yields a homotopic map. If V resp. V is a refinement of U resp. U such that
f maps each V ∈ V into some V ∈ V , then we have a square of simplicial maps
between simplicial complexes
N (U )
p
N (V)
f
f
f
N (U )
p
N (V )
(9)
that commute up to homotopy and induce commuting maps on the level of their
simplicial (co)homology. Thus, if β ∈ H∗(X, B) is represented by (βU )U , then f∗v can
be defined to by the net that assigns to each U the f∗-image of βU ∈ H(X, A ∪ B, U )
where U is fine enough so that f maps elements of U to U . This is well defined, because
if V is a refinement of U , then the commutativity of the induced maps in the upper
left triangle of (9) implies f∗βV = f∗ p∗βV = f∗βU ; further, any U˜ such that f maps
its element to elements of U has a common refinement with U . The commutativity of
the induced maps in the lower right triangle of (9) implies that p∗( f∗β)V = ( f∗β)U ,
hence f∗β is a well defined element of H∗(X, B).
Let φ ∈ H ∗(X , A ) be represented by some φ ˜ ∈ H∗(X , A , U˜ ). By
construcU
tion, φ f∗β is represented by the net that assigns to any refinement V˜ of U˜ the
cap product of
φ
˜
V
f∗(βU ) ∈ H∗( X , B , V˜ ),
where U is fine enough so that f maps each U ∈ U to some V˜ ∈ V˜ . It remains to
show that f∗( f ∗φ β) is represented by the same object. Let V˜ be a refinement
of U˜ , U be as in (10) and let V be a refinement of U fine enough so that f maps
each of its sets to some element of V˜ . The term f∗(β ) in (10) equals f∗(β ) and
U V
f ∗(φ ) can be represented by f ∗(φ ) ∈ H∗( X, A, V ). It follows that f∗( f ∗φ β)
˜
V
can be represented by a net that assigns to V˜ the element f∗( f ∗(φ )
βV ) ∈
22. Kosinski, A.: Differential Manifolds. Dover Books on Mathematics Series. Dover Publications,
Mineola (2007)
23. Mardešic´, S.: Thirty years of shape theory. Math. Commun. 2(1), 1–12 (1997)
24. Milnor, J.W.: On the Steenrod Homology Theory. Novikov Conjectures, Index Theorems and Rigidity,
vol. 1. Cambridge University Press, Cambridge (1960)
25. Milnor, J.W.: Topology from the Differential Viewpoint. Princeton University Press, Princeton (1997)
26. Munkres, J.: Elementary Differential Topology. Annals of Mathematics Studies. Princeton University
Press, Princeton (1966)
27. Neumaier, A.: Interval Methods for Systems of Equations. Cambridge University Press, Cambridge
(1990)
28. Prasolov, V.V.: Elements of Homology Theory. Graduate Studies in Mathematics. American
Mathematical Society, Providence, RI (2007)
29. Škraba, P., Wang, B., Guoning, C., Rosen, P.: 2D vector field simplification based on robustness. In:
IEEE Pacific Visualization (PacificVis) (2014) http://www.sci.utah.edu/~beiwang/publications/VF_
Robustness_Simplification_BeiWang_2014. To appear in IEEE Pacific Visualization (PacificVis)
30. Spanier, E.H.: Algebraic Topology. Springer, New York (1994)
31. Wallace, A.: Algebraic Topology: Homology and Cohomology. Dover Books on Mathematics Series.
Dover Publications, Mineola (2007)
32. Whitehead, J.: On the theory of obstructions. Ann. Math. 54, 68–84 (1951)
1. Abraham , R. , Robbin , J.: Transversal Mappings and Flows. W.A. Benjamin , New York ( 1967 )
2. Alefeld , G.E. , Potra , F.A. , Shen , Z. : On the existence theorems of Kantorovich, Moore and Miranda . Tech. Rep . 01 /04, Institut für Wissenschaftliches Rechnen und Mathematische Modellbildung ( 2001 )
3. Ben-Tal , A. , Ghaoui , L. , Nemirovski , A.: Robust Optimization . Princeton Series in Applied Mathematics. Princeton University Press, Princeton ( 2009 )
4. Bendich , P. , Edelsbrunner , H. , Morozov , D. , Patel , A. : Homology and robustness of level and interlevel sets . Homology Homotopy Appl . 15 ( 1 ), 51 - 72 ( 2013 )
5. Bryant , J.: Piecewise Linear Topology . In: Daverman, R.J. , et al. (eds.) Handbook of Geometric Topology, pp. 219 - 259 . Elsevier, Amsterdam ( 2002 )
6. Cˇ adek, M., Krcˇál , M. , Matoušek , J. , Sergeraert , F. , Vokrˇínek , L. , Wagner , U. : Computing all maps into a sphere . J. ACM 61 ( 3 ), 17 : 1 - 17 : 44 ( 2014 )
7. Carlsson , G.: Topology and data . Bull. Am. Math. Soc. (New Ser.) 46 ( 2 ), 255 - 308 ( 2009 )
8. Chazal , F. , Patel , A. , Škraba , P. : Computing the robustness of roots . Appl. Math. Lett. 25 ( 11 ), 1725 - 1728 ( 2012 )
9. Collins, P. : Computability and representations of the zero set . Electron. Notes Theor. Comput. Sci . 221 , 37 - 43 ( 2008 )
10. Edelsbrunner , H. , Harer , J.L.: Computational Topology . American Mathematical Society , Providence, RI ( 2010 )
11. Edelsbrunner , H. , Morozov , D. , Patel , A. : Quantifying transversality by measuring the robustness of intersections . Found. Comput. Math . 11 ( 3 ), 345 - 361 ( 2011 )
12. Eilenberg , S. , Steenrod , N.E. : Axiomatic approach to homology theory . Proc. Natl. Acad. Sci. USA 31 ( 4 ), 117 - 120 ( 1945 )
13. Franek , P. , Krcˇál, M.: Persistence of Zero Sets. ( 2015 ) Preprint http://arxiv.org/abs/1507.04310
14. Franek , P. , Krcˇál, M.: Robust satisfiability of systems of equations . In: Proceedings of the Annual ACM-SIAM Symposium on Discrete Algorithms (SODA) ( 2014 ). Extended version accepted to J. ACM. Preprint in http://arxiv.org/abs/1402.0858
15. Franek , P. , Ratschan , S. , Zgliczynski , P. : Quasi-decidability of a fragment of the analytic first-order theory of real numbers ( 2012 ). Preprint in http://arxiv.org/abs/1309.6280
16. Frommer , A. , Lang , B. : Existence tests for solutions of nonlinear equations using Borsuk's theorem . SIAM J. Numer. Anal . 43 ( 3 ), 1348 - 1361 ( 2005 )
17. Goudail , F. , Réfrégier , P. : Statistical Image Processing Techniques for Noisy Images: An ApplicationOriented Approach . Kluwer Academic/Plenum Publishers, New York ( 2004 )
18. Hatcher , A.: Algebraic Topology . Cambridge University Press, Cambridge ( 2001 )
19. Higham , N.: Accuracy and Stability of Numerical Algorithms, 2nd edn . Society for Industrial and Applied Mathematics , Philadelphia ( 2002 )
20. Hirsch , M. : Differential Topology . Graduate Texts in Mathematics. Springer, New York ( 2012 )
21. Kearfott , R.B.: On existence and uniqueness verification for non-smooth functions . Reliab. Comput . 8 ( 4 ), 267 - 282 ( 2002 )