The Central Idea
A topological invariant is a property, number, or algebraic object that is preserved under homeomorphism.
If
\[ X\cong Y, \]then every topological invariant must agree for \(X\) and \(Y\).
Symbolically,
\[ \boxed{ X\cong Y \Longrightarrow I(X)=I(Y) }. \]Invariants as Tests for Non-Homeomorphism
The most important practical use of invariants is often negative.
If one invariant differs, then the spaces cannot be homeomorphic:
\[ \boxed{ I(X)\ne I(Y) \Longrightarrow X\not\cong Y }. \]A single difference can settle the comparison.
Matching Invariants Do Not Always Prove Homeomorphism
The converse is usually false.
Two spaces may share several invariants and still fail to be homeomorphic.
\[ I(X)=I(Y) \centernot\Longrightarrow X\cong Y. \]Invariants are often incomplete descriptions of topology.
Qualitative vs. Numerical Invariants
Some invariants are qualitative.
Examples:
- connected or disconnected;
- compact or noncompact;
- orientable or nonorientable;
- path connected or not path connected.
Others assign numerical values, such as:
- number of connected components;
- genus;
- Euler characteristic;
- crossing number of a knot.
Algebraic Invariants
More advanced topology assigns algebraic structures to spaces.
Examples include:
- fundamental groups;
- homology groups;
- cohomology groups;
- knot groups;
- polynomial invariants.
These invariants can detect distinctions that simpler properties miss.
Connectedness
Connectedness is a topological invariant.
If
\[ X\cong Y, \]then
\[ X\text{ connected} \iff Y\text{ connected}. \]A homeomorphism cannot tear one connected piece into two.
Example: Connectedness Distinguishes Spaces
Compare
\[ X=(0,1) \]and
\[ Y=(0,1)\cup(2,3). \]The first is connected.
The second is disconnected.
Therefore,
\[ \boxed{X\not\cong Y}. \]Number of Connected Components
The number of connected components is also a topological invariant.
If one space has \(3\) components and another has \(5\), then they cannot be homeomorphic.
\[ \boxed{ \#\operatorname{Comp}(X) \ne \#\operatorname{Comp}(Y) \Longrightarrow X\not\cong Y }. \]Path Connectedness
Path connectedness is preserved under homeomorphism.
If every pair of points in \(X\) can be joined by a continuous path, then the corresponding points in \(Y\) can be joined by the image of that path.
Thus,
\[ \boxed{ X\cong Y \Longrightarrow \bigl( X\text{ path connected} \iff Y\text{ path connected} \bigr) }. \]Compactness
Compactness is a topological invariant.
If \(X\) is compact and
\[ f:X\to Y \]is a homeomorphism, then \(Y=f(X)\) is compact because continuous images of compact spaces are compact.
Applying the inverse gives the converse.
Example: Compactness Distinguishes Intervals
Compare
\[ [0,1] \]and
\[ (0,1). \]The closed interval is compact.
The open interval is not compact.
Therefore,
\[ \boxed{[0,1]\not\cong(0,1)}. \]Boundedness Is Not a Topological Invariant
Boundedness depends on a metric or geometric embedding.
The interval
\[ (0,1) \]is bounded, while
\[ \mathbb{R} \]is unbounded.
Yet
\[ (0,1)\cong\mathbb{R}. \]Therefore boundedness is not topological.
Length Is Not a Topological Invariant
The intervals
\[ [0,1] \]and
\[ [0,1000] \]have very different lengths.
But a linear map gives a homeomorphism between them.
Therefore length is not a topological invariant.
Angle Is Not a Topological Invariant
Homeomorphisms may bend and distort a space.
They do not need to preserve angles.
A square boundary is homeomorphic to a circle even though one has corners and the other does not.
Area and Volume Are Not Topological Invariants
A homeomorphism may stretch regions and change their area or volume.
Therefore ordinary Euclidean measure is not preserved in general.
Topology ignores these metric quantities unless additional structure is imposed.
Local Dimension
Dimension is a topological invariant for manifolds and Euclidean spaces.
In particular,
\[ \mathbb{R}^m\not\cong\mathbb{R}^n \]whenever
\[ m\ne n. \]This fact is supported by deep results such as invariance of domain and invariance of dimension.
Example: The Line and Plane
The spaces
\[ \mathbb{R} \]and
\[ \mathbb{R}^2 \]are not homeomorphic.
One elementary clue comes from deleting a point:
\[ \mathbb{R}\setminus\{0\} \]is disconnected, while
\[ \mathbb{R}^2\setminus\{(0,0)\} \]is path connected.
Cut-Point Behavior
A cut point is a point whose removal disconnects a connected space.
Cut-point behavior is preserved under homeomorphism.
If \(f:X\to Y\) is a homeomorphism and \(x\in X\), then
\[ X\setminus\{x\} \cong Y\setminus\{f(x)\}. \]Thus a cut point must correspond to a cut point.
Example: Circle vs. Closed Interval
Both
\[ S^1 \]and
\[ [0,1] \]are compact and connected.
But interior points of \([0,1]\) are cut points, while no point of \(S^1\) is a cut point.
Therefore,
\[ \boxed{S^1\not\cong[0,1]}. \]Orientability
Orientability is a topological invariant of surfaces.
A homeomorphism cannot convert an orientable surface into a nonorientable one.
Therefore orientability is a powerful classification tool.
Example: Torus vs. Klein Bottle
The torus and Klein bottle both have Euler characteristic
\[ 0. \]But the torus is orientable and the Klein bottle is nonorientable.
Hence,
\[ \boxed{T^2\not\cong K}. \]Genus
For connected closed orientable surfaces, genus is a topological invariant.
It records the number of handles.
Thus:
\[ g(S^2)=0, \] \[ g(T^2)=1, \] \[ g(\text{double torus})=2. \]Genus Distinguishes Closed Orientable Surfaces
For connected closed orientable surfaces,
\[ g_1\ne g_2 \]implies
\[ \boxed{S_{g_1}\not\cong S_{g_2}}. \]In fact, genus completely classifies connected closed orientable surfaces up to homeomorphism.
Euler Characteristic
The Euler characteristic is a numerical topological invariant.
For a finite cell decomposition of a surface,
\[ \boxed{\chi=V-E+F}. \]Different subdivisions may change \(V\), \(E\), and \(F\), but not their alternating combination.
Euler Characteristic of Common Surfaces
For the sphere,
\[ \chi(S^2)=2. \]For the torus,
\[ \chi(T^2)=0. \]For a double torus,
\[ \chi=-2. \]Different Euler characteristics prove non-homeomorphism.
Euler Characteristic and Genus
For a connected closed orientable surface,
\[ \boxed{\chi=2-2g}. \]Therefore Euler characteristic and genus encode the same classification information within this restricted family.
Boundary Components
The number of boundary components of a compact surface is a topological invariant.
A disk has
\[ b=1, \]while an annulus has
\[ b=2. \]Therefore they cannot be homeomorphic.
Example: Disk vs. Annulus
A disk and annulus are both compact, connected, and orientable.
But their boundary-component counts differ:
\[ b(D^2)=1, \qquad b(A)=2. \]Therefore,
\[ \boxed{D^2\not\cong A}. \]Boundary Type Is Intrinsic for Manifolds with Boundary
For manifolds with boundary, homeomorphisms send boundary points to boundary points.
Interior points have neighborhoods modeled on
\[ \mathbb{R}^n, \]while boundary points have neighborhoods modeled on a half-space.
This local distinction is topological.
Compact Surface Classification Data
For compact connected surfaces, the main classification data are:
- orientability;
- genus or nonorientable genus;
- number of boundary components.
Euler characteristic is closely related to these quantities.
Classification as a Complete Invariant System
Within the class of compact connected surfaces, the combination of orientability, genus type, and boundary count forms a complete classification system.
In that setting, matching full classification data is enough to conclude homeomorphism.
The Fundamental Group
The fundamental group
\[ \pi_1(X) \]is an algebraic invariant that records how loops in \(X\) can be continuously deformed.
It becomes especially useful when simpler invariants fail to distinguish spaces.
Simply Connected Spaces
A path connected space is simply connected if every loop can be continuously contracted to a point.
In terms of the fundamental group,
\[ \boxed{ X\text{ simply connected} \Longrightarrow \pi_1(X)\text{ is trivial} }. \]Example: Circle vs. Disk
Both the circle and disk are path connected.
But:
\[ \pi_1(D^2) \]is trivial, while
\[ \pi_1(S^1)\cong\mathbb{Z}. \]Therefore the circle and disk are not homeomorphic.
Homology Groups
Homology assigns a sequence of algebraic groups
\[ H_0(X),H_1(X),H_2(X),\ldots \]to a topological space.
Very roughly:
- \(H_0\) detects connected components;
- \(H_1\) detects independent loop-like holes;
- \(H_2\) detects enclosed surface-type voids;
- higher groups detect higher-dimensional structure.
Betti Numbers
The ranks of homology groups are called Betti numbers.
They are written
\[ b_0,b_1,b_2,\ldots \]and give numerical summaries of topological features in different dimensions.
Euler Characteristic from Betti Numbers
For suitable spaces,
\[ \boxed{ \chi = b_0-b_1+b_2-b_3+\cdots }. \]This connects elementary counting with algebraic topology.
Example: Sphere Betti Numbers
For the two-sphere,
\[ b_0=1, \qquad b_1=0, \qquad b_2=1. \]Therefore,
\[ \chi=1-0+1 \] \[ \boxed{\chi=2}. \]Example: Torus Betti Numbers
For the torus,
\[ b_0=1, \qquad b_1=2, \qquad b_2=1. \]Therefore,
\[ \chi=1-2+1 \] \[ \boxed{\chi=0}. \]Knot Invariants
Knot theory studies embeddings of circles in three-dimensional space.
Useful invariants include:
- minimal crossing number;
- unknotting number;
- knot genus;
- Alexander polynomial;
- Jones polynomial;
- knot group.
Crossing Number
The crossing number
\[ c(K) \]is the smallest number of crossings in any diagram of a knot.
For the unknot,
\[ c=0. \]For the trefoil,
\[ c=3. \]Therefore the two knot types are distinct.
Linking Number
For an oriented two-component link, linking number is a topological invariant.
The standard Hopf link has
\[ |\operatorname{lk}|=1, \]while the two-component unlink has
\[ \operatorname{lk}=0. \]Therefore they are distinct links.
Invariants Can Be Coarse or Fine
A coarse invariant distinguishes only broad classes of spaces.
A finer invariant distinguishes more cases.
Connectedness is coarse.
Fundamental groups and homology groups are often finer.
No single invariant is automatically best for every problem.
A Complete Invariant
A complete invariant for a class of objects is an invariant that agrees exactly when the objects are equivalent.
In other words,
\[ I(X)=I(Y) \iff X\cong Y \]within the specified class.
Complete invariants are rare and usually depend on restricting the class of spaces being studied.
Example of a Complete Classification System
For compact connected surfaces, the combination of:
- orientability;
- genus type;
- boundary-component count;
gives complete classification up to homeomorphism.
Thus the invariant system is complete within that family.
Invariants and Deformation
A valid invariant must survive every homeomorphism.
Stretching, bending, twisting, or smoothly reshaping a space cannot change its topological invariants.
Cutting, gluing, or changing identifications may change them.
Example: Sphere to Ellipsoid
Stretching a sphere into an ellipsoid changes lengths and curvature.
But:
\[ \text{connectedness}, \quad \text{compactness}, \quad g=0, \quad \chi=2 \]remain unchanged.
The sphere and ellipsoid are homeomorphic.
Example: Sphere to Torus
Creating a handle changes:
\[ g:0\to1 \]and
\[ \chi:2\to0. \]Therefore a sphere cannot be turned into a torus by homeomorphism alone.
Invariants and Quotient Spaces
Quotient constructions can change topology by identifying points.
Such identifications may alter:
- connectedness;
- fundamental group;
- Euler characteristic;
- orientability;
- boundary structure.
Comparing invariants before and after a quotient helps reveal what changed.
Invariants and Local Structure
Some invariants describe global structure.
Others capture local behavior.
Examples of local topological information include:
- local dimension;
- whether a point is a manifold boundary point;
- local connectedness;
- local compactness.
Homeomorphisms preserve both local and global topological structure.
Choosing the Right Invariant
When comparing spaces, start with the simplest invariants.
A practical order might be:
- number of components;
- connectedness;
- compactness;
- path connectedness;
- boundary behavior;
- orientability;
- Euler characteristic or genus;
- fundamental group or homology when needed.
Why Start with Simple Invariants?
If compactness already distinguishes two spaces, there is no need to compute a fundamental group.
Good topological reasoning uses the least complicated invariant that solves the problem.
Common Error: Treating a Geometric Quantity as Topological
Length, angle, area, and curvature are generally not topological invariants.
A homeomorphism may change all of them.
Always ask whether the quantity is preserved under arbitrary homeomorphism.
Common Error: Thinking Equal Invariants Prove Homeomorphism
Equal values of one invariant usually provide only necessary evidence.
For example, the torus and Klein bottle both have
\[ \chi=0, \]yet they are not homeomorphic.
Common Error: Thinking Every Useful Property Is an Invariant
A property is topological only if every homeomorphism preserves it.
Visual appearance alone is not enough.
One must test the property against the definition of homeomorphism.
Common Error: Confusing Intrinsic and Extrinsic Properties
Intrinsic properties belong to the space itself.
Extrinsic properties depend on how the space sits inside another space.
Knotting, for example, concerns an embedding of \(S^1\) in three-dimensional space rather than the intrinsic topology of the circle alone.
Common Error: Assuming Euler Characteristic Is Universal
Euler characteristic is extremely useful but does not distinguish every topological space.
It is one invariant among many.
Stronger invariants may be needed when Euler characteristics agree.
Common Error: Assuming Fundamental Group Is Complete
Even fundamental groups do not classify all topological spaces.
Distinct spaces can have isomorphic fundamental groups.
Topology often requires several layers of invariants.
A Reliable Procedure for Comparing Spaces
- Identify the category of spaces being compared.
- List easy invariants first.
- Compare connectedness and component structure.
- Compare compactness.
- Compare local or boundary behavior.
- For surfaces, compare orientability, genus, boundary count, and Euler characteristic.
- Use algebraic invariants if simpler tests agree.
- If one invariant differs, conclude non-homeomorphism.
A Reliable Procedure for Evaluating a Proposed Invariant
- State the property or quantity clearly.
- Ask whether it is preserved by every homeomorphism.
- Try simple homeomorphic examples with different geometry.
- If the property changes under a homeomorphism, it is not topological.
- If it is known to be preserved, it can be used as an invariant.
Example: Is Boundedness an Invariant?
Compare
\[ (0,1) \]and
\[ \mathbb{R}. \]They are homeomorphic, but one is bounded and the other is not.
Therefore,
\[ \boxed{\text{boundedness is not a topological invariant}}. \]Example: Is Compactness an Invariant?
Yes.
Homeomorphisms are continuous in both directions, and continuous images of compact spaces are compact.
Therefore,
\[ \boxed{\text{compactness is a topological invariant}}. \]Example: Is Genus an Invariant?
For connected closed orientable surfaces, yes.
A sphere and torus have different genus:
\[ 0\ne1. \]Therefore they are not homeomorphic.
Example: Is Crossing Count of One Diagram an Invariant?
No.
Reidemeister moves can add or remove crossings in a knot diagram without changing knot type.
The invariant is the minimal crossing number, not the crossing count of an arbitrary diagram.
Practice 1: Definition
Complete:
\[ X\cong Y \Longrightarrow \boxed{\phantom{I(X)=I(Y)}}. \]Practice 2: Different Invariants
If
\[ I(X)\ne I(Y), \]what can be concluded?
Practice 3: Compactness
Can
\[ [0,1] \]be homeomorphic to
\[ (0,1)? \]Name the invariant that answers the question.
Practice 4: Connectedness
Can
\[ (0,1) \]be homeomorphic to
\[ (0,1)\cup(2,3)? \]Practice 5: Sphere vs. Torus
Give one numerical invariant that distinguishes
\[ S^2 \]from
\[ T^2. \]Practice 6: Torus vs. Klein Bottle
Both have Euler characteristic \(0\). What invariant distinguishes them?
Practice 7: Disk vs. Annulus
What invariant distinguishes a disk from an annulus?
Practice 8: Non-Invariant
Is boundedness a topological invariant?
Practice 9: Knot Theory
Is the number of crossings in one arbitrary knot diagram a knot invariant?
Practice 10: Complete Invariant
What does it mean for an invariant to be complete within a class of spaces?
Practice Answers
Practice 1
\[ \boxed{I(X)=I(Y)}. \]Practice 2
The spaces cannot be homeomorphic:
\[ \boxed{X\not\cong Y}. \]Practice 3
No. Compactness distinguishes them.
\[ [0,1]\text{ is compact}, \qquad (0,1)\text{ is not}. \]Practice 4
No. The first space is connected, while the second is disconnected.
Practice 5
Euler characteristic:
\[ \chi(S^2)=2, \qquad \chi(T^2)=0. \]Genus also works:
\[ g(S^2)=0, \qquad g(T^2)=1. \]Practice 6
Orientability distinguishes them. The torus is orientable, while the Klein bottle is nonorientable.
Practice 7
Boundary-component count distinguishes them:
\[ b(D^2)=1, \qquad b(A)=2. \]Practice 8
No. For example,
\[ (0,1)\cong\mathbb{R}, \]but only the first is bounded.
Practice 9
No. Reidemeister moves can change the crossing count of a particular diagram.
The minimal crossing number is the knot invariant.
Practice 10
A complete invariant agrees exactly when the objects are equivalent within the chosen class:
\[ \boxed{ I(X)=I(Y) \iff X\cong Y }. \]What to Remember
- Topological invariants are preserved under homeomorphism.
- Different invariant values prove that spaces are not homeomorphic.
- Matching one or several invariants may still fail to prove homeomorphism.
- Connectedness, compactness, path connectedness, and component count are topological invariants.
- Length, angle, area, volume, and boundedness are generally not topological invariants.
- Orientability, genus, boundary count, and Euler characteristic are important invariants for surfaces.
- Fundamental groups and homology provide stronger algebraic invariants.
- Knot theory uses invariants such as minimal crossing number, knot group, and knot polynomials.
- A complete invariant classifies objects completely within a specified category.
- The best strategy is usually to begin with the simplest invariant that can settle the question.
The core logic is:
\[ \boxed{ X\cong Y \Longrightarrow \text{all topological invariants agree} }. \]Therefore,
\[ \boxed{ \text{one differing invariant} \Longrightarrow X\not\cong Y }. \]Where We Go Next
Topological invariants unify many of the ideas developed throughout topology. Connectedness, compactness, genus, orientability, Euler characteristic, boundary data, and knot invariants all serve the same purpose: they preserve structural information while ignoring irrelevant geometric distortion.
From here, the natural next steps include fundamental groups, homotopy, homology, cohomology, covering spaces, manifolds, and algebraic topology. These subjects develop increasingly powerful invariants for distinguishing spaces that look topologically similar at first glance.