Hubbry Logo
Order topologyOrder topologyMain
Open search
Order topology
Community hub
Order topology
logo
7 pages, 0 posts
0 subscribers
Be the first to start a discussion here.
Be the first to start a discussion here.
Order topology
Order topology
from Wikipedia

In mathematics, an order topology is a specific topology that can be defined on any totally ordered set. It is a natural generalization of the topology of the real numbers to arbitrary totally ordered sets.

If X is a totally ordered set, the order topology on X is generated by the subbase of "open rays"

for all a, b in X. Provided X has at least two elements, this is equivalent to saying that the open intervals

together with the above rays form a base for the order topology. The open sets in X are the sets that are a union of (possibly infinitely many) such open intervals and rays.

A topological space X is called orderable or linearly orderable[1] if there exists a total order on its elements such that the order topology induced by that order and the given topology on X coincide. The order topology makes X into a completely normal Hausdorff space.

The standard topologies on R, Q, Z, and N are the order topologies.

Induced order topology

[edit]

If Y is a subset of X, X a totally ordered set, then Y inherits a total order from X. The set Y therefore has an order topology, the induced order topology. As a subset of X, Y also has a subspace topology. The subspace topology is always at least as fine as the induced order topology, but they are not in general the same.

For example, consider the subset Y = {−1} ∪ {1/n}nN of the rationals. Under the subspace topology, the singleton set {−1} is open in Y, but under the induced order topology, any open set containing −1 must contain all but finitely many members of the space.

Example of a subspace of a linearly ordered space whose topology is not an order topology

[edit]

Though the subspace topology of Y = {−1} ∪ {1/n}nN in the section above is shown not to be generated by the induced order on Y, it is nonetheless an order topology on Y; indeed, in the subspace topology every point is isolated (i.e., singleton {y} is open in Y for every y in Y), so the subspace topology is the discrete topology on Y (the topology in which every subset of Y is open), and the discrete topology on any set is an order topology. To define a total order on Y that generates the discrete topology on Y, simply modify the induced order on Y by defining −1 to be the greatest element of Y and otherwise keeping the same order for the other points, so that in this new order (call it say <1) we have 1/n <1 −1 for all n ∈ N. Then, in the order topology on Y generated by <1, every point of Y is isolated in Y.

We wish to define here a subset Z of a linearly ordered topological space X such that no total order on Z generates the subspace topology on Z, so that the subspace topology will not be an order topology even though it is the subspace topology of a space whose topology is an order topology.

Let in the real line. The same argument as before shows that the subspace topology on Z is not equal to the induced order topology on Z, but one can show that the subspace topology on Z cannot be equal to any order topology on Z.

An argument follows. Suppose by way of contradiction that there is some strict total order < on Z such that the order topology generated by < is equal to the subspace topology on Z (note that we are not assuming that < is the induced order on Z, but rather an arbitrarily given total order on Z that generates the subspace topology).

Let M = Z \ {−1} = (0,1), then M is connected, so M is dense on itself and has no gaps, in regards to <. If −1 is not the smallest or the largest element of Z, then and separate M, a contradiction. Assume without loss of generality that −1 is the smallest element of Z. Since {−1} is open in Z, there is some point p in M such that the interval (−1,p) is empty, so p is the minimum of M. Then M \ {p} = (0,p) ∪ (p,1) is not connected with respect to the subspace topology inherited from R. On the other hand, the subspace topology of M \ {p} inherited from the order topology of Z coincides with the order topology of M \ {p} induced by <, which is connected since there are no gaps in M \ {p} and it is dense. This is a contradiction.

Left and right order topologies

[edit]

Several variants of the order topology can be given:

  • The right order topology[2] on X is the topology having as a base all intervals of the form , together with the set X.
  • The left order topology on X is the topology having as a base all intervals of the form , together with the set X.

These topologies naturally arise when working with semicontinuous functions, in that a real-valued function on a topological space is lower semicontinuous if and only if it is continuous when the reals are equipped with the right order.[3] The (natural) compact open topology on the resulting set of continuous functions is sometimes referred to as the semicontinuous topology[4].

Additionally, these topologies can be used to give counterexamples in general topology. For example, the left or right order topology on a bounded set provides an example of a compact space that is not Hausdorff.

The left order topology is the standard topology used for many set-theoretic purposes on a Boolean algebra.[clarification needed]

Ordinal space

[edit]

For any ordinal number λ one can consider the spaces of ordinal numbers

together with the natural order topology. These spaces are called ordinal spaces. (Note that in the usual set-theoretic construction of ordinal numbers we have λ = [0, λ) and λ + 1 = [0, λ]). Obviously, these spaces are mostly of interest when λ is an infinite ordinal; for finite ordinals, the order topology is simply the discrete topology.

When λ = ω (the first infinite ordinal), the space [0,ω) is just N with the usual (still discrete) topology, while [0,ω] is the one-point compactification of N.

Of particular interest is the case when λ = ω1, the set of all countable ordinals, and the first uncountable ordinal. The element ω1 is a limit point of the subset [0,ω1) even though no sequence of elements in [0,ω1) has the element ω1 as its limit. In particular, [0,ω1] is not first-countable. The subspace [0,ω1) is first-countable however, since the only point in [0,ω1] without a countable local base is ω1. Some further properties include

Topology and ordinals

[edit]

Ordinals as topological spaces

[edit]

Any ordinal number can be viewed as a topological space by endowing it with the order topology (indeed, ordinals are well-ordered, so in particular totally ordered). Unless otherwise specified, this is the usual topology given to ordinals. Moreover, if we are willing to accept a proper class as a topological space, then we may similarly view the class of all ordinals as a topological space with the order topology.

The set of limit points of an ordinal α is precisely the set of limit ordinals less than α. Successor ordinals (and zero) less than α are isolated points in α. In particular, ω (meaning the set [0, ω)) is a discrete topological space, as is each finite ordinal, but no ordinal greater than ω is discrete. The ordinal α is compact as a topological space if and only if α is either a successor ordinal or zero.

The closed sets of an ordinal α are those that contain a limit ordinal less than α whenever every open interval that contains the limit ordinal intersects the set.

Any ordinal is, of course, an open subset of any larger ordinal. We can also define the topology on the ordinals in the following inductive way: 0 is the empty topological space, α+1 is obtained by taking the one-point compactification of α, and for δ a limit ordinal, δ is equipped with the inductive limit topology. Note that if α is a successor ordinal, then α is compact, in which case its one-point compactification α+1 is the disjoint union of α and a point.

As topological spaces, all the ordinals are Hausdorff and even normal. They are also totally disconnected (connected components are points), scattered (every non-empty subspace has an isolated point; in this case, just take the smallest element), zero-dimensional (the topology has a clopen basis: here, write an open interval (β,γ) as the union of the clopen intervals (β,γ'+1) = [β+1,γ'] for γ'<γ). However, they are not extremally disconnected in general (there are open sets, for example the even numbers from ω, whose closure is not open).

The topological spaces ω1 and its successor ω1+1 are frequently used as textbook examples of uncountable topological spaces. For example, in the topological space ω1+1, the element ω1 is in the closure of the subset ω1 even though no (countably-long) sequence of elements in ω1 has the element ω1 as its limit: an element in ω1 is a countable set; for any sequence of such sets, the union of these sets is the union of countably many countable sets, so still countable; this union is an upper bound of the elements of the sequence, and therefore of the limit of the sequence, if it has one.

The space ω1 is first-countable but not second-countable, and ω1+1 has neither of these two properties, despite being compact. It is also worthy of note that any continuous function from ω1 to R (the real line) is eventually constant: so the Stone–Čech compactification of ω1 is ω1+1, just as its one-point compactification (in sharp contrast to ω, whose Stone–Čech compactification is much larger than ω).

Ordinal-indexed sequences

[edit]

If α is a limit ordinal and X is a set, an α-indexed sequence of elements of X merely means a function from α to X. This concept, a transfinite sequence or ordinal-indexed sequence, is a generalization of the concept of a sequence. An ordinary sequence corresponds to the case α = ω.

If X is a topological space, we say that an α-indexed sequence of elements of X converges to a limit x when it converges as a net, in other words, when given any neighborhood U of x there is an ordinal β < α such that xι is in U for all ιβ.

Ordinal-indexed sequences are more powerful than ordinary (ω-indexed) sequences to determine limits in topology: for example, ω1 is a limit point of ω1+1 (because it is a limit ordinal), and, indeed, it is the limit of the ω1-indexed sequence which maps any ordinal less than ω1 to itself: however, it is not the limit of any ordinary (ω-indexed) sequence in ω1, since any such limit is less than or equal to the union of its elements, which is a countable union of countable sets, hence itself countable.

However, ordinal-indexed sequences are not powerful enough to replace nets (or filters) in general: for example, on the Tychonoff plank (the product space ), the corner point is a limit point (it is in the closure) of the open subset , but it is not the limit of an ordinal-indexed sequence of ordinals in .

See also

[edit]

Notes

[edit]

References

[edit]
Revisions and contributorsEdit on WikipediaRead on Wikipedia
from Grokipedia
In mathematics, the order topology is a topology defined on a totally ordered set XX (equipped with a strict order relation <<) by generating open sets from intervals determined by the order. The basis for the order topology consists of all open intervals (a,b)={xXa<x<b}(a, b) = \{x \in X \mid a < x < b\} for a,bXa, b \in X with a<ba < b, along with half-open intervals of the form [a0,b)={xXa0x<b}[a_0, b) = \{x \in X \mid a_0 \leq x < b\} if XX has a least element a0a_0, and (a,b0]={xXa<xb0}(a, b_0] = \{x \in X \mid a < x \leq b_0\} if XX has a greatest element b0b_0. This collection forms a basis because it covers XX and satisfies the intersection condition for basis elements. Notable examples include the standard topology on the real numbers R\mathbb{R}, which coincides with the order topology induced by the usual less-than order. Another example is the order topology on R×R\mathbb{R} \times \mathbb{R} induced by the dictionary (lexicographic) order. Additionally, the order topology on the positive integers Z+\mathbb{Z}^+ yields the discrete topology, as singletons are open. Key properties of the order topology include the openness of unbounded rays such as (a,+)={xXa<x}(a, +\infty) = \{x \in X \mid a < x\} and (,b)={xXx<b}(-\infty, b) = \{x \in X \mid x < b\}, which form a subbasis for the topology. This structure allows the order topology to capture continuity and compactness in ordered spaces, making it foundational for studying ordered topological spaces like ordinals and the long line.

Fundamentals

Definition

A totally ordered set, also known as a linearly ordered set, consists of a set XX equipped with a binary relation \leq that is reflexive, antisymmetric, transitive, and total: for any x,yXx, y \in X, either xyx \leq y or yxy \leq x. The strict order << is defined by x<yx < y if xyx \leq y and xyx \neq y. Common intervals in such a set include the open interval (a,b)={xXa<x<b}(a, b) = \{x \in X \mid a < x < b\}, the half-open intervals [a,b)={xXax<b}[a, b) = \{x \in X \mid a \leq x < b\} and (a,b]={xXa<xb}(a, b] = \{x \in X \mid a < x \leq b\}, and the closed interval [a,b]={xXaxb}[a, b] = \{x \in X \mid a \leq x \leq b\}, where a,bXa, b \in X. The order topology τ\tau on a totally ordered set (X,)(X, \leq) is the topology generated by taking all open intervals as a basis. More precisely, τ\tau is the unique topology having as a basis the collection of all open intervals (a,b)(a, b) together with, if XX has a least element a0a_0, the intervals [a0,b)[a_0, b), and if XX has a greatest element b0b_0, the intervals (a,b0](a, b_0]. Formally, for a totally ordered set (X,)(X, \leq), a subset UXU \subseteq X is open in the order topology if and only if it is a union of open intervals of the form (a,b)={xXa<x<b}(a, b) = \{x \in X \mid a < x < b\} for a,bXa, b \in X with a<ba < b, along with rays (,b)={xXx<b}(-\infty, b) = \{x \in X \mid x < b\} and (a,+)={xXx>a}(a, +\infty) = \{x \in X \mid x > a\} for a,bXa, b \in X. This construction ensures that the basis elements cover XX and satisfy the intersection property required for a basis. This concept was introduced by Felix Hausdorff in his 1914 monograph Grundzüge der Mengenlehre, where he developed foundational ideas in set theory and topology based on ordered structures.

Basis Elements

In the order topology on a totally ordered set XX, a standard basis B\mathcal{B} consists of all open intervals of the form (a,b)={xXa<x<b}(a, b) = \{ x \in X \mid a < x < b \}, where a,bX{,+}a, b \in X \cup \{-\infty, +\infty\} with a<ba < b. This collection includes unbounded rays such as (,b)={xXx<b}(-\infty, b) = \{ x \in X \mid x < b \} for bXb \in X and (a,+)={xXx>a}(a, +\infty) = \{ x \in X \mid x > a \} for aXa \in X, accommodating sets without minimal or maximal elements. These basis elements generate all open sets as arbitrary unions, so every open set UXU \subseteq X can be expressed as U=iI(ai,bi)U = \bigcup_{i \in I} (a_i, b_i) for some index set II and suitable ai,bia_i, b_i. To verify that B\mathcal{B} forms a basis for the order topology, two conditions must hold: first, for every xXx \in X, there exists some BBB \in \mathcal{B} containing xx; second, if B1,B2BB_1, B_2 \in \mathcal{B} and xB1B2x \in B_1 \cap B_2, then there exists B3BB_3 \in \mathcal{B} such that xB3B1B2x \in B_3 \subseteq B_1 \cap B_2. The first condition is satisfied because, for any xXx \in X, one can choose successors and predecessors in the order (or rays if at endpoints) to form an interval containing xx, such as (a,b)(a, b) where aa is the greatest element less than xx (if it exists) and bb the least greater than xx. For the second, suppose x(a1,b1)(a2,b2)x \in (a_1, b_1) \cap (a_2, b_2); then set a=max(a1,a2)a = \max(a_1, a_2) and b=min(b1,b2)b = \min(b_1, b_2), yielding (a,b)B(a, b) \in \mathcal{B} with x(a,b)(a1,b1)(a2,b2)x \in (a, b) \subseteq (a_1, b_1) \cap (a_2, b_2), since the order ensures a<x<ba < x < b. This confirms B\mathcal{B} satisfies the basis axioms. A subbasis S\mathcal{S} for the order topology is given by the collection of all open rays {(,b)bX}{(a,+)aX}\{ (-\infty, b) \mid b \in X \} \cup \{ (a, +\infty) \mid a \in X \}. The basis B\mathcal{B} arises as all finite intersections of elements from S\mathcal{S}, since the intersection of (a,+)(a, +\infty) and (,b)(-\infty, b) yields (a,b)(a, b) when a<ba < b, and single rays remain when no intersection is taken. Arbitrary unions of these finite intersections then produce the full topology. For example, consider the totally ordered set of rational numbers Q\mathbb{Q} with the standard order. The basis elements are open intervals (p,q)(p, q) where p,qQp, q \in \mathbb{Q} and p<qp < q; these are open in the order topology on Q\mathbb{Q}, which coincides with the subspace topology induced from the Euclidean topology on R\mathbb{R}, though the order construction highlights the role of rays in generating unbounded opens like (p,+)Q(p, +\infty) \cap \mathbb{Q}.

Properties

Separation Axioms

The order topology on a totally ordered set satisfies the T0 separation axiom, as distinct points can be separated by open sets: if x<yx < y, then x(,y)x \in (-\infty, y) but y(,y)y \notin (-\infty, y), and symmetrically for the other ray. It is also T1, since singletons are closed; the complement of {x}\{x\} is the union of the open rays (,x)(x,)(-\infty, x) \cup (x, \infty), which is open. Equivalently, {x}=y>x(y,)z<x(,z)\{x\} = \bigcap_{y > x} (y, \infty) \cap \bigcap_{z < x} (-\infty, z), showing singletons are GδG_\delta sets, though T1 follows directly from the closedness. This holds for totally ordered sets, which are antisymmetric by definition. The order topology is Hausdorff (T2). For distinct points x<yx < y, disjoint open neighborhoods exist: if there is cc with x<c<yx < c < y, take (,c)(-\infty, c) containing xx and (c,)(c, \infty) containing yy; if no such cc exists (i.e., x,yx, y are consecutive), take (,y)(-\infty, y) containing xx and (x,)(x, \infty) containing yy, which are disjoint since nothing lies between xx and yy. Thus, the property holds regardless of density; for example, the order topology on the integers Z\mathbb{Z} is the discrete topology, which is Hausdorff. Order topologies are regular (T3). Given a point xx and a closed set CC with xCx \notin C, the basis elements allow separation: since the space is T1 and the order structure ensures closed sets are "order-closed," there exist disjoint open sets UU containing xx and VV containing CC. Specifically, basis elements like open intervals or rays can be chosen to avoid overlap, leveraging the linear order to isolate xx from the "gaps" in CC. This holds generally, though in dense orders like R\mathbb{R}, the standard basis simplifies the separation. As a consequence of being regular and Hausdorff, order topologies are Tychonoff (completely regular, T3.5). Points and closed sets can be separated by continuous functions, often via order-preserving embeddings into products of intervals. For instance, the coordinate functions in the order can be used to construct such separations. Order topologies are also normal (T4) and even completely normal (hereditarily normal). Disjoint closed sets can be separated by disjoint open sets, with the property holding in subspaces. A non-metrizable example is the long line, obtained as the lexicographic order on [0,ω1)×[0,1)[0, \omega_1) \times [0,1) excluding the endpoint, equipped with the order topology; it satisfies all these separation axioms but fails metrizability due to lacking a countable basis.

Compactness and Connectedness

In the order topology induced by a total order on a set XX, the space is connected if and only if XX is densely ordered (for any a<ba < b in XX, there exists cXc \in X with a<c<ba < c < b) and Dedekind-complete (every nonempty subset of XX that is bounded above has a least upper bound in XX). Such an ordered set is called a , and connectedness follows from the fact that any disconnection would correspond to a without a supremum or a gap in the ordering. For instance, the real numbers R\mathbb{R} under the standard ordering form a linear continuum, so the order topology on R\mathbb{R} (which coincides with the standard topology) is connected. Order topologies on linear continua are locally connected, as the basis consists of open intervals and rays, each of which is itself a connected subspace (a linear continuum without endpoints or one-sided). This local connectedness arises because small neighborhoods around any point are homeomorphic to open intervals in R\mathbb{R}, which are connected. The order topology on a totally ordered set XX is compact if and only if XX has a minimum and maximum element and satisfies the least upper bound property (Dedekind-completeness). Under these conditions, every open cover has a finite subcover, as the completeness ensures that suprema exist to "cap" chains of basis elements in any cover. For example, the closed interval [0,1][0,1] with the order topology is compact, and this topology agrees with the subspace topology from R\mathbb{R}, satisfying the Heine-Borel theorem. In contrast, R\mathbb{R} itself is not compact in the order topology, as the open cover {(n,n+2)nZ}\{(n, n+2) \mid n \in \mathbb{Z}\} has no finite subcover, mirroring the failure of Heine-Borel for unbounded sets. When XX is well-ordered, the order topology is compact if and only if XX has a maximum element (i.e., its order type is a successor ordinal). Examples include finite ordinals and [0,ω][0, \omega] (order type ω+1\omega + 1), as well as larger successor ordinals like ω1+1\omega_1 + 1. In this case, the space is sequentially compact, and since it is Hausdorff, compactness follows; larger well-orderings like [0,ω1][0, \omega_1] (order type ω1+1\omega_1 + 1? Wait, no: [0, ω₁] has type ω₁ +1, compact; but [0, ω₁) type ω₁, not compact due to uncountable open covers without finite subcovers. The Lindelöf property—that every open cover has a countable subcover—holds in the order topology on spaces of countable ordinals (e.g., [0,ω][0, \omega] or finite ordinals) but fails for uncountable ordinals like ω1\omega_1, the first uncountable ordinal. For ω1\omega_1, the open cover {(0,α)α<ω1}\{(0, \alpha) \mid \alpha < \omega_1\} requires uncountably many sets to cover the space, as any countable subcollection covers only up to a countable supremum.

One-Sided Variants

Left Ray Topology

The left ray topology on a totally ordered set (X, <) is the topology generated by the subbasis consisting of all left rays (−∞, b) = {x ∈ X | x < b} for b ∈ X. The open sets are arbitrary unions of these left rays, which, owing to their nested structure, are either the empty set, the whole space X, or individual left rays (−∞, b) for some b ∈ X. This construction yields a topology coarser than the standard order topology, as the latter's subbasis includes both left and right rays, producing a larger collection of open sets; consequently, the identity map from (X with the order topology) to (X with the left ray topology) is continuous, but the reverse map is not, since bounded open intervals (a, b) from the order topology are not open in the left ray topology. The finite intersections of subbasis elements are left rays with the minimum upper bound, so the collection of all left rays {(−∞, b) | b ∈ X} ∪ {X} forms a basis for the topology. On the real line ℝ, this topology has open sets ∅, ℝ, and (−∞, b) for b ∈ ℝ, providing a model for one-sided (left) continuity in ordered spaces, where functions that are continuous with respect to left limits can be analyzed using these unbounded open sets. The order topology on X is generated by taking both left rays and right rays as a subbasis. The lower limit topology (Sorgenfrey line) on ℝ is a different refinement with basis [a, b) for a < b ∈ ℝ. The left ray topology satisfies the T0 separation axiom but is not Hausdorff. For distinct points x < y in X, the (−∞, y) contains x but not y, satisfying T0, but every open set containing y is of the form (−∞, b) with b > y and thus contains x, preventing separation of x and y by disjoint opens. On the integers ℤ with the usual order, this manifests as points not being separable from points to their left, with any open set containing n containing all m < n. The topology is also not regular, as there exist closed sets and points outside them that cannot be separated by disjoint open sets; for example, in ℝ, the closed set [0, +∞) and the point -1 cannot be separated in this manner.

Right Ray Topology

The right ray topology on a totally ordered set (X,<)(X, <) is the topology generated by taking as a subbasis the collection of all right rays {(a,+)aX}\{(a, +\infty) \mid a \in X \}, where (a,+)={xXx>a}(a, +\infty) = \{x \in X \mid x > a \}, along with the and XX itself; the open sets are arbitrary unions of these right rays. Owing to their nested structure, the open sets are either \emptyset, XX, or individual right rays (a,+)(a, +\infty) for some aXa \in X. This topology is coarser than the , and the identity map from (X(X with order topology)) to (X(X with right ray topology)) is continuous, but not conversely, as bounded intervals are not open in the right ray topology. The finite intersections of subbasis elements are right rays with the maximum lower bound, so the collection of all right rays {(a,+)aX}{X}\{(a, +\infty) \mid a \in X\} \cup \{X\} forms a basis for the . On the real line R\mathbb{R}, this has open sets \emptyset, R\mathbb{R}, and (a,+)(a, +\infty) for aRa \in \mathbb{R}, providing a model for one-sided (right) continuity in ordered spaces. The upper limit on R\mathbb{R} is a distinct refinement generated by the basis {(a,b]a,bR,a<b}\{(a, b] \mid a, b \in \mathbb{R}, a < b\}. The right ray topology satisfies the T0T_0 separation axiom but is not T1T_1 or Hausdorff. For distinct points x<yx < y in XX, the open set (x,+)(x, +\infty) contains yy but not xx, satisfying T0T_0, but every open set containing xx is of the form (a,+)(a, +\infty) with a<xa < x and thus contains yy, preventing separation by disjoint opens. Any two non-empty open sets intersect, as they are cofinal to ++\infty. On the integers Z\mathbb{Z} with the usual order, points cannot be separated from points to their right, with any open set containing nn containing all m>nm > n. The topology is also not regular; for example, in R\mathbb{R}, the closed set (,0](-\infty, 0] and the point $1$ cannot be separated by disjoint open sets. There is a natural duality with the left ray topology: the order-reversing map f:XXf: X \to X, or on R\mathbb{R} the xxx \mapsto -x, sends right rays (a,+)(a, +\infty) to left rays (,a)(-\infty, -a), establishing a between the two topologies. On R\mathbb{R}, the right ray topology emphasizes right-unbounded opens and is coarser than the standard order topology, as standard open intervals cannot be expressed as unions of right rays without overflowing to ++\infty. In contrast to the left ray topology, which emphasizes left-unbounded opens, the right ray topology focuses on right-unbounded ones, and their combination as subbasis generates the full order topology.

Subspace Considerations

Induced Topology on Subsets

In a linearly ordered set (X,)(X, \leq) equipped with the order topology τX\tau_X, consider a subset YXY \subseteq X ordered by the induced order Y\leq_Y. The subspace topology τY\tau_Y on YY is defined as τY={UYUτX}\tau_Y = \{ U \cap Y \mid U \in \tau_X \}. The subspace topology τY\tau_Y coincides with the order topology generated by Y\leq_Y for certain subsets YY. A key case is when YY is convex in XX (i.e., an interval, so that if a,bYa, b \in Y and a<x<ba < x < b with xXx \in X, then xYx \in Y); in this situation, the open sets in the order topology on YY—such as intervals (a,b)Y={yYa<y<b}(a, b)_Y = \{ y \in Y \mid a < y < b \} with a,bYa, b \in Y—align precisely with intersections of open intervals from τX\tau_X with YY. Convexity is sufficient but not necessary for coincidence; for example, the subspace topology on the disconnected set (0,1)(2,3)R(0,1) \cup (2,3) \subset \mathbb{R} also matches its order topology, as both are unions of open intervals in each component. Another case is when YY is order-dense in XX (i.e., between any two elements of XX, there lies an element of YY). A representative example is the rational numbers Q\mathbb{Q} as a subset of the real numbers R\mathbb{R}, where R\mathbb{R} carries its standard order topology (equivalent to the Euclidean topology). Although Q\mathbb{Q} is not convex in R\mathbb{R}, it is order-dense, and the subspace topology on Q\mathbb{Q} coincides with the order topology on Q\mathbb{Q}. Basis elements in the subspace topology, such as (c,d)Q(c, d) \cap \mathbb{Q} for c,dRc, d \in \mathbb{R}, can be expressed as unions of open intervals (p,q)Q(p, q) \cap \mathbb{Q} with p,qQp, q \in \mathbb{Q}, which form the basis for the order topology on Q\mathbb{Q}. To see this more generally, note that the basis for the order topology on YY consists of sets of the form (a,b)Y(a, b) \cap Y with a,bYa, b \in Y. Each such set is open in τY\tau_Y because (a,b)Y=((a,b)X)Y(a, b) \cap Y = ((a, b) \cap X) \cap Y and (a,b)X(a, b) \cap X is open in τX\tau_X. Conversely, for convex or order-dense YY, every basis element of τY\tau_Y—namely, (c,d)Y(c, d) \cap Y with c,dXc, d \in X—contains, for each y(c,d)Yy \in (c, d) \cap Y, an order basis element (a,b)Y(c,d)Y(a', b') \cap Y \subseteq (c, d) \cap Y with a,bYa', b' \in Y, ensuring the topologies match.

Non-Order Subspace Examples

A classic example illustrating the discrepancy between the subspace topology and the induced order topology arises in the real line R\mathbb{R} equipped with its standard order topology. Consider the subset Y=[0,1){2}RY = [0, 1) \cup \{2\} \subseteq \mathbb{R}. In the subspace topology on YY, the singleton {2}\{2\} is open, as it equals the intersection of YY with the open interval (1.5,2.5)(1.5, 2.5) in R\mathbb{R}. However, in the order topology induced on YY by its natural linear order, {2}\{2\} is not open. Any basis element containing 2 takes the form (a,2](a, 2] where aYa \in Y and a<2a < 2; since all such aa lie in [0,1)[0, 1), this interval includes points from [0,1)[0, 1) arbitrarily close to 1, making it impossible to isolate 2 without including elements from the left accumulation. This failure occurs because the gap between 1 and 2 in YY is recognized by the ambient topology of R\mathbb{R}, allowing open sets to separate the isolated point 2 from the accumulating interval [0,1)[0, 1), whereas the induced order on YY treats the space as a single linear order without "filling" the gap, forcing neighborhoods of 2 to extend leftward into the accumulation at 1. In the context of ordinal spaces, an analogous example is found in the ordered set ω+2=ω{ω,ω+1}\omega + 2 = \omega \cup \{\omega, \omega + 1\} with its order topology, considering the subspace Y=ω{ω+1}Y = \omega \cup \{\omega + 1\} (omitting ω\omega). In the subspace topology, {ω+1}\{\omega + 1\} is open, as it equals (ω,ω+1]Y(\omega, \omega + 1] \cap Y, leveraging the ambient basis element that isolates the successor point after the limit ordinal ω\omega. In contrast, the induced order topology on YY renders {ω+1}\{\omega + 1\} not open, since basis neighborhoods of ω+1\omega + 1 are rays (α,ω+1](\alpha, \omega + 1] for α<ω+1\alpha < \omega + 1 in YY, which always include cofinal tails of ω\omega (the natural numbers). The general reason for these mismatches lies in subspaces where accumulation points or gaps do not align with the order intervals of the induced linear order; the ambient topology can exploit gaps or limits external to the subset to create finer open sets, such as isolated points, that the intrinsic order topology cannot replicate without including extraneous accumulation. Consequently, the order topology is not always hereditary: while the subspace topology inherits the ambient structure faithfully, it may fail to coincide with (and can be strictly finer than) the order topology induced on the subset.

Ordinal Applications

Ordinal Spaces

The order topology on an ordinal α is defined on the set [0, α] of all ordinals less than or equal to α, using as a basis the collection of all open intervals (β, γ) = {δ ∈ [0, α] | β < δ < γ} for ordinals β < γ ≤ α, along with the open rays (−∞, γ) = {δ ∈ [0, α] | δ < γ} and (β, +∞) = {δ ∈ [0, α] | β < δ} as a subbasis. This topology makes [0, α] a linearly ordered topological space (LOTS) where the order is preserved. Ordinal spaces are Hausdorff, as the order topology on any linear order separates distinct points with disjoint open intervals. They are first-countable at successor ordinals, where each such point σ = ρ + 1 is isolated via the clopen singleton {σ} = (ρ, σ + 1) ∩ [0, α] (adjusting for the endpoint if necessary), and thus admits a countable local basis consisting of that singleton. At limit ordinals with countable cofinality, points are also first-countable, as a countable cofinal sequence below the point yields a countable local basis of neighborhoods. However, at limit ordinals with uncountable cofinality, such as the endpoint α = ω₁ in [0, ω₁], the space is not first-countable, since any local basis at ω₁ requires uncountably many distinct open rays (β, ω₁] for β < ω₁ to separate the cofinal structure. The space [0, α] is compact for every ordinal α. For instance, [0, ω₁] is compact. Ordinal spaces are scattered, meaning no nonempty subspace is dense-in-itself (or equivalently, every nonempty subspace has an isolated point), a consequence of the well-ordering: in any nonempty subset, the least element is isolated relative to that subset via an open ray from below. This property holds for all ordinals, including uncountable ones like ω₁. The long line provides an example of a non-separable ordinal-like space, constructed as the set ω₁ × [0, 1) equipped with the lexicographic order (comparing first by ordinal component, then by real), and thus the order topology; it is sequentially compact and hereditarily normal but fails separability due to the uncountable "length" without a countable dense subset. Ordinal spaces are zero-dimensional, as every basic open interval (β, γ) is clopen: its complement is the union of the clopen initial segment [0, β] (open as (−∞, β + 1) and closed by well-ordering) and the terminal segment [γ, α] (symmetrically clopen). For countable α, the space is second-countable (hence metrizable and separable), with a countable basis of rational-like intervals in the countable order. Uncountable ordinal spaces exhibit uncountable cellularity: for example, in [0, ω₁), the family of singletons {σ} for all successor ordinals σ < ω₁ forms an uncountable collection of pairwise disjoint nonempty open sets.

Sequences in Ordinals

In the order topology on an ordinal α\alpha, the convergence of a sequence (βn)nN(\beta_n)_{n \in \mathbb{N}} to a point γα\gamma \in \alpha is determined by the basis neighborhoods of γ\gamma. For a successor ordinal γ=δ+1\gamma = \delta + 1, the singleton {γ}\{\gamma\} is open, so convergence requires the sequence to be eventually constant at γ\gamma. For a limit ordinal γ\gamma, the basis neighborhoods are of the form (δ,γ+1)(\delta, \gamma + 1) for δ<γ\delta < \gamma, so (βn)(\beta_n) converges to γ\gamma if and only if for every δ<γ\delta < \gamma, there exists NN such that βn>δ\beta_n > \delta for all n>Nn > N. This means the tail of the sequence is unbounded below γ\gamma in the order, and the limit superior of the sequence equals γ\gamma. In countable ordinals like ω\omega, sequences behave similarly to those in the natural numbers with the discrete topology, where convergence is eventual constancy except for increasing sequences approaching limit points such as ω\omega itself (in ω+1\omega + 1). However, in uncountable ordinals such as the first uncountable ordinal ω1\omega_1, countable sequences cannot converge to ω1\omega_1 because any countable collection of ordinals less than ω1\omega_1 has a countable supremum, which is itself a countable ordinal strictly less than ω1\omega_1. Thus, in the space [0,ω1][0, \omega_1], no non-constant sequence from below converges to ω1\omega_1, making {ω1}\{\omega_1\} sequentially open but not open, illustrating that the topology is not sequential. Ordinal spaces also exhibit interesting sequential compactness properties. The space [0,ω1)[0, \omega_1) with the order topology is sequentially compact: every sequence has a convergent subsequence. This follows from the well-ordering, which allows extraction of a non-decreasing subsequence whose supremum lies in [0,ω1)[0, \omega_1) and serves as the limit. In contrast, [0,ω1][0, \omega_1] is sequentially compact, as it is a compact Hausdorff space. More generally, the space [0, γ] with the order topology is always sequentially compact, since it is compact.

References

Add your contribution
Related Hubs
User Avatar
No comments yet.