Romyar SharifiLECTURE NOTES
READING EDITIONPDF

LECTURE NOTES / Chapter 4

Point-Set Topology

Romyar Sharifi

Chapter 4 Countability and separation axioms

Book contents

Chapter 4
Countability and separation axioms

4.1. Countability axioms

Definition 4.1.1.

A topological space X is said to be first-countable if every point of X has a countable basis of open neighborhoods.

Example 4.1.2.

If X is a metrizable space, then X is first-countable. For instance, if d is a metric on X, then x X has the countable basis {B(x, 1 n)n 1} of open neighborhoods.

Proposition 4.1.3.

Let A be a subset of a first-countable topological space X. Then A¯ is the set of limits of convergent sequences in A.

Proof.

If there exists a sequence (an)n1 in A with limit x X, then every neighborhood of x contains all an for n sufficiently large, so x A¯ by definition. Conversely, if x A¯ and {Unn 1} is a basis of open neighborhoods of x, then we may pick an AUn for each n 1, and the sequence (an)n1 has limit x.

Proposition 4.1.4.

Let X be a first-countable space. Let x X. If f : X Y is a function such that for every convergent sequence (xn)n1 in X with limit x, the sequence (f(xn))n1 converges to f(x), then f is continuous at x.

Proof.

Let V be an open set in Y, let x f1(V ), and let {Unn 1} be a countable basis of open neighborhoods of x. If f1(V ) does not contain an open neighborhood of x, then it cannot contain any Un. Assume this, and for each n 1, choose xn Un such that xnf1(V ). Then the sequence (xn)n1 converges to x, and by assumption, (f(xn))n1 converges to f(x). Therefore, there exists N 1 such that f(xn) V for all n N, so xn f1(V ) for such n, contradicting our assumption.

Definition 4.1.5.

A topological space X is second-countable if X has a countable base for its topology.

Example 4.1.6.

The space n is second-countable, since it has the base {B(x, 1 k)x n,k 1}.

Example 4.1.7.

The space X = i=1 is second-countable in the product topology but not in the uniform topology. In the product topology, we can take the countable base consisting of products

i=1N(a i,bi)×i=N+1

with ai,bi and ai < bi for 1 i N for some N 0.

In the uniform topology, we have the uncountable discrete subspace

A = {(an)n1an {0,1}}.

every two points being distance 1 apart in the uniform metric. For any base of X, the open balls of radius 12 about the points of A would each contain an element of the base, which is therefore uncountable. In particular, metrizable spaces are not necessarily second-countable.

Definition 4.1.8.

A topological space X is Lindelöf if every open cover of X has a countable subcover.

Proposition 4.1.9.

Second-countable spaces are Lindelöf.

Proof.

Let 𝒰 be an open cover of X, and let B = {Bnn 1} be a countable base of X. Let S be the set of positive integers n such that Bn is contained in an element of U. For each n S, let Un 𝒰 with Bn Un. We claim that {Unn S} covers X.

If x X, then x U for some U 𝒰. Since 𝔹n is a base for the topology on X, there exists n 1 such that x Bn U. By definition of S, we then have n S, and so x Un. Thus, we have proven the claim.

Recall that a subset A of a topological space is X is dense if A¯ = X.

Definition 4.1.10.

A topological space X is separable if it has a countable dense subset.

Proposition 4.1.11.

Second-countable spaces are separable.

Proof.

Let B = {Unn 1} be a countable base of X. For each n 1, choose xn Un, and set A = {xnn 1}. Then for any x X and open neighborhood V of x, there exists n 1 such that x Un V, so in particular xn V. Thus, x A¯, so A is dense in X.

For metric spaces, the converses to Propositions 4.1.9 and 4.1.11 hold.

Theorem 4.1.12.

For a metric space X, the following are equivalent:

i.

X is second-countable,

ii.

X is Lindelöf, and

iii.

X is separable.

Proof.

By Propositions 4.1.9 and 4.1.11, it suffices to show that Lindeöf spaces are second-countable and separable spaces are second-countable. Let X be a metric space.

Suppose first that X is Lindelöf. For positive integers n 1, let Bn be a countable subcover of the open cover of X by balls of radius 1n. Let B = n=1Bn, which is also countable. We claim that B is a base for the metric topology on X. Let x X and 𝜖 > 0, and consider the open ball B(x,𝜖). Choose n with 2n < 𝜖, and let y X be such that x B(y, 1 n) Bn. Then B(y, 1 n) B(x,𝜖) by the triangle inequality, hence the claim.

Next, suppose that X is separable, and let S X be a countable dense subset. We claim that the countable collection

B = {B(y,1n)y S,n 1}

of open balls is a base for the metric topology on X. Let x X and 𝜖 > 0. As S is dense, B(x, 𝜖 3) contains a point y S, and if we take n 1 such that 𝜖3 1 n 2𝜖 3, then x B(y, 1 n) B(x,𝜖) by the triangle inequality, hence the claim.

4.2. Separation axioms

In this section, we consider both weaker and stronger conditions in terms of the separation of points in a topological space than our much-used Hausdorff condition. We begin by defining the word “separated” in the sense of topology.

Definition 4.2.1.

We say that two subsets A and B of a topological space are separated if A¯B = and AB¯ = .

Definition 4.2.2.

A separation of X is a set of two disjoint open subspaces of X with union X.

Remark 4.2.3.

A space X has a separation if and only if it is disconnected, in which case it is the topological disjoint union of the two spaces A and B in the separation. Moreover, these A and B are clearly separated. Similarly, if X is the union of two separated subsets, then they are both closed, hence open, and therefore constitute a separation of X.

Here then are several weaker conditions that being Hausdorff.

Definition 4.2.4.

a.

A space X is said to be T 0 if every two distinct points of X have distinct sets of open neighborhoods.

b.

A space X is said to be T 1 if points of X are closed.

c.

A space X is said to be T 2 if it is Hausdorff.

Example 4.2.5.

Spaces that are T 0 but not T 1 are ubiquitous in modern algebraic geometry. For instance, what is known as the spectrum Spec of the ring of integers of is a space consisting of one point (0) and another point (p) for each prime number p. It has a base consisting of (0) and the complements of finite sets {(p1),,(pn)} with each pi prime. The closure of {(0)} is Spec. Unfortunately, the explanation for why algebraic geometers consider this space lies beyond the scope of this course.

Definition 4.2.6.

A space in which every two points with distinct sets of open neighborhoods are separated is called symmetric.

Lemma 4.2.7.

A space is T 1 if and only if it is symmetric and T 0, i.e., every pair of distinct points is separated.

Proof.

Suppose X is T 1. If x,y X are distinct, then X {y} and X {x} are distinct open neighborhoods of x and y, respectively, so X is T 0. Moreover, {x}¯ = {x} for all x X, so for yx, the sets {x} and {y} are separated. That is, X is symmetric.

Conversely, if X is T 0 and x X, then for any y X, there exists an open neighborhood U containing x and not y. If X is also symmetric, then since x and y have distinct sets of neighborhoods, {x} and {y} are separated. In particular, y{x}¯, which is to say that {x} is closed.

Definition 4.2.8.

a.

A topological space X is regular if for every point x X and closed subset A of X with xA, there exist disjoint open sets U and V with x U and A V.

b.

A topological space X is normal if for every two closed, disjoint subsets A and B of X, there exist disjoint open subsets U and V of X such that A U and B V.

Since in a topological space, points need not be closed (i.e., the space need not be T 1), it is not clear that either regular or normal should implies Hausdorff. In general, they do not, and normal does not imply regular. So, we make the following definitions.

Definition 4.2.9.

a.

A space X is said to be T 3 if it is regular and T 0.

b.

A space X is said to be T 4 if it is normal and T 1.

The following tells us that a T 3-space is Hausdorff and a T 4-space is regular.

Lemma 4.2.10.

If a space is T i for some 1 i 4, then it is T i1.

Proof.

By Lemma 4.2.7, a T 1-space is T 0, and by Lemma 1.5.9, a T 2-space is T 1.

Suppose X is T 3. Take x,y X with xy. By the T 0-axiom, there is without loss of generality an open neighborhood W of x than does not contain y. Then Wc and x are disjoint, hence by the regularity condition, are contained in disjoint open sets U and V, respectively. Then x U and y V, so X is Hausdorff.

Similarly, suppose that X is T 4. Let x X, and let A X be closed with xA. Since X is T 1, the set {x} is closed. Therefore, by normality there exist open disjoint sets U containing {x} and V containing A, as desired.

The reader can check the following, which is most interesting for and stops at T 3-spaces.

Lemma 4.2.11.

For 0 i 3, every subspace of a T i-space is T i.

Example 4.2.12.

The space X = with the topology consisting of sets of the form U C, where U is open in in the Euclidean topology and C is a countable subset of U, is Hausdorff but not regular. It is Hausdorff since its topology is finer than the Euclidean topology on . It is not regular, as the set is closed in its topology, and the open sets containing in are the complements U in of countable sets of irrational numbers. An open neighborhood V of an irrational number contains all but countably many points in an open interval around it, and so does U, but such an interval is uncountable, so U V cannot be empty.

Lemma 4.2.13.

A space X is regular (resp., normal) if and only if every open set of X containing a point x (resp., closed set A) contains the closure of an open set containing x (resp., A).

Proof.

Let X be regular (resp., normal). Let A be a singleton (resp., closed) subset of X, and let W be an open subset containing A. Then B = Wc is disjoint from A, so by regularity (resp., normality) of X, there exist disjoint open sets U containing A and V containing B. Then A U Vc W. Since Vc is closed, W contains U¯, so U is the desired open set.

Conversely, suppose that for any singleton (resp., closed) subset A of X, every open set of X containing A contains the closure of an open set containing A. Let B be a closed set disjoint from A, and let U be an open subset of Bc containing A and such that U¯ Bc. Take V = U¯c so that B V, and note that U and V are disjoint. Thus, X is regular (resp., normal).

We have seen that a direct product of Hausdorff spaces is Hausdorff. The analogous statement holds for regular Hausdorff spaces (i.e., T 3-spaces).

Proposition 4.2.14.

A product of T 3-spaces is T 3 in the product topology.

Proof.

Let X = iIXi, where {Xii I} is a collection of T 3-topological spaces. Then X is Hausdorff, and in particular, its points are closed. Let x = (xi)iI X, and let U be an open neighborhood of x. Then U contains a basic open neighborhood iIUi, where each Ui is open in Xi and all but finitely many Ui equal Xi. For every i, we may choose an open neighborhood Vi of xi with Vi¯ Ui by Lemma 4.2.13. If Ui = Xi, we take Vi = Xi as well. Then iIVi contains x and has closure iIVi¯ iIUi, as desired.

The following example shows that even a finite direct product of T 4-spaces need not be T 4.

Example 4.2.15.

Consider X = with the lower-limit topology generated by the base of open sets [a,b) with a < b. Note that these sets are closed as well, since (,a)[b,) has complement [a,b). Then X is normal Hausdorff. To see normality, take disjoint closed sets A and B in X. For each a A, we pick xa > a with [a,xa) in the complement of B and let U be the union of these half-open intervals. Similarly, for each b B, we pick yb > b with [b,yb) in the complement of A and let V be the union of these intervals. Take a A and b B, and suppose without loss of generality that a < b. Then xa < b since b[a,xa), so the intersection [a,xa)[b,yb) is empty, and therefore U and V are disjoint.

On the other hand, the product X2 = X ×X is regular Hausdoff by Proposition 4.2.14, but it is not normal. The subspace D = {(x,x)x } of X ×X has the discrete topology and is closed in X2, and if we take the subset A = {(x,x)x }, then A and DA are closed subsets of X2 that are not contained in disjoint open neighborhoods of X2. We omit the nontrivial proofs of these facts.

Theorem 4.2.16.

Regular, second-countable spaces are normal.

Proof.

Let X be regular and second-countable. Let B be a countable base for the topology on X. Let A and B be disjoint closed subsets of X. By regularity, for each x X we can find an open neighborhood U of x with U¯ disjoint from B, and contained in U we can find some neighborhood in B of x. Together, these basis elements give a countable covering 𝒰 = {Unn 1} of A with Un¯B = for each n 1. Similarly, we can find a countable covering 𝒱 = {Vnn 1} of B with Vn¯A = . For each n 1, consider the open sets

Un = U n i=1nV i¯c and V n = V n i=1nU i¯c.

If a A, then a Un for some n 1, and aVi¯ for all i, so a Un. The open sets U = n1Unand V = n1Vn contain A and B, respectively, and they are disjoint, since if u Un for some n 1, then uVi for i n by definition of Un and uVifor i > n by definition of Vn.

Theorem 4.2.17.

Metrizable spaces are normal.

Proof.

Let A and B be disjoint, closed subsets of a metrizable space X, and let d be a metric on X. For each a A, there exists 𝜖a > 0 with B(a,𝜖a)B = , and similarly, for each b B, there exists δb > 0 with B(b,δb)A = . Set

U = aAB (a, 𝜖a 2 ) and V = bBB (b, δb 2 ).

Then U and V are open containing A and B, respectively. If x U V, then there exist a A and b B such that d(a,x) < 𝜖a 2 and d(x,b) < δb 2 . By the triangle inequality, we then have d(a,b) < 𝜖a 2 + δb 2 max(𝜖a,δb). If d(a,b) < 𝜖a, then b B(a,𝜖a), a contradiction, and if d(a,b) < δb, then a B(b,δb), contradiction. Thus U and V are disjoint.

Theorem 4.2.18.

Compact Hausdorff spaces are normal.

Proof.

Let X be a compact Hausdorff space, and let A and B be disjoint closed subsets of X, which are necessarily compact. Lemma 3.2.7 implies that compact Hausdorff spaces are regular. That is, for each b B, we may choose disjoint open sets Ub containing A and Vb containing b. Then the sets Vb cover B, hence have a finite subcover, say by Vb1,,Vbn B. Then U = i=1nUb i and V = i=1nVb i are disjoint open sets containing A and B, respectively.

Using the one-point compactification, we may use Theorem 4.2.13 to give a quick proof of the analogous result for locally compact Hausdorff spaces.

Corollary 4.2.19.

Locally compact Hausdorff spaces are regular.

Proof.

Let X be locally compact and Hausdorff. Let Y be its one-point compactification, which is compact Hausdorff and therefore normal. As Y is T 4, it is also T 3, and X is T 3 as a subspace of Y.

4.3. Urysohn’s lemma

Theorem 4.3.1 (Urysohn’s lemma).

A topological space X is normal if and only if for every pair of disjoint nonempty closed sets A and B of X, there exists a continuous function f : X [0,1] such that f(A) = {0} and f(B) = {1}.

Proof.

Let X be a normal space. Set U1 = Bc, and by normality of X, pick an open set U0 containing A with U0¯ U1. We will construct open sets Uq for q (0,1) such that if q,r [0,1] with q < r, then Uq¯ Ur.

Fix any bijection ϕ : 0 [0,1] with ϕ(0) = 0 and ϕ(1) = 1. For each n 2, let qn be the largest value of ϕ(m) with m < n that is less than ϕ(n), and let rn be the smallest value of ϕ(m) with m < n that is greater than ϕ(n). Suppose by induction that we have constructed Uϕ(m) for all m < n. By Lemma 4.2.13, we may choose an open set Uϕ(n) containing Uqn¯ and such that Uϕ(n)¯ is contained in Urn. For negative r , set Ur = , and for r greater than 1, set Ur = X. In this way, we have constructed open sets Uq for all rational numbers q such that Uq¯ Ur if q < r are rational numbers.

We now define f by setting

f(x) = inf{q x Uq}

for x X. If x A, then x U0, so f(x) = 0. If x B, then xU1 but x Ur for all rationals r > 1, so f(x) = 1. If x Uq¯, then x Ur for all rationals r > q, so f(x) q. Given a nonempty open interval (a,b) in and x in its inverse image, we then have q,r with a < q < f(x) < r < b. Since f(x) < r, we have x Ur, and since f(x) > q, we have xUq¯. That is, we have x UrUq¯, which is an open neighborhood with image under f contained in [q,r] (a,b). Thus, f is continuous.

As for the converse, note that if A and B are disjoint closed sets in X and f : X [0,1] is a continuous function with f(A) = {0} and f(B) = {1}, then U = f1([0,12)) and V = f1((12,1]) are disjoint open sets containing A and B, respectively.

Definition 4.3.2.

A space X is completely regular if for every closed set A in X and x Ac, there exists a continuous function f : X [0,1] with f(A) = {0} and f(x) = 1. A space X is said to be Tychonoff, or T 31 2 , if it is completely regular and T 1.

Remark 4.3.3.

A normal Hausdorff space is Tychonoff, and a completely regular space is regular.

We leave it to the reader to check that the following hold, the condition on subspaces between the motivation for the word “completely”.

Proposition 4.3.4.

Subspaces and products of completely regular spaces are completely regular.

Regularity, together with second-countability, implies complete regularity (since it implies normality). In fact, we have the following.

Lemma 4.3.5.

Let X be a regular, second-countable space. Then there exists a countable collection {fnn 1} of continuous functions fn: X [0,1] such that for each pair (A,x) of a closed subset A of X and x Ac, there exists n such that fn(A) = {0} and fn(x) = 1.

Proof.

Since X is second-countable and regular, it is normal. Let {Unn 1} be a countable base of open sets in X. If Um¯ Un, then by Urysohn’s lemma, there exists a continuous function fm,n: X [0,1] with fm,n(Um¯) = {1} and f(X Un) = {0}.

The set Ac contains some neighborhood Un of x, which contains the closure of some neighborhood Um of x by the regularity of X. Then fm,n(x) = 1 and fm,n(A) = {0}. Since the collection of functions fm,n is countable, we have the lemma.

Theorem 4.3.6 (Urysohn metrization theorem).

Second-countable regular Hausdorff spaces are metrizable and are exactly the spaces that are homeomorphic to subspaces of the product space [0,1]J, where J is countably infinite.

Proof.

We may take J = {nn 1}. Recall that J is metrizable by Proposition 2.2.16, so to show metrizability, it suffices to embed our second-countable regular space X in [0,1]J. If we can do this for a given X, then [0,1]J is regular Hausdorff and second-countable, and then so is X.

Let (fn)n1 be a sequence of functions fn: X [0,1] as in Lemma 4.3.5 and use them to define a function f = (fn)n1: X [0,1]J, which is continuous as each fn is continuous. Since X is Hausdorff, for any x,y X with xy, the set {x}c is an open neighborhood of y, so we can for each xy in X find an n 1 such that fn(x) = 0 and fn(y) = 1. Thus, f is injective.

It remains to show that f is an open map to its image. Let U be an open set in X, let a = (an)n1 f(U), and let x U with f(x) = a. Choose an m 1 such that fm(x) = 1 and fm(y) = 0 for all yU. Let V be the open set πm1((0,1])f(X) in the image of f, where πm is the mth projection map from πm: [0,1]J [0,1]. Then πm(a) = fm(x) = 1, so a V. For b V, we have b = f(y) for some y with πm(b) = fm(y) > 0, so y U and therefore b f(U). That is, we have a f(U) V. Thus, f(U) is open.

Another application of Urysohn’s lemma is found in the following theorem.

Theorem 4.3.7 (Tietze extension theorem).

Let A be a closed subspace of a normal topological space X. Any continuous map of A into an interval in can be extended to a continuous map of X into the same interval in .

Proof.

It is sufficient to prove this result for maps into [1,1], [1,1) and (1,1), since all intervals in are homeomorphic to one of these. We next show how the result reduces to maps to [1,1]. Given any g: X [1,1], let B = g1(1) and C = g1(1). By Urysohn’s lemma, there exist continuous functions ϕ : X [0,1] and ϕ: X [0,1] with ϕ(B) {0}, ϕ(C) = ϕ(C) {0}, and ϕ(A) = ϕ(A) {1}. If g(A) [1,1), then set g = 𝜙𝑔, and note that g|A = g|A. Moreover, it satisfies g(x) = 0 if x C and g(x) < 1 if xC, so g(X) [1,1). Similarly, if g(A) (1,1), then set g = ϕg. Then g|A = g|A, and g(X) (1,1).

We next prove a weaker result. Suppose that F : A [1,1] is continuous. Let

B = F1([1,1 3]) and C = F1([1 3,1]).

By Urysohn’s lemma, there exists a continuous function

G: X [13,13]

with G(B) = {13} and G(C) = {13}. Then

|G(a)F (a)|2 3

for all a A. That is, for a BC, we have |G(a)F (a)| 11 3 = 2 3, and for xBC, we have |G(a)F (a)|1 3 (13) = 2 3.

Now let f : A [1,1] be continuous. Suppose by induction that for some n 0 we have found continuous functions

gi: X [1 3 (2 3 )n1, 1 3 (2 3 )n1]

for 1 i n with

|f(a)i=1ng i(a)| (2 3 )n

for all a A. Then the argument of the previous paragraph applied to F = (32)n(f i=1ngi) yields the next function gn+1 = (23)nG in the induction. Since n=11 3(2 3)n1 = 1, the infinite series

g =i=1g i: X [1,1]

is well-defined, and by definition, it restricts to f on A. Moreover, g is continuous as the partial sums i=1ngi converge uniformly to g on X: that is, |i=n+1gn(x)| (23)n for all x X.

Definition 4.3.8.

The support of a continuous function f from a topological space X to is the closure

suppf = {x Xf(x)0}¯.

Definition 4.3.9.

Let 𝒰 = {Ui1 i n} be an open cover of X for some n 1. A finite collection {ϕi: X [0,1]1 i n} of continuous functions is a partition of unity on X subordinate to, or dominated by, 𝒰 if suppϕi Ui for each i and i=1nϕi(x) = 1 for all x X.

Lemma 4.3.10.

Let 𝒰 = {Ui1 i n} be a finite open cover of a normal space X for some n 1. Then there exists an open cover 𝒱 = {Vi1 i n} of X with Vi¯ Ui for all 1 i n.

Proof.

It suffices to show the existence of V1 open with V1¯ U1 and such that {U2,,Un,V1} covers X, since then we can repeat with this new cover, replacing U2 and so forth. Let A = X i=2nUi. By normality of X, there exists an open set V1 containing A with A V1 V1¯ U1. Then the collection {V1,U2,,Un} covers X.

Proposition 4.3.11.

Let 𝒰 = {Ui1 i n} be a finite open cover of a normal space X. Then there exists a partition of unity on X subordinate 𝒰.

Proof.

By Lemma 4.3.10, we can find an open cover 𝒱 = {Vi1 i n} with Vi¯ Ui for 1 i n and an open cover 𝒲 = {Wi1 i n} with Wi¯ Vi for each 1 i n. By Urysohn’s lemma, there exist functions ψi: X [0,1] such that ψi(Wi¯) {1} and ψi(Vic) {0}. Since ψi1({0}) Vi, we have suppψi Vi¯ Ui. For each x X, we have x Wi for some i, and therefore jψj(x) ψi(x) = 1 > 0. We may then define

ϕi(x) = ψi(x) j=1nψj(x)

for x X. Then the ϕi form the desired partition of unity.

Definition 4.3.12.

A second-countable Hausdorff space M is said to be a manifold if there exists n 0 such that every point of M has an open neighborhood homeomorphic to n. We then say that M is n-dimensional, or an n-manifold.

Remark 4.3.13.

To say that a point in M has an open neighborhood homeomorphic to n is to say that it has an open neighborhood homeomorphic to an open subset of n (in particular, as open balls in n are homeomorphic to n). Equivalently, there is a local homeomorphism n M with image containing the point.

Example 4.3.14.

Open sets in n and the sphere Sn are n-manifolds. The torus is an example of a 2-manifold, or surface.

We have the following application of Urysohn’s lemma and partitions of unity to manifolds.

Theorem 4.3.15.

Any compact manifold can be embedded in N for some N 1.

Proof.

Since X is a manifold, it has an open cover 𝒰 by open sets that may be embedded in n for some fixed n. Since X is compact, there exist U1,,Um 𝒰 that together cover X. Let fi: Ui n be an open embedding. By Theorem 4.2.18, the space X is normal, so there exists a partition of unity {ϕi1 i m} subordinate to {Ui1 i m}. For each i, set Ai = suppϕi and define gi: X n by gi(x) = ϕi(x)fi(x) for x Ui and gi(x) = 0 for x Aic. Note that gi is well-defined as ϕi(x) = 0 for x Ac.

We set

F = ((ϕ1,g1),,(ϕm,gm)): X (1+n)m

which is continuous as a product of continuous functions. We claim that F is an embedding, which will finish the proof with N = m(1+n). Since X is compact, it is enough to show it is injective. If x,y X and F (x) = F (y), then ϕi(x) = ϕi(y) and gi(x) = gi(y) for all 1 i n. Let i be such that ϕi(x) > 0 so that x,y Ui. Then gi(x) = ϕi(x)fi(x) and gi(y) = ϕi(x)fi(y), so fi(x) = fi(y). As fi: Ui n is injective, we have x = y.

Find in the notes