Bourbaki

§ 2. Reflections

In this paragraph, K denotes a commutative field, assumed not to be a characteristic 2 from no. 2 onwards. We denote by V a vector space over K.

1. Pseudo-reflections

Definition 1

An endomorphism s of the vector space V is said to be a pseudo-reflection if 1−s1 - s is of rank 1.

Let s be a pseudo-reflection in V, and let D be the image of 1−s1 - s. By definition, D is of dimension 1; thus, given a≠0a \neq 0 in D, there exists a non-zero linear form a∗a^* on V such that x−s(x)=⟨x,a∗⟩.ax - s(x) = \langle x, a^* \rangle . a for all x∈Vx \in V.

Conversely, given a≠0a \neq 0 in V and a linear form a∗≠0a^* \neq 0 on V, the formula

sa,a∗(x)=x−⟨x,a∗⟩.a(x∈V) s_{a, a^*}(x) = x - \langle x, a^* \rangle . a \quad (x \in V)

defines a pseudo-reflection sa,a∗s_{a, a^*}; the image of 1−sa,a∗1 - s_{a, a^*} is generated by a and the kernel of 1−sa,a∗1 - s_{a, a^*} is the hyperplane of V consisting of those x such that ⟨x,a∗⟩=0\langle x, a^* \rangle = 0. If V∗V^* is the dual of V, it is immediate that the transpose sa∗,as_{a^*, a} of sa,a∗s_{a, a^*} is the pseudo-reflection of V∗V^* given by the formula

sa∗,a(x∗)=x∗−⟨x∗,a⟩.a∗(x∗∈V∗). s_{a^*, a}(x^*) = x^* - \langle x^*, a \rangle . a^* \quad (x^* \in V^*).

If a is a non-zero vector, a pseudo-reflection with vector a is any pseudo-reflection s such that a belongs to the image of 1−s1 - s. The hyperplane of a pseudo-reflection s is the kernel of 1−s1 - s, the set of vectors x such that s(x)=xs(x) = x.

Proposition 1

Let G be a group and ρ\rho an irreducible linear representation of G on a vector space V; assume that there exists an element g of G such that ρ(g)\rho(g) is a pseudo-reflection.

(i) Every endomorphism of V commuting with ρ(G)\rho(G) is a homothety, and ρ\rho is absolutely irreducible.

(ii) Assume that V is finite dimensional. Let B be a non-zero bilinear form on V invariant under ρ(G)\rho(G). Then B is non-degenerate, either symmetric or skew-symmetric, and every bilinear form on V invariant under ρ(G)\rho(G) is proportional to B.

Let u be an endomorphism of V commuting with ρ(G)\rho(G). Let g be an element of G such that ρ(g)\rho(g) is a pseudo-reflection and let D be the image of 1−ρ(g)1 - \rho(g). Since D is of dimension 1 and u(D)⊂Du(D) \subset D, there exists α\alpha in K such that u−α.1u - \alpha . 1 vanishes on D; the kernel N of u−α.1u - \alpha . 1 is then a vector subspace of V invariant under ρ(G)\rho(G) and is non-zero as it contains D; since ρ\rho is irreducible, N=VN = V and u=α.1u = \alpha . 1. The second part of (i) follows from the first by Algebra, Chap. VIII, § 13, no. 4, Cor. of Prop. 5.

Let N (resp. N′N') be the subspace of V consisting of those x such that B(x,y)=0B(x, y) = 0 (resp. B(y,x)=0B(y, x) = 0) for all y in V; since B is invariant under ρ(G)\rho(G), the subspaces N and N′N' of V are stable under ρ(G)\rho(G) and distinct from V since B≠0B \neq 0. Since ρ\rho is irreducible, N=N′=0N = N' = 0 and B is non-degenerate.

Since V is finite dimensional, every bilinear form on V is given by the formula

B′(x,y)=B(u(x),y) B'(x, y) = B(u(x), y)
for some endomorphism u of V. If B′B' is invariant under ρ(G)\rho(G), the endomorphism u commutes with ρ(G)\rho(G). Indeed, let x,yx, y be in V and let g be in G; since B and B′B' are invariant under ρ(G)\rho(G), we have
B(u(ρ(g)(x)),y)=B′(ρ(g)(x),y)=B′(x,ρ(g−1)(y))=B(u(x),ρ(g−1)(y))=B(ρ(g)(u(x)),y), \begin{align*} B(u(\rho(g)(x)), y) &= B'(\rho(g)(x), y) = B'(x, \rho(g^{-1})(y)) \\ &= B(u(x), \rho(g^{-1})(y)) = B(\rho(g)(u(x)), y), \end{align*}
hence u(ρ(g)(x))=ρ(g)(u(x))u(\rho(g)(x)) = \rho(g)(u(x)) since B is non-degenerate. By (i), there exists α\alpha in K with u=α.1u = \alpha.1, so B′=α.BB' = \alpha.B.

In particular, we can apply this to the bilinear form B′(x,y)=B(y,x)B'(x, y) = B(y, x); then B(y,x)=α.B(x,y)=α2.B(y,x)B(y, x) = \alpha.B(x, y) = \alpha^2.B(y, x) for all x,yx, y in V, and since B is nonzero we have α2=1\alpha^2 = 1, hence α=1\alpha = 1 or α=−1\alpha = -1. Thus, B is either symmetric or skew-symmetric.

2. Reflections

Recall that from now on, unless stated otherwise, the field K is assumed to be of characteristic different from 2. A reflection in V is a pseudo-reflection s such that s2=1s^2 = 1; if s is a reflection, we denote by Vs+V_s^+ the kernel of s−1s - 1 and by Vs−V_s^- that of s+1s + 1.

Proposition 2

Let s be an endomorphism of V.

(i) If s is a reflection, V is the direct sum of the hyperplane Vs+V_s^+ and the line Vs−V_s^-.

(ii) Conversely, assume that V is the direct sum of a hyperplane H and a line D such that s(x)=xs(x) = x and s(y)=−ys(y) = -y for x∈Hx \in H and y∈Dy \in D. Then s is a reflection and H=Vs+H = V_s^+, D=Vs−D = V_s^-. Finally, D is the image of 1−s1 - s.

(i) If s is a reflection, Vs+V_s^+ is a hyperplane. If x belongs to Vs+∩Vs−V_s^+ \cap V_s^-, then x=s(x)=−xx = s(x) = -x, so x=0x = 0 since K is of characteristic ≠2\neq 2. Finally, for xx in V, the vector x′=s(x)+xx' = s(x) + x (resp. x′′=s(x)−xx'' = s(x) - x) belongs to Vs+V_s^+ (resp. Vs−V_s^-) since s2=1s^2 = 1, and 2x=x′−x′′2x = x' - x''. Thus V is the direct sum of Vs+V_s^+ and Vs−V_s^-, and Vs−V_s^- is necessarily of dimension 1 since Vs+V_s^+ is a hyperplane.

(ii) Under the stated hypotheses, every element of V can be written uniquely in the form v=x+yv = x + y with x∈Hx \in H and y∈Dy \in D and we have s(v)=x−ys(v) = x - y; assertion (ii) follows immediately from this.

Corollary

If V is finite dimensional, every reflection is of determinant −1.

Let s be a reflection in V. Prop. 2, (i) shows that there exists a basis (e1,…,en)(e_1, \ldots, e_n) of V such that s(e1)=e1,…,s(en−1)=en−1s(e_1) = e_1, \ldots, s(e_{n-1}) = e_{n-1} and s(en)=−ens(e_n) = -e_n, hence det⁡(s)=−1\det(s) = -1.

Proposition 3

Let s be a reflection in V.

(i) A subspace V' of V is stable under s if and only if Vs−⊂V′V_s^- \subset V' or V′⊂Vs+V' \subset V_s^+.

(ii) An endomorphism u of V commutes with s if and only if Vs+V_s^+ and Vs−V_s^- are stable under u.

(i) If V′⊂Vs+V' \subset V_s^+, then s(x)=xs(x) = x for all xx in V′V', so s(V′)⊂V′s(V') \subset V'. Assume that Vs−⊂V′V_s^- \subset V'; then, for any xx in V′V', s(x)−x∈Vs−⊂V′s(x) - x \in V_s^- \subset V', hence s(x)∈V′s(x) \in V'; thus s(V′)⊂V′s(V') \subset V'. Conversely, assume that s(V′)⊂V′s(V') \subset V'; if V′⊄Vs+V' \not\subset V_s^+, there exists xx in V′V' with s(x)≠xs(x) \neq x; the non-zero vector a=s(x)−xa = s(x) - x belongs to the line Vs−V_s^-, and hence generates this space; since a∈V′a \in V', we have Vs−⊂V′V_s^- \subset V'.

(ii) Assume first that uu commutes with ss. If xx is a vector such that s(x)=ε.xs(x) = \varepsilon.x (where ε=±1\varepsilon = \pm 1), then s(u(x))=u(s(x))=ε.u(x)s(u(x)) = u(s(x)) = \varepsilon.u(x), so Vs+V_s^+ and Vs−V_s^- are stable under uu. Conversely, assume that Vs+V_s^+ and Vs−V_s^- are stable under uu; it is clear that us−suus - su vanishes on Vs+V_s^+ and on Vs−V_s^-, and since V is the direct sum of Vs+V_s^+ and Vs−V_s^-, we have us−su=0us - su = 0.

Corollary

Two distinct reflections s and u commute if and only if Vs−⊂Vu+V_s^- \subset V_u^+ and Vu−⊂Vs+V_u^- \subset V_s^+.

If Vs−⊂Vu+V_s^- \subset V_u^+ and Vu−⊂Vs+V_u^- \subset V_s^+, Prop. 3, (i) shows that Vu+V_u^+ and Vu−V_u^- are stable under ss, hence su=ussu = us by Prop. 3, (ii).

Conversely, if su=ussu = us, the subspace Vs−V_s^- is stable under uu by (ii); by (i), there are two possible cases:

a) We have Vu−⊂Vs−V_u^- \subset V_s^-: since they are both of dimension 1, these spaces are therefore equal, hence Vs−⊄Vu+V_s^- \not\subset V_u^+; since Vu+V_u^+ is stable under ss, we have Vu+⊂Vs+V_u^+ \subset V_s^+ and so these two hyperplanes are equal. But then s=us = u, contrary to our assumptions.

b) We have Vs−⊂Vu+V_s^- \subset V_u^+: the image of 1−s1 - s is thus contained in the kernel of 1−u1 - u, so (1−u).(1−s)=0(1 - u).(1 - s) = 0. Since ss and uu commute, we have (1−s).(1−u)=0(1 - s).(1 - u) = 0, in other words Vu−⊂Vs+V_u^- \subset V_s^+.

Remark

Let a≠0a \neq 0 be in V and let a∗a^* be a non-zero linear form on V: it follows from formula (1) that

sa,a∗2(x)=x+(⟨a,a∗⟩−2)⟨x,a∗⟩.a s_{a,a^*}^2(x) = x + (\langle a, a^* \rangle - 2)\langle x, a^* \rangle.a

and hence that sa,a∗s_{a,a^*} is a reflection if and only if ⟨a,a∗⟩=2\langle a, a^* \rangle = 2. In this case, we have sa,a∗(a)=−as_{a,a^*}(a) = -a.

3. Orthogonal reflections

Assume that V is finite dimensional. Let B be a non-degenerate bilinear form on V. By Algebra, Chap. IX, § 6, no. 3, Prop. 4, B is invariant under a reflection s in V if and only if the subspaces Vs+V_s^+ and Vs−V_s^- of V are orthogonal with respect to B; they are then non-isotropic. Moreover, for any non-isotropic hyperplane H in V, there is a unique reflection s that preserves B and induces the identity on H; this is the symmetry with respect to H, cf. Algebra, Chap. IX, §6, no. 3. If aa is a non-zero vector orthogonal to H, we have B(a,a)≠0B(a, a) \neq 0 and the reflection s is given by the formula

s(x)=x−2B(x,a)B(a,a)⋅afor any x∈V, s(x) = x - 2 \frac{B(x, a)}{B(a, a)} \cdot a \quad \text{for any } x \in V,

by Algebra, Chap. IX, § 6, no. 4, formula (6). The reflection s is also called the orthogonal reflection with respect to H.

Proposition 4

Assume that V is finite dimensional. Let B be a non-degenerate symmetric bilinear form on V, X a subspace of V and X0X^0 the orthogonal complement of X with respect to B; finally, let s be the orthogonal reflection with respect to a non-isotropic hyperplane H of V. The following conditions are equivalent:

(i) X is stable under s. (ii) X0X^0 is stable under s. (iii) H contains X or X0X^0.

We have Vs+=HV_s^+ = H, and by what we have said, Vs−V_s^- is the orthogonal complement H0H^0 of H with respect to B. By Prop. 3, X is stable under s if and only if X⊂HX \subset H or H0⊂XH^0 \subset X; but the relation H0⊂XH^0 \subset X is equivalent to X0⊂HX^0 \subset H by Algebra, Chap. IX, § 1, no. 6, Cor. 1 to Prop. 4. This proves the equivalence of (i) and (iii); that of (ii) and (iii) follows by interchanging the roles of X and X0X^0, since (X0)0=X(X^0)^0 = X.

4. Orthogonal reflections in a euclidean affine space

We retain the notation of the preceding number, and let E be an affine space of which V is the space of translations. Giving the form B on V provides E with the structure of a euclidean space (Algebra, Chap. IX, § 6, no. 6).

Let H be a non-isotropic hyperplane of E. The symmetry with respect to H (Algebra, Chap. IX, § 6, no. 6) is also called the orthogonal reflection with respect to H; we often denote it by sHs_H. We have sH2=1s_H^2 = 1 and sHs_H is the unique displacement (loc. cit., Def. 3) of E, distinct from the identity and leaving fixed the elements of H. The automorphism of V associated to sHs_H is the orthogonal reflection with respect to the direction of H (which is a non-isotropic hyperplane of V).

Every xx in EE can be written uniquely in the form x=h+vx = h + v, with h∈Hh \in H and v∈Vv \in V orthogonal to HH; we have

sH(h+v)=h−v. s_H(h + v) = h - v.

Proposition 5

Let HH and H′H' be two parallel, non-isotropic hyperplanes of EE. There exists a unique vector v∈Vv \in V orthogonal to HH and such that H′=H+vH' = H + v. The displacement sH′sHs_{H'} s_H is the translation by the vector 2v2v.

The existence and uniqueness of vv are immediate. The automorphism of VV associated to sH′sHs_{H'} s_H is the identity; thus sH′sHs_{H'} s_H is a translation. On the other hand, let a∈H′a \in H'; then a−v∈Ha - v \in H and

sH′sH(a−v)=sH′(a−v)=a+v=(a−v)+2v, s_{H'} s_H(a - v) = s_{H'}(a - v) = a + v = (a - v) + 2v,

showing that sH′sHs_{H'} s_H is the translation by the vector 2v2v.

Corollary

Let HH and H′H' be two distinct, parallel, non-isotropic hyperplanes. If KK is of characteristic zero (resp. p>0p > 0, with p≠2p \neq 2), the group of displacements of EE generated by sHs_H and sH′s_{H'} is an infinite dihedral group (resp. a dihedral group of order 2p2p).

Indeed, by Prop. 2 of Chap. IV, § 1, no. 2, it suffices to show that sH′sHs_{H'} s_H is of infinite order (resp. order 2p2p), which is clear.

Remark

We retain the notation of Prop. 5 and assume in addition that K=RK = \mathbf{R}. Put s=sHs = s_H and s′=sH′s' = s_{H'}. Let HnH_n be the hyperplane H+n.vH + n.v and let CnC_n be the set of points of EE of the form a+ξ.va + \xi.v with a∈Ha \in H and n<ξ<n+1n < \xi < n + 1. The CnC_n are connected open sets forming a partition of E−⋃nHnE - \bigcup_n H_n. They are therefore the chambers defined by the system H=(Hn)n∈Z\mathfrak{H} = (H_n)_{n \in \mathbf{Z}} in EE. The translation (s′s)n(s's)^n transforms the chamber C=C0C = C_0 into the chamber C2nC_{2n}, and since s(C0)=C−1s(C_0) = C_{-1}, we have (s′s)ns(C)=C2n−1(s's)^n s(C) = C_{2n-1}. It follows that the dihedral group W generated by s and s' permutes the chambers Cn simply-transitively*. Moreover, as we shall now show, if the chambers C and w(C) are on opposite sides of H (for w∈Ww \in W), we have l(sw) = l(w) - 1 (the lengths being taken with respect to S={s,s′}S = \{s, s'\} (Chap. IV, § 1, no. 1)). Indeed, we then have w(C)=Cnw(C) = C_n for some n<0n < 0. If n=−2kn = -2k, then w=(ss′)kw = (ss')^k and sw=(s′s)k−1s′sw = (s's)^{k-1}s', so l(w)=2kl(w) = 2k and l(sw)=2k−1l(sw) = 2k - 1 (Chap. IV, § 1, no. 2, Remark). If n=−2k−1n = -2k - 1, then w=(ss′)ksw = (ss')^k s and sw=(s′s)ksw = (s's)^k, so l(w)=2k+1l(w) = 2k + 1 and l(sw)=2kl(sw) = 2k.

5. Complements on plane rotations

In this no., VV denotes a real vector space of dimension 2, provided with a scalar product (that is, a non-degenerate, positive, symmetric bilinear form) and an orientation. The measures of angles will be taken with respect to the base 2π2\pi; the principal measure of an angle between half-lines (resp. lines) is thus a real number θ\theta such that 0≤θ<2π0 \leq \theta < 2\pi (resp. 0≤θ<π0 \leq \theta < \pi) (General Topology, Chap. VIII, § 2, no. 3 and no. 6). By abuse of language, for any real number θ\theta, we shall use θ\theta to denote an angle whose measure is θ\theta and denote by ρθ\rho_\theta the rotation with angle θ\theta (Algebra, Chap. IX, § 10, no. 3).

Proposition 6

Let s be the orthogonal reflection with respect to a line D of V. If Δ\Delta and Δ′\Delta' are two half-lines starting at the origin (resp. two lines passing through the origin) of V, we have

(s(Δ),s(Δ′)^)≡−(Δ^,Δ′^)(mod2π)(resp. (modπ))). (s(\Delta), \widehat{s(\Delta')}) \equiv - (\widehat{\Delta}, \widehat{\Delta'}) \pmod{2\pi} \quad \text{(resp. } \pmod{\pi})).

Let uu be a rotation transforming Δ\Delta into Δ′\Delta'. Since susu is an orthogonal transformation of V of determinant −1-1, it is a reflection and thus (su)2=1(su)^2 = 1. Consequently, u−1=sus−1u^{-1} = sus^{-1} transforms s(Δ)s(\Delta) into s(Δ′)s(\Delta'), hence the proposition.

Corollary

Let D and D' be two lines of V and let θ\theta be a measure of the angle (D^,D′^)(\widehat{D}, \widehat{D'}). Then sD′sD=ρ2θs_{D'} s_D = \rho_{2\theta}.

We know that sD′sDs_{D'} s_D is a rotation since it is of determinant 1. Let Δ\Delta and Δ′\Delta' be two half-lines starting at the origin carried by D and D'. We have

(Δ,sD′sD^(Δ))≡(Δ,sD′^(Δ))≡(Δ^,Δ′^)+(Δ′,sD′^(Δ))≡(Δ^,Δ′^)+(sD′(Δ′),sD′^(Δ))≡(Δ^,Δ′^)−(Δ′,Δ)≡2(Δ^,Δ′^)(mod2π), \begin{align*} (\Delta, \widehat{s_{D'} s_D}(\Delta)) &\equiv (\Delta, \widehat{s_{D'}}(\Delta)) \equiv (\widehat{\Delta}, \widehat{\Delta'}) + (\Delta', \widehat{s_{D'}}(\Delta)) \\ &\equiv (\widehat{\Delta}, \widehat{\Delta'}) + (s_{D'}(\Delta'), \widehat{s_{D'}}(\Delta)) \\ &\equiv (\widehat{\Delta}, \widehat{\Delta'}) - (\Delta', \Delta) \equiv 2(\widehat{\Delta}, \widehat{\Delta'}) \pmod{2\pi}, \end{align*}
hence the corollary.

Now let Δ\Delta and Δ′\Delta' be two half-lines of V such that

Δ≠Δ′andΔ≠−Δ′, \Delta \neq \Delta' \quad \text{and} \quad \Delta \neq -\Delta',
and let s and s′s' be the orthogonal reflections with respect to the lines D and D' containing Δ\Delta and Δ′\Delta'. Let θ\theta be the principal measure of the angle (D^,D′^)(\widehat{D}, \widehat{D'}). If θ∈πQ\theta \in \pi \mathbf{Q}, denote by m the smallest integer ≥1\geq 1 such that mθ∈πZm\theta \in \pi \mathbf{Z}. If θ∉πQ\theta \notin \pi \mathbf{Q}, put m=∞m = \infty. Let W be the group generated by s and s′s'.

Proposition 7

The group W is dihedral* (Chap. IV, § 1, no. 2) *of order 2m2m. It consists of the rotations ρ2nθ\rho_{2n\theta} and the products ρ2nθs\rho_{2n\theta} s for n∈Zn \in \mathbf{Z}. The transforms of D and D' by the elements of W are the transforms of D by the rotations ρnθ\rho_{n\theta} for n∈Zn \in \mathbf{Z}.

The Corollary to Prop. 6 shows that s′ss's is of order m, which gives the first assertion. The elements of W are thus of the form (s′s)n=ρ2nθ(s's)^n = \rho_{2n\theta} or (s′s)ns=ρ2nθs(s's)^n s = \rho_{2n\theta} s; the last assertion follows from this, since D′=ρθ(D)D' = \rho_\theta(D).

Corollary

Let C be the open angular sector formed by the union of the open half-lines Δ1\Delta_1 starting at the origin and such that 0<(Δ^,Δ1^)<θ0 < (\widehat{\Delta}, \widehat{\Delta_1}) < \theta. Then* no transform of D or D' by an element of W meets C if and only if m is finite and

θ=π/m. \theta = \pi / m.

If m=∞m = \infty, the image of the set nθn \theta (n∈Zn \in \mathbf{Z}) is dense in R/2πZ\mathbf{R}/2\pi \mathbf{Z} (General Topology, Chap. VII, § 1, Prop. 11); the union of the transforms of D by the elements of W is thus dense in V and meets C. If m is finite and if θ=kπ/m\theta = k \pi / m with 1<k<m1 < k < m, the integers k and m being relatively prime, there exists an integer h such that hk≡1hk \equiv 1 mod. m; then (D,ρhθ(D))≡π/m(D, \rho_{h \theta}(D)) \equiv \pi / m (mod. π\pi), and ρhθ(D)\rho_{h \theta}(D) meets C. This shows that the condition is necessary. The converse is immediate.

Remark

Assume that m is finite and that θ=π/m\theta = \pi / m. If n∈Zn \in \mathbf{Z}, let CnC_n be the union of the open half-lines Δ1\Delta_1 starting at 0 such that

nθ<(Δ‾,Δ1‾)<(n+1)θ. n \theta < (\overline{\Delta}, \overline{\Delta_1}) < (n + 1) \theta.

The CnC_n for −m≤n<m-m \leq n < m are connected open subsets forming a partition of E−⋃nDnE - \bigcup_n D_n (where Dn=ρnθ(D)D_n = \rho_{n \theta}(D)). These are therefore the chambers determined in E by the system of m lines DnD_n (1≤n≤m1 \leq n \leq m). We have C2k=ρ2kθ(C)C_{2k} = \rho_{2k \theta}(C) and C2k−1=ρ2kθs(C)C_{2k-1} = \rho_{2k \theta s}(C). Moreover, Cn=CC_n = C if and only if n∈2mZn \in 2m \mathbf{Z}. Consequently, the group W permutes the chambers CnC_n simply-transitively.

We show finally that, if w∈Ww \in W is such that the chambers C and w(C)w(C) are on opposite sides of the line D, then l(sw)=l(w)−1l(sw) = l(w) - 1 (the lengths being taken with respect to S={s,s′}S = \{s, s'\}). Indeed, the assumption implies that w(C)=Cnw(C) = C_n with −m≤n<0-m \leq n < 0. If n=−2kn = -2k, we have w=(ss′)kw = (ss')^k and sw=s′(ss′)k−1sw = s'(ss')^{k-1}, so l(w)=2kl(w) = 2k and l(sw)=2k−1l(sw) = 2k - 1 (Chap. IV, § 1, no. 2, Remark). If n=−2k+1n = -2k + 1, we have

w=(ss′)k−1sandsw=(s′s)k−1, w = (ss')^{k-1}s \quad \text{and} \quad sw = (s's)^{k-1},

hence l(w)=2k−1l(w) = 2k - 1 and l(sw)=2k−2l(sw) = 2k - 2. Q.E.D.

Exercises

See the exercises for § 2.

Transcribed from the printed volume. Not translated.