For n≥1n\geq1, consider the standard action of on Rn+1\mathbb R^{n+1}, hence on the unit sphere

Sn={x∈Rn+1:∥x∥=1}.S^n=\{x\in\mathbb R^{n+1}:\|x\|=1\}.

This is a smooth .

Transitivity and stabilizer

Fix the “north pole” p=en+1=(0,…,0,1)p=e_{n+1}=(0,\dots,0,1). For any x∈Snx\in S^n, there exists g∈SO(n+1)g\in SO(n{+}1) with g⋅p=xg\cdot p=x (choose an orthonormal basis sending en+1e_{n+1} to xx), so the action is .

The stabilizer of pp consists of rotations preserving the orthogonal complement p⊥≅Rnp^\perp\cong\mathbb R^n, hence

Stab(p)≅SO(n),\mathrm{Stab}(p)\cong SO(n),

as an embedded (see ).

Identification with a coset space

Define

Φ:SO(n+1)/SO(n)⟶Sn,Φ(g SO(n)):=g⋅p.\Phi: SO(n{+}1)/SO(n) \longrightarrow S^n,\qquad \Phi(g\,SO(n)):=g\cdot p.

This is well-defined because elements of SO(n)SO(n) fix pp. It is bijective by transitivity and the stabilizer description, and it is a diffeomorphism once we use the standard smooth structure on the SO(n+1)/SO(n)SO(n{+}1)/SO(n) (which exists because SO(n)SO(n) is closed; compare the ). Thus

Sn≅SO(n+1)/SO(n)S^n \cong SO(n{+}1)/SO(n)

as a .

Dimension check

Using dim⁡SO(m)=m(m−1)2\dim SO(m)=\frac{m(m-1)}{2},

dim⁡(SO(n+1)/SO(n))=(n+1)n2−n(n−1)2=n=dim⁡Sn,\dim\bigl(SO(n{+}1)/SO(n)\bigr) =\frac{(n{+}1)n}{2}-\frac{n(n-1)}{2}=n=\dim S^n,

consistent with the identification.

Special case. For n=2n=2, this reads S2≅SO(3)/SO(2)S^2\cong SO(3)/SO(2), connecting directly to .