Extending Lie homomorphisms. The subalgebra : correspondence. GIven a Lie homomorphism φ : G → H ’we have a corresponding homomorphism φ0 : g → h. Here are three of the four ways to define this φ0 0 1. Using g = T1G. Then φ = dφ1 : T1G → T1H. 2. Using g = left-invariant 0 vector fields. Then φ (ξ) = φ∗ξ. 3. Using one-parameter γ(t) = exp(tX). Then φ0(γ) = φ ◦ γ. Now we want to go the other way: given φ0 : g → h with G, H connected Lie groups find a φ : G → H inducing φ0. If φ is to exist one must have that φ(exp(tX)) = exp(tφ0(X)). This uniquely defines φ in a neighborhood of the identity. But it may fail to yield a good definition everywhere. Example. g = h = R. G = H = S1. φ0(x) = cx. The exponential g → G is x 7→ eix so that this φ0 would have to induce φ(eix = eicx which is not defined as a map from the circle to the circle unless c is an integer. The solution is to pass to the universal cover. Theorem 1. If G is simply connected and H is connected then any Lie algebra homomorphism φ0 : g → h induces a unique Lie group homomorphism φ : G → H. A key to proving this theorem is another important theorem Theorem 2 (subalgebra-subgroup theorem). To each subalgebra K ⊂ g of a con- nected Lie group G there is a unique injectively immersed connected subgroup i : K → G with di1(T1K) = K. Warning; K need not be embedded, i.e. it need not be closed. For example Irrational flow on the two torus T 2 defines an immersion of K = R into G = T 2. Proof of subalgebra-subgroup theorem This theorem is a special case of the Frobenius integrability theorem. The leaf through 1 ∈ G is the immersed subgroup. Proof of simply connected lift theorem. G × H forms a Lie group with Lie algebra g ⊕ h. The graph of the map φ0 defines a Lie subalgebra of this direct sum Lie algebra. Consequently, by the subalgebra-subgroup theorem, there is a unique injectively embedded i : K → G × H whose image realizes this graph. Consider the composition i ◦ π : K → G × H → G where π : G × H → G is the projection. Its differential is 1-1 and onto, being just the compositon g → g⊕K → g. Consequently it defines a covering map K → G. But G is simply connected, so this covering map is an isomorphism and we can identify K with G It follows that if π2 : G × H → H is the projection then i ◦ π2 : G → H is a Lie group homomorphism and induces 0 φ . This i ◦ π2 is φ. QED.

1