John Stillwell - Naive Lie Theory.pdf - Index of
John Stillwell - Naive Lie Theory.pdf - Index of
John Stillwell - Naive Lie Theory.pdf - Index of
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
5.4 Algebraic properties <strong>of</strong> the tangent space 105<br />
Definition. A matrix <strong>Lie</strong> algebra is a vector space <strong>of</strong> matrices that is closed<br />
under the <strong>Lie</strong> bracket [X,Y ]=XY −YX.<br />
All the <strong>Lie</strong> algebras we have seen so far have been matrix <strong>Lie</strong> algebras,<br />
and in fact there is a theorem (Ado’s theorem) saying that every <strong>Lie</strong> algebra<br />
is isomorphic to a matrix <strong>Lie</strong> algebra. Thus it is not wrong to say simply<br />
“<strong>Lie</strong> algebra” rather than “matrix <strong>Lie</strong> algebra,” and we will usually do so.<br />
Perhaps the most important idea in <strong>Lie</strong> theory is to study <strong>Lie</strong> groups<br />
by looking at their <strong>Lie</strong> algebras. This idea succeeds because vector spaces<br />
are generally easier to work with than curved objects—which <strong>Lie</strong> groups<br />
usually are—and the <strong>Lie</strong> bracket captures most <strong>of</strong> the group structure.<br />
However, it should be emphasized at the outset that g does not always<br />
capture G entirely, because different <strong>Lie</strong> groups can have the same <strong>Lie</strong><br />
algebra. We have already seen one class <strong>of</strong> examples. For all n, O(n) is<br />
different from SO(n), but they have the same tangent space at 1 and hence<br />
the same <strong>Lie</strong> algebra. There is a simple geometric reason for this: SO(n)<br />
is the subgroup <strong>of</strong> O(n) whose members are connected by paths to 1. The<br />
tangent space to O(n) at 1 is therefore the tangent space to SO(n) at 1.<br />
Exercises<br />
If, instead <strong>of</strong> considering the path C s (t) =A(s)B(t)A(s) −1 in G we consider the<br />
path<br />
D s (t)=A(s)B(t)A(s) −1 B(t) −1 for some fixed value <strong>of</strong> s,<br />
then we can relate the <strong>Lie</strong> bracket [X,Y] <strong>of</strong> X,Y ∈ T 1 (G) to the so-called commutator<br />
A(s)B(t)A(s) −1 B(t) −1 <strong>of</strong> smooth paths A(s) and B(t) through 1 in G.<br />
5.4.1 Find D ′ s(t), and hence show that D ′ s(0)=A(s)YA(s) −1 −Y.<br />
5.4.2 D ′ s (0) ∈ T 1(G) (why?) and hence, as s varies, we have a smooth path E(s)=<br />
D ′ s(0) in T 1 (G) (why?).<br />
5.4.3 Show that the velocity E ′ (0) equals XY −YX, and explain why E ′ (0) is in<br />
T 1 (G).<br />
The tangent space at 1 is the most natural one to consider, but in fact all<br />
elements <strong>of</strong> G have the “same” tangent space.<br />
5.4.4 Show that the smooth paths through any g ∈ G are <strong>of</strong> the form gA(t),where<br />
A(t) is a smooth path through 1.<br />
5.4.5 Deduce from Exercise 5.4.4 that the space <strong>of</strong> tangents to G at g is isomorphic<br />
to the space <strong>of</strong> tangents to G at 1.