Lie Theory
Table of Contents
1. Lie Group
2. Lie Algebra
2.1. Logarithmic Map
- Lie group can be reduced to Lie algebra through logarithmic map, in which it is easier to do calculations.
2.2. Differential on the Action Space
- Lie algebra is the vector field toward the transformation direction on the vector space on which the transformation is applied.2
2.3. Tangent Space at the Identity
2.4. Lie Bracket
- Let us calculate group multiplication of Lie group, within Lie algebra.
- Its properties comes from the properties of group multiplication.
- It is kind of a second derivative.
- It is defined to satisfy the Jacobi identity:
- \[ [X, [Y,Z]] + [Y, [Z,X]] + [Z, [X, Y]] = 0 \]
- We want to define some kind of multiplication on \(T_eG\)
- but \(ab=A'(0)B'(0)=\partial_t\partial_s(A(t)B(s))|_{t=0,s=0}\not\in T_eG\), because \(\partial_t (A(s)B(t))|_{t=0}\in T_{A(s)}G\).
- Define \(C_s(t)=A(s)B(t)A^{-1}(s)\) and take derivative, we have \(\partial_t C_s(t)|_{t=0}=A(s)bA^{-1}(s)\in T_eG\) and \(\partial_s\partial_t C_s(t)|_{t=0,s=0}=ab-ba \in T_eG\). We make it a definition: \([a,b]\equiv ab-ba\).
- For an element of Lie algebra \(A\), \(gAg^{-1} = e^{tB}Ae^{-tB}\) is also in the Lie algebra, which is the vector field \(A\) but conjugated along a different axis. The derivative evaluated at \(t=0\) is \([A,B]=AB-BA\).5
- Notable example of this is the \([\log\mathbf{k}, \log\mathbf{j}] = \log\mathbf{i}\). Rotating \(\log\mathbf{k}\) about \(\mathbf{j}\) axis, the rate of change at each point is \(\log\mathbf{i}\).
- Conjugation diagram6 excalidraw:./conjugation.excalidraw.svg
2.4.1. Properties
- Distribution Rule
- \[
[X,[Y,Z]] = [[X, Y], Z] + [Y, [X, Z]]
\]
- This is the generalization of the adjoint operation.7
- \[ [[X,Y],Z] = [X, [Y, Z]] - [Y, [X, Z]] \]
- \[
[X,[Y,Z]] = [[X, Y], Z] + [Y, [X, Z]]
\]
2.5. Classification
2.5.1. Simple Finite Lie Algebra
- Four infinite families \(A_n,\ B_n,\ C_n,\ D_n\)
- Five exceptional Lie algebras \(E_6, E_7, E_8, F_4, G_2\).
- Theorem: There is no more.
- \(n\) is the rank which is the number of nodes in the Dynkin diagram, and also the number of simple roots.
- Heavy use of matrix calculus
3. Root System
- Some of the Lie algebra can be represented in a finite way, which in turn form a simple set of vectors that represents the structure of the Lie algebra.
3.1. \(\mathfrak{sl}(2)\)
- \[ \textcolor{RoyalBlue}{H} \equiv \begin{pmatrix} 1 & 0 \\ 0 & -1\end{pmatrix} \quad \textcolor{OrangeRed}{E} \equiv \begin{pmatrix} 0 & 1 \\ 0 & 0\end{pmatrix} \quad \textcolor{ForestGreen}{F} \equiv \begin{pmatrix} 0 & 0 \\ 1 & 0\end{pmatrix} \]
- acting on the vector space with five basis vectors:
- Notice that they satisfies: \[ [H, E] = 2E,\quad [H, F] = -2F,\quad [E, F] = H. \]
3.2. \(\mathfrak{sl}(3)\)
- \[ H_1, H_2, \textcolor{OrangeRed}{E_1}, \textcolor{ForestGreen}{E_2}, F_1, F_2 \]
- Lie algebra can also act on themselves
3.3. Root
- The vectors that corresponds to the basis elements and their Lie brackets within a Lie algebra.
- The Lie bracket corresponds to the vector addition of roots.
3.3.1. Positive Roots
- \(\Delta_+\)
- Roots that are on the upper half, including the ones on the positive axes.
3.3.2. Simple Roots
- The one that corresponds to the actual basis elements within a Lie algebra.
3.4. Dynkin Diagram
3.4.1. Construction
\(i\quad j\) | \(d_i\) | \(a_{ij}\) | \(\langle \alpha_i, \alpha_j\rangle = d_ia_{ij}\) |
\(i = j\) | --- | 2 | \(2d_i\) |
\(\bullet \quad\bullet\) | --- | 0 | 0 |
\(\bullet—\bullet\) | \(d_i = d_j\) | -1 | \(-d_i\) |
\(\bullet\!\mathord{\Longrightarrow}\!\bullet\) | 2 | -1 | -2 |
\(\bullet\!\mathord{\equiv\!\!>\!\!\!\!\!\!\equiv\!\equiv}\!\bullet\) | 3 | -1 | -3 |
\(\bullet\!\mathord{\Longleftarrow}\!\bullet\) | 1 | -2 | -2 |
\(\bullet\!\mathord{\equiv\!\!<\!\!\!\!\!\!\equiv\!\equiv}\!\bullet\) | 1 | -3 | -3 |
- Dynkin diagram determines a root system
- Note \(\|\alpha_i\| = \sqrt{2d_i}\)
- \(a_{ij}\) is the entries of the Cartan matrix and can equivalently be defined: \[ a_{ji} = 2\frac{\langle \alpha_i, \alpha_j\rangle}{\langle \alpha_j, \alpha_j\rangle} \]
3.4.2. Exceptional G2
- \(\mathfrak{g}_2\)
- The Cross Product and the Exceptional G2 - YouTube
- The result of the 7-Dimensional cross product generally does not follow the transformation rule.
- But there's subgroup of the \(\mathfrak{so}(7)\), which contains 14 families of two-plane rotation that preserves the relationship. This is the G2.
- The existence is based on the multiplicative property of the octonion. G2 is the automorphism group of octonions.
4. Baker-Campbell-Hausdorff Formula
- The value of \(Z\) that solves \(e^Xe^Y = e^Z\) with \(X, Y\) in the Lie algebra.
- It is not necessarily convergent.
- Baker, Campbell, Hausdorff only stated the qualitative form of the formula, that it can be written in terms of possibly infinitely nested commutators yielding another element in the Lie algebra if it is convergent.
5. Dynkin's Formula
- First explicit formula
- \[
\log(\exp X\exp Y) = \sum_{n=1}^\infty\frac{(-1)^{n-1}}{n}\sum_{\begin{smallmatrix}r_1+s_1 >0\\[-.5em] \vdots\\[.3em] r_n+s_n >0\end{smallmatrix}} \frac{[X^{r_1}Y^{s_1}X^{r_2}Y^{s_2}\cdots X^{r_n}Y^{s_n}]}{\left(\sum_{j=1}^n(r_j+s_j)\right)\cdot \prod_{i=1}^nr_i!s_i!}
\]
- where the bracket is the notation for the Lie bracket nested on the right iterated by the exponent, and \(r_i, s_i\) are assumed to be nonnegative, with the understanding that \([X] := X\).
- The first few terms are well-known to be:
- \[ Z(X,Y) = X+Y + \frac{1}{2}[X,Y] + \frac{1}{12}([X,[X,Y]] + [Y,[Y,X]]) -\frac{1}{24}[Y,[X,[X,Y]]] -\cdots \]
6. Reference
Footnotes:
1
Mathemaniac. https://youtu.be/IlqVo3sJFLE. Introduction
2
Mathemaniac. Lie Algebra. https://youtu.be/gj4kvpy1eCE?si=38MLHlZbAscbyMfJ
3
Michael Penn. https://youtu.be/2_KVbtyufq8?si=liu02OhUrZ5JYeg-
4
Eigenchris. https://youtu.be/IPzwqAVfce4?si=3teaHPXQgJsNIwaq
5
Mathemaniac. Lie Algebra. https://youtu.be/gj4kvpy1eCE?si=38MLHlZbAscbyMfJ
6
Eigenchris. https://youtu.be/IPzwqAVfce4?si=3teaHPXQgJsNIwaq
7
Mathemaniac. Lie Algebra. https://youtu.be/gj4kvpy1eCE?si=38MLHlZbAscbyMfJ