Masteries · Jason GlowneyArithmoí Book 1

Masteries

1.2.8 Dihedral Group Tangent Spaces

In this sec­tion, we apply the tan­gent space from Lie theory to our de­fined D3ℝ, D3ℂ, D3𝕔D_{3\mathbb{R}},\ D_{3\mathbb{C}},\ D_{3\mathbb{c}} groups, and to the co­in­cid­ing group de­riv­a­tives D3ℝ′, D3ℂ′, D3𝕔′D_{3\mathbb{R}}^{'},\ D_{3\mathbb{C}}^{'},\ D_{3\mathbb{c}}^{'}. For D3ℝ, D3ℂ, D3𝕔D_{3\mathbb{R}},\ D_{3\mathbb{C}},\ D_{3\mathbb{c}}, we find that the skew-sym­me­try con­di­tions hold Xj+Xj∗=0X_{j} + X_{j}^{*} = 0, for their tan­gent vec­tors XjX_{j}. Pre­vi­ous­ly, we have shown the ex­is­tence of de­riv­a­tives for the D3ℝ, D3ℂ, D3𝕔D_{3\mathbb{R}},\ D_{3\mathbb{C}},\ D_{3\mathbb{c}} groups, and given this, we can define the tan­gent space through the smooth paths these groups have pass­ing by the iden­ti­ty.

De­f­i­n­i­tion (1.2.146) – The Tan­gent Space 𝔗𝛜(𝐆(𝐀(𝐭)))\mathfrak{T}_{\mathbf{\epsilon}}\left( \mathbf{G}\left( \mathbf{A}\left( \mathbf{t} \right) \right) \right) is de­fined as the tan­gent vec­tors of smooth paths in G(A(t))G\left( A(t) \right), where they pass through the iden­ti­ty ϵ\epsilon, which rep­re­sents

𝔗ϵ(G(Aj(tαk)))=ddt G(Aj(tαk))|t=0,\begin{array}{r} \mathfrak{T}_{\epsilon}\left( G\left( A_{j}\left( t\alpha_{k} \right) \right) \right) = \frac{d}{dt}\left. \ G\left( A_{j}\left( t\alpha_{k} \right) \right) \right|_{t = 0}, \end{array}

where, G(Aj(tαk))G\left( A_{j}\left( t\alpha_{k} \right) \right), des­ig­nates the group with the set of al­ge­bras, Aj(tαk)={tAjℝ,tAjℂ,tAj𝕔}A_{j}\left( t\alpha_{k} \right) = \left\{ tA_{j}^{\mathbb{R}},tA_{j}^{\mathbb{C}},{tA}_{j}^{\mathbb{c}} \right\}, and tt rep­re­sents the vari­ables, {xj,yj}\{ x_{j},y_{j}\}, of the Lie al­ge­bras {Xj,Yj}∈A.\left\{ X_{j},Y_{j} \right\} \in A.

From the pre­vi­ous sec­tions, we know that the matrix-by-scalar de­riv­a­tive of the G(Aj(tαk))G\left( A_{j}\left( t\alpha_{k} \right) \right) group will be equiv­a­lent to

G′(Aj(tαk))=ddtG(Aj(tαk))=αkAjG(Aj(tαk)),\begin{array}{r} G^{'}\left( A_{j}\left( t\alpha_{k} \right) \right) = \frac{d}{dt}G\left( A_{j}\left( t\alpha_{k} \right) \right) = \alpha_{k}A_{j}G\left( A_{j}\left( t\alpha_{k} \right) \right), \end{array}
(1.2.147)

mean­ing that all de­riv­a­tives will be equal to the group being dif­fer­en­ti­at­ed, mul­ti­plied by the prod­uct of its scalar factor αk∈ℂ\alpha_{k}\in \mathbb{C} with the AjA_{j} al­ge­bra. The AjA_{j} al­ge­bras are the sym­me­tries of the equi­lat­er­al tri­an­gle, rep­re­sent­ing the core char­ac­ter­is­tics of these higher-di­men­sion­al num­bers. Their role should not be un­der­stat­ed and leads us to define AjA_{j}, as the pri­mor­dial al­ge­bras of the equi­lat­er­al tri­an­gle.

De­f­i­n­i­tion (1.2.148) – A pri­mor­dial al­ge­bra 𝐀𝐣[𝐧]\mathbf{A}_{\mathbf{j}}^{\mathbf{\lbrack n\rbrack}} is the finite set of per­mu­ta­tion el­e­ments ∈Dnℝ\in D_{n\mathbb{R}}, rep­re­sent­ing the sym­me­tries of its re­spec­tive reg­u­lar nn-gon. Aj[n]A_{j}^{\lbrack n\rbrack} exists in the co­in­cid­ing ℂn\mathbb{C}^{n} space, with strict­ly finite and pos­i­tive uni­tary values, which cor­re­spond to the iden­ti­ty values of each of the nn com­plex unit discs com­pris­ing its nn- di­men­sion­al space of planes.

To be clear, the Aj[3]A_{j}^{\lbrack 3\rbrack} set rep­re­sents the pri­mor­dial al­ge­bras that form the foun­da­tions of the D3ℝ, D3ℂ, D3𝕔D_{3\mathbb{R}},\ D_{3\mathbb{C}},\ D_{3\mathbb{c}} group dy­nam­ics. Aj[3]A_{j}^{\lbrack 3\rbrack} is the set of per­mu­ta­tions Aℝ={Xjℝ,Yjℝ}∈D3ℝA^{\mathbb{R}} = \left\{ X_{j}^{\mathbb{R}},Y_{j}^{\mathbb{R}} \right\} \in D_{3\mathbb{R}}, com­posed of the fol­low­ing matrix per­mu­ta­tions

Aj[3]={(100001010),   (001010100),   (010100001),(100010001),   (010001100),   (001100010)}.\begin{array}{r} A_{j}^{\lbrack 3\rbrack} = \left\{ \begin{array}{r} \begin{pmatrix} 1 & 0 & 0 \\ 0 & 0 & 1 \\ 0 & 1 & 0 \end{pmatrix},\ \ \ \begin{pmatrix} 0 & 0 & 1 \\ 0 & 1 & 0 \\ 1 & 0 & 0 \end{pmatrix},\ \ \ \begin{pmatrix} 0 & 1 & 0 \\ 1 & 0 & 0 \\ 0 & 0 & 1 \end{pmatrix}, \\ \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix},\ \ \ \begin{pmatrix} 0 & 1 & 0 \\ 0 & 0 & 1 \\ 1 & 0 & 0 \end{pmatrix},\ \ \ \begin{pmatrix} 0 & 0 & 1 \\ 1 & 0 & 0 \\ 0 & 1 & 0 \end{pmatrix} \end{array} \right\}. \end{array}
(1.2.149)

So, in the ex­pres­sion G′(A j[3](tαk))=αkAj[3]G(Aj(tαk))G^{'}\left( A_{\ j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) = \alpha_{k}A_{j}^{\lbrack 3\rbrack}G\left( A_{j}\left( t\alpha_{k} \right) \right), the Aj[3]A_{j}^{\lbrack 3\rbrack} pri­mor­dial al­ge­bras rep­re­sent the sym­me­tries of the equi­lat­er­al tri­an­gle acting on the group G(Aj(tαk))G\left( A_{j}\left( t\alpha_{k} \right) \right). These pri­mor­dial sym­me­tries are ex­tend­ed into the other de­grees of free­dom ∈ℂ3\in \mathbb{C}^{3} by the αk\alpha_{k} vari­able. In the fol­low­ing ex­pres­sion, we con­sid­er the tan­gent space in terms of the action of its pri­mor­dial al­ge­bra

𝔗ϵ(G(Aj[n](tαk)))=αkAj[n][ G(Aj[n](tαk))|t=0],\begin{array}{r} \mathfrak{T}_{\epsilon}\left( G\left( A_{j}^{\lbrack n\rbrack}\left( t\alpha_{k} \right) \right) \right) = \alpha_{k}A_{j}^{\lbrack n\rbrack}\left\lbrack \left. \ G\left( A_{j}^{\lbrack n\rbrack}\left( t\alpha_{k} \right) \right) \right|_{t = 0} \right\rbrack, \end{array}
(1.2.150)

where for the case

𝔗ϵ(G(Aj[3](tαk)))=αkAj[3][ G(Aj[3](tαk))|t=0],\begin{array}{r} \mathfrak{T}_{\epsilon}\left( G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) \right) = \alpha_{k}A_{j}^{\lbrack 3\rbrack}\left\lbrack \left. \ G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) \right|_{t = 0} \right\rbrack, \end{array}
(1.2.151)

the group por­tion eval­u­at­ed at t=0t = 0, will be equal to the iden­ti­ty for all G(Aj[3](tαk))G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right). The proof for this comes from the di­ag­o­nal­iza­tion de­com­po­si­tion for each

G(Aj[3](tαk))=VeDV−1,\begin{array}{r} G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) = Ve^{D}V^{- 1}, \end{array}
(1.2.152)

for which, we al­ready found that the di­ag­o­nal ma­tri­ces for eDe^{D}, have the form

eD=(etαkλ1000etαkλ2000etαkλ3),\begin{array}{r} e^{D} = \begin{pmatrix} e^{t\alpha_{k}\lambda_{1}} & 0 & 0 \\ 0 & e^{t\alpha_{k}\lambda_{2}} & 0 \\ 0 & 0 & e^{t\alpha_{k}\lambda_{3}} \end{pmatrix}, \end{array}
(1.2.153)

for the eigen­val­ues λn\lambda_{n} from the Aj[3]A_{j}^{\lbrack 3\rbrack} pri­mor­dial al­ge­bras. When t=0t = 0, eDe^{D} will be equal to the iden­ti­ty matrix

 eD|t=0=(100010001)=ϵ,\begin{array}{r} \left. \ e^{D} \right|_{t = 0} = \begin{pmatrix} 1 & 0 & 0 \\ 0 & 1 & 0 \\ 0 & 0 & 1 \end{pmatrix} = \epsilon, \end{array}
(1.2.154)

and there­fore, all G(Aj[3](tαk))G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) eval­u­at­ed at t=0t = 0, will also be equal to the iden­ti­ty

 G(Aj[3](tαk))|t=0=ϵ.\begin{array}{r} \left. \ G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) \right|_{t = 0} = \epsilon. \end{array}
(1.2.155)

This leads to the result that the tan­gent space, for any group G(Aj[3](tαk)),G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right),is equal to

𝔗ϵ(G(Aj[3](tαk)))=αkAj[3],\begin{array}{r} \mathfrak{T}_{\epsilon}\left( G\left( A_{j}^{\lbrack 3\rbrack}\left( t\alpha_{k} \right) \right) \right) = \alpha_{k}A_{j}^{\lbrack 3\rbrack}, \end{array}
(1.2.156)

defin­ing the Aj[3]A_{j}^{\lbrack 3\rbrack} pri­mor­dial al­ge­bras and their given de­grees of free­dom in the ℂ3\mathbb{C}^{3} space, and is based upon the scalar mul­ti­ple, αj∈ℂ3\alpha_{j} \in \mathbb{C}^{3}. To present an ex­am­ple, we eval­u­ate the fol­low­ing group de­riv­a­tive

G′(A1[3](it))=iAj[3]G(A1[3](it))=(i0000i0i0)(eit000cos⁡(t)isin(t)0isin(t)cos⁡(t)),\begin{array}{r} G^{'}\left( A_{1}^{\lbrack 3\rbrack}(it) \right) = iA_{j}^{\lbrack 3\rbrack}G\left( A_{1}^{\lbrack 3\rbrack}(it) \right) = \begin{pmatrix} i & 0 & 0 \\ 0 & 0 & i \\ 0 & i & 0 \end{pmatrix}\begin{pmatrix} e^{it} & 0 & 0 \\ 0 & \cos(t) & isin(t) \\ 0 & isin(t) & \cos(t) \end{pmatrix}, \end{array}
(1.2.157)

for which, (A1[3](it))′\left( A_{1}^{\lbrack 3\rbrack}(it) \right)^{'}, rep­re­sents its de­riv­a­tive Lie al­ge­bra

(A1[3](it))′=(i+iπ20000i+iπ20i+iπ20).\begin{array}{r} \left( A_{1}^{\lbrack 3\rbrack}(it) \right)^{'} = \begin{pmatrix} i + \frac{i\pi}{2} & 0 & 0 \\ 0 & 0 & i + \frac{i\pi}{2} \\ 0 & i + \frac{i\pi}{2} & 0 \end{pmatrix}. \end{array}
(1.2.158)

In Figure 1.2.6, we present the tan­gent space, 𝔗ϵ(G(A1[3](it)))\mathfrak{T}_{\epsilon}\left( G\left( A_{1}^{\lbrack 3\rbrack}(it) \right) \right), as the three eigen­vec­tors xˆj\ {\widehat{x}}_{j} that form it when pass­ing through the iden­ti­ty at ϵ\epsilon. Tri­an­gle-1 rep­re­sents the Lie al­ge­bra A1[3](it)A_{1}^{\lbrack 3\rbrack}(it), and the de­riv­a­tive al­ge­bra (A1[3](it))′\left( A_{1}^{\lbrack 3\rbrack}(it) \right)^{'} is rep­re­sent­ed by tri­an­gle-22, which is ro­tat­ed by ϕ=iπ2\phi = \frac{i\pi}{2} in the phase space from A1[3](it)A_{1}^{\lbrack 3\rbrack}(it). The blue-plane sec­tion is tan­gent to the sphere in the neigh­bor­hood of the iden­ti­ty ϵ\epsilon, for weight­ed vec­tors rep­re­sent­ing the path the tri­an­gle's ver­tices take pass­ing through ϵ\epsilon. The Eu­clid­ean flat­ness of the tan­gent space in the figure above can be read­i­ly ap­pre­ci­at­ed, cap­tur­ing the Lie al­ge­bra giving rise to the curved geom­e­try of the G(A1[3](it))G\left( A_{1}^{\lbrack 3\rbrack}(it) \right) group.

Figure
Figure 1.2.6 – The tan­gent space for 𝔗ϵ(G(Aj[3](it)))\mathfrak{T}_{\epsilon}\left( G\left( A_{j}^{\lbrack 3\rbrack}(it) \right) \right), where Aj[3](it)A_{j}^{\lbrack 3\rbrack}(it) is the Lie al­ge­bra, (Aj[3](it))′\left( A_{j}^{\lbrack 3\rbrack}(it) \right)^{'}is

the de­riv­a­tive Lie al­ge­bra, ϵ\epsilon is the iden­ti­ty, and xˆ(Aj[3](it))\widehat{x}(A_{j}^{\lbrack 3\rbrack}(it)) are the tan­gent space vec­tors.

The Lie al­ge­bra rep­re­sen­ta­tions in the figure differ from pre­vi­ous images in the book, which were ren­der­ings of the gen­er­a­tor's column vec­tors. The Lie al­ge­bras in Figure 1.2.6 are plots of the re­sult­ing row vec­tors in the tan­gent space from the prod­ucts

iAj[3]V=V(i): (i0000i0i0)(100012−1201212)=(i000i2i20i2−i2),\begin{array}{r} iA_{j}^{\lbrack 3\rbrack}V = V_{(i)}:\ \begin{pmatrix} i & 0 & 0 \\ 0 & 0 & i \\ 0 & i & 0 \end{pmatrix}\begin{pmatrix} 1 & 0 & 0 \\ 0 & \frac{1}{\sqrt{2}} & - \frac{1}{\sqrt{2}} \\ 0 & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \end{pmatrix} = \begin{pmatrix} i & 0 & 0 \\ 0 & \frac{i}{\sqrt{2}} & \frac{i}{\sqrt{2}} \\ 0 & \frac{i}{\sqrt{2}} & - \frac{i}{\sqrt{2}} \end{pmatrix}, \end{array}
(1.2.159)
−Aj[3]V=V(−1): (−10000−10−10)(100012−1201212)=(−1000−12120−12−12),\begin{array}{r} - A_{j}^{\lbrack 3\rbrack}V = V_{( - 1)}:\ \begin{pmatrix} - 1 & 0 & 0 \\ 0 & 0 & - 1 \\ 0 & - 1 & 0 \end{pmatrix}\begin{pmatrix} 1 & 0 & 0 \\ 0 & \frac{1}{\sqrt{2}} & - \frac{1}{\sqrt{2}} \\ 0 & \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \end{pmatrix} = \begin{pmatrix} - 1 & 0 & 0 \\ 0 & - \frac{1}{\sqrt{2}} & \frac{1}{\sqrt{2}} \\ 0 & - \frac{1}{\sqrt{2}} & - \frac{1}{\sqrt{2}} \end{pmatrix}, \end{array}
(1.2.160)

where the eigen­vec­tors ma­tri­ces VV are equiv­a­lent for the iA1[3]iA_{1}^{\lbrack 3\rbrack} and −A1[3]- A_{1}^{\lbrack 3\rbrack} al­ge­bras. The re­sult­ing V(i)V_{(i)}, V(−1)V_{( - 1)} vector ma­tri­ces rep­re­sent eigen­vec­tor ma­tri­ces that solve the eigen­val­ue/eigen­vec­tor de­com­po­si­tion

αkAj[3]=VDV−1.\begin{array}{r} \alpha_{k}A_{j}^{\lbrack 3\rbrack} = VDV^{- 1}. \end{array}
(1.2.161)

For ex­am­ple, the V(i)V_{(i)} eigen­vec­tor matrix forms the fol­low­ing prod­uct with the −A1[3]- A_{1}^{\lbrack 3\rbrack} di­ag­o­nal eigen­val­ue matrix

V(i)D(−1)(V(i))−1=−A1[3]:(i000i2i20i2−i2)(−1000−10001)(−i000−i2−i20−i2i2)=(−10000−10−10),\begin{array}{r} V_{(i)}D_{( - 1)}\left( V_{(i)} \right)^{- 1} = - A_{1}^{\lbrack 3\rbrack}:\begin{pmatrix} i & 0 & 0 \\ 0 & \frac{i}{\sqrt{2}} & \frac{i}{\sqrt{2}} \\ 0 & \frac{i}{\sqrt{2}} & - \frac{i}{\sqrt{2}} \end{pmatrix}\begin{pmatrix} - 1 & 0 & 0 \\ 0 & - 1 & 0 \\ 0 & 0 & 1 \end{pmatrix}\begin{pmatrix} - i & 0 & 0 \\ 0 & - \frac{i}{\sqrt{2}} & - \frac{i}{\sqrt{2}} \\ 0 & - \frac{i}{\sqrt{2}} & \frac{i}{\sqrt{2}} \end{pmatrix} = \begin{pmatrix} - 1 & 0 & 0 \\ 0 & 0 & - 1 \\ 0 & - 1 & 0 \end{pmatrix}, \end{array}
(1.2.162)

which equals −A1[3].- A_{1}^{\lbrack 3\rbrack}.We have not yet ex­plic­it­ly stated that

G′(At)=AetA,\begin{array}{r} G^{'}(At) = Ae^{tA}, \end{array}
(1.2.163)

but this does hold for the Lie groups with XX and YY-type Lie al­ge­bras, which is very sim­i­lar to the non-matrix de­riv­a­tives. This im­plies that the G′(At)G^{'}(At) group de­riv­a­tive is equiv­a­lent to the G(At)G(At) group being acted on by its Lie al­ge­bra via matrix mul­ti­pli­ca­tion. This makes in­tu­itive sense since a de­riv­a­tive de­fines how the group changes with re­spect to a cal­cu­la­tive time. Given that a matrix group G(At)G(At) com­mutes with itself, it holds that

G′(At)=AetA=etAA,\begin{array}{r} G^{'}(At) = Ae^{tA} = e^{tA}A, \end{array}
(1.2.164)

where mul­ti­pli­ca­tion on either side by the Lie al­ge­bra is equiv­a­lent. The group de­riv­a­tive above is equal to the eigen­de­com­po­si­tion tech­nique that we have been uti­liz­ing, where G′(At)G^{'}(At) holds as a result of the fol­low­ing the­o­rem.

The­o­rem (1.2.165) – For a di­ag­o­nal­iz­able group G(At)G(At), with a Lie al­ge­bra AA, the fol­low­ing iden­ti­ty holds for the group de­riv­a­tive G′(At)G'(At)

G′(At)=V(ddt[eD(At)])V−1=AetA.\begin{array}{r} G'(At) = V\left( \frac{d}{dt}\left\lbrack e^{D(At)} \right\rbrack \right)V^{- 1} = Ae^{tA}. \end{array}

Proof – For a di­ag­o­nal­iz­able Lie al­ge­bra AtAt, the fol­low­ing eigen­de­com­po­si­tion holds

At=V(D(At))V−1=A=V(λ1t000λ2t000λ3t)V−1,At = V\left( D(At) \right)V^{- 1} = A = V\begin{pmatrix} \lambda_{1}t & 0 & 0 \\ 0 & \lambda_{2}t & 0 \\ 0 & 0 & \lambda_{3}t \end{pmatrix}V^{- 1},

where D(At)D(At) is a di­ag­o­nal­ized matrix of eigen­val­ues, for t∈ℝt\in \mathbb{R}, and the matrix VV is the set of eigen­vec­tors of the matrix At.At. The ex­po­nen­tial map­ping of AtAt yields the Lie group, G(At)=eAtG(At) = e^{At}, which has the eigen­de­com­po­si­tion

G(At)=eAt=VeD(At)V−1=V(eλ1t000eλ2t000eλ3t)V−1.G(At) = e^{At} = Ve^{D(At)}V^{- 1} = V\begin{pmatrix} e^{\lambda_{1}t} & 0 & 0 \\ 0 & e^{\lambda_{2}t} & 0 \\ 0 & 0 & e^{\lambda_{3}t} \end{pmatrix}V^{- 1}.

Find­ing the group's de­riv­a­tive G′(At)G^{'}(At) using the de­com­po­si­tion tech­nique yields

G′(At)=V(ddt[eD(At)])V−1=V(ddt(eλ1t000eλ2t000eλ3t))V−1=V(λ1eλ1t000λ2eλ2t000λ3eλ3t)V−1,G^{'}(At) = V\left( \frac{d}{dt}\left\lbrack e^{D(At)} \right\rbrack \right)V^{- 1} = V\left( \frac{d}{dt}\begin{pmatrix} e^{\lambda_{1}t} & 0 & 0 \\ 0 & e^{\lambda_{2}t} & 0 \\ 0 & 0 & e^{\lambda_{3}t} \end{pmatrix} \right)V^{- 1} = V\begin{pmatrix} {\lambda_{1}e}^{\lambda_{1}t} & 0 & 0 \\ 0 & \lambda_{2}e^{\lambda_{2}t} & 0 \\ 0 & 0 & \lambda_{3}e^{\lambda_{3}t} \end{pmatrix}V^{- 1},

which, after fac­tor­ing out the eigen­val­ues λj\lambda_{j}, be­comes

G′(At)=V(λ1000λ2000λ3)(eλ1t000eλ2t000eλ3t)V−1.G^{'}(At) = V\begin{pmatrix} \lambda_{1} & 0 & 0 \\ 0 & \lambda_{2} & 0 \\ 0 & 0 & \lambda_{3} \end{pmatrix}\begin{pmatrix} e^{\lambda_{1}t} & 0 & 0 \\ 0 & e^{\lambda_{2}t} & 0 \\ 0 & 0 & e^{\lambda_{3}t} \end{pmatrix}V^{- 1}.

Given the iden­ti­ty

VV−1=V−1V=ϵ,VV^{- 1} = V^{- 1}V = \epsilon,

where ϵ\epsilon is the group's iden­ti­ty, G′(At)G^{'}(At) is equiv­a­lent to

G′(At)=V(λ1000λ2000λ3)V−1V(eλ1t000eλ2t000eλ3t)V−1.G^{'}(At) = V\begin{pmatrix} \lambda_{1} & 0 & 0 \\ 0 & \lambda_{2} & 0 \\ 0 & 0 & \lambda_{3} \end{pmatrix}V^{- 1}V\begin{pmatrix} e^{\lambda_{1}t} & 0 & 0 \\ 0 & e^{\lambda_{2}t} & 0 \\ 0 & 0 & e^{\lambda_{3}t} \end{pmatrix}V^{- 1}.

Set­ting the tt vari­able to one, we find that AA is equal to

 At|t=1=A=V(λ1000λ2000λ3)V−1,\left. \ At \right|_{t = 1} = A = V\begin{pmatrix} \lambda_{1} & 0 & 0 \\ 0 & \lambda_{2} & 0 \\ 0 & 0 & \lambda_{3} \end{pmatrix}V^{- 1},

and given that

G(At)=etA=V(eλ1t000eλ2t000eλ3t)V−1,G(At) = e^{tA} = V\begin{pmatrix} e^{\lambda_{1}t} & 0 & 0 \\ 0 & e^{\lambda_{2}t} & 0 \\ 0 & 0 & e^{\lambda_{3}t} \end{pmatrix}V^{- 1},

it holds that

G′(At)=AetA.G^{'}(At) = Ae^{tA}.
□\square

The Lie al­ge­bra of a group seen in the con­text of the tan­gent space allows one to vi­su­al­ize a matrix group's defin­ing al­ge­bra geo­met­ri­cal­ly. The fol­low­ing sec­tion con­tem­plates the de­pen­dent con­nec­tions a Lie group has with its Lie al­ge­bra and sub­al­ge­bras beyond the tan­gent space.