Masteries · Jason GlowneyArithmoí Book 1

Masteries

1.5.2 The Symmetric Cone Group Lie Algebra

The S3ℂΔS_{3\mathbb{C}}^{\Delta} group's tan­gent space, its Lie brack­et op­er­a­tions, and its vector space struc­ture is the Lie al­ge­bra of the group de­fined as 𝔰3ℂΔ\mathfrak{s}_{3\mathbb{C}}^{\Delta}. The Lie al­ge­bra 𝔰3ℂΔ\mathfrak{s}_{3\mathbb{C}}^{\Delta} is a matrix Lie al­ge­bra by way of Ado's the­o­rem1, which states that every finite-di­men­sion­al Lie al­ge­bra can be en­com­passed as a Lie al­ge­bra of square ma­tri­ces under the com­mu­ta­tor brack­et (Lie brack­et). A matrix Lie al­ge­bra is a vector space of ma­tri­ces that is closed under a Lie brack­et [X,Y]=XY−YX.\lbrack X,Y\rbrack = XY - YX.2 For an el­e­ment XX in the Lie al­ge­bra 𝔰3ℂΔ\mathfrak{s}_{3\mathbb{C}}^{\Delta}, the Lie brack­et with itself equals zero,

[X,X]=XX−XX=0.\begin{array}{r} \lbrack X,X\rbrack = XX - XX = 0. \end{array}
(1.5.6)

We can rep­re­sent our matrix el­e­ments 𝒳i,𝒴i\mathcal{X}_{i},\mathcal{Y}_{i} ∈𝔰3ℂΔ\in \mathfrak{s}_{3\mathbb{C}}^{\Delta}, as fol­lows

𝒳j={(a0000b0c0)1,(00a0b0c00)2,(0a0b0000c)3},\mathcal{X}_{j} = \left\{ \begin{pmatrix} a & 0 & 0 \\ 0 & 0 & b \\ 0 & c & 0 \end{pmatrix}_{1},\begin{pmatrix} 0 & 0 & a \\ 0 & b & 0 \\ c & 0 & 0 \end{pmatrix}_{2},\begin{pmatrix} 0 & a & 0 \\ b & 0 & 0 \\ 0 & 0 & c \end{pmatrix}_{3} \right\},
𝒴j={(d000e000f)1,(0d000ef00)2,(00de000f0)3},\begin{array}{r} \mathcal{Y}_{j} = \left\{ \begin{pmatrix} d & 0 & 0 \\ 0 & e & 0 \\ 0 & 0 & f \end{pmatrix}_{1},\begin{pmatrix} 0 & d & 0 \\ 0 & 0 & e \\ f & 0 & 0 \end{pmatrix}_{2},\begin{pmatrix} 0 & 0 & d \\ e & 0 & 0 \\ 0 & f & 0 \end{pmatrix}_{3} \right\}, \end{array}
(1.5.7)

which for 𝒳j\mathcal{X}_{j}, the a,b,ca,b,c en­tries remain row con­sis­tent, as do the en­tries d,e,fd,e,f of 𝒴j\mathcal{Y}_{j}. The axes of ro­ta­tion for the 𝒳j\mathcal{X}_{j} el­e­ments are de­fined as 𝒳1=a,𝒳2=b\mathcal{X}_{1} = a,\mathcal{X}_{2} = b, and 𝒳3=c\mathcal{X}_{3} = c. The reader should recall that the set of en­tries, {a,b,c,d,e,f}\{ a,b,c,d,e,f\}, may take any value from the set {±1,±i,±(−1)(14),±(−1)(34)}\left\{ \pm 1, \pm i, \pm ( - 1)^{\left( \frac{1}{4} \right)}, \pm ( - 1)^{\left( \frac{3}{4} \right)} \right\}. Lie brack­ets of the form [𝒳j,𝒳k]\left\lbrack \mathcal{X}_{j},\mathcal{X}_{k} \right\rbrack yield re­sults ∈𝒴l\in \mathcal{Y}_{l}. Start­ing with [𝒳1,𝒳2]\left\lbrack \mathcal{X}_{1},\mathcal{X}_{2} \right\rbrack

[𝒳1,𝒳2]=(0−aca2bc0−b2−acbc0)=𝒴3({d=a2e=bcf=bc )−𝒴2 ({d=ace=b2f=ac ),\begin{array}{r} \left\lbrack \mathcal{X}_{1},\mathcal{X}_{2} \right\rbrack = \begin{pmatrix} 0 & - ac & a^{2} \\ bc & 0 & - b^{2} \\ - ac & bc & 0 \end{pmatrix} = \mathcal{Y}_{3}\left( \left\{ \begin{array}{r} d = a^{2} \\ e = bc \\ f = bc \end{array} \right.\ \right) - \mathcal{Y}_{2}\ \left( \left\{ \begin{array}{r} d = ac \\ e = b^{2} \\ f = ac \end{array} \right.\ \right), \end{array}
(1.5.8)

we find the Lie brack­et result takes the form 𝒴3−𝒴2\mathcal{Y}_{3} - \mathcal{Y}_{2}, and switch­ing the brack­et order, [𝒳2,𝒳1]\left\lbrack \mathcal{X}_{2},\mathcal{X}_{1} \right\rbrack, re­sults in the neg­a­tive of [𝒳1,𝒳2]\left\lbrack \mathcal{X}_{1},\mathcal{X}_{2} \right\rbrack

[𝒳2,𝒳1]=(0ac−a2−bc0b2ac−bc0)=𝒴2 ({d=ace=b2f=ac )−𝒴3({d=a2e=bcf=bc ).\begin{array}{r} \left\lbrack \mathcal{X}_{2},\mathcal{X}_{1} \right\rbrack = \begin{pmatrix} 0 & ac & {- a}^{2} \\ - bc & 0 & b^{2} \\ ac & - bc & 0 \end{pmatrix} = \mathcal{Y}_{2}\ \left( \left\{ \begin{array}{r} d = ac \\ e = b^{2} \\ f = ac \end{array} \right.\ \right) - \mathcal{Y}_{3}\left( \left\{ \begin{array}{r} d = a^{2} \\ e = bc \\ f = bc \end{array} \right.\ \right). \end{array}
(1.5.9)

This neg­a­tive re­la­tion­ship holds for any Lie brack­et of el­e­ments ∈𝒳i\in \mathcal{X}_{i}, [𝒳i,𝒳j]=−[𝒳j,𝒳i].\left\lbrack \mathcal{X}_{i},\mathcal{X}_{j} \right\rbrack = - \left\lbrack \mathcal{X}_{j},\mathcal{X}_{i} \right\rbrack.For [𝒳2,𝒳3]\left\lbrack \mathcal{X}_{2},\mathcal{X}_{3} \right\rbrack and [𝒳3,𝒳1]\left\lbrack \mathcal{X}_{3},\mathcal{X}_{1} \right\rbrack, the Lie brack­et re­sults are

[𝒳2,𝒳3]=(0−abacb20−ab−c2ac0)=𝒴3({d=ace=b2f=ac )−𝒴2 ({d=abe=abf=c2 ),\begin{array}{r} \left\lbrack \mathcal{X}_{2},\mathcal{X}_{3} \right\rbrack = \begin{pmatrix} 0 & - ab & ac \\ b^{2} & 0 & - ab \\ - c^{2} & ac & 0 \end{pmatrix} = \mathcal{Y}_{3}\left( \left\{ \begin{array}{r} d = ac \\ e = b^{2} \\ f = ac \end{array} \right.\ \right) - \mathcal{Y}_{2}\ \left( \left\{ \begin{array}{r} d = ab \\ e = ab \\ f = c^{2} \end{array} \right.\ \right), \end{array}
(1.5.10)
[𝒳3,𝒳1]=(0−a2abab0−bc−bcc20)=𝒴3({d=abe=abf=c2 )−𝒴2 ({d=a2e=bcf=bc ),\begin{array}{r} \left\lbrack \mathcal{X}_{3},\mathcal{X}_{1} \right\rbrack = \begin{pmatrix} 0 & - a^{2} & ab \\ ab & 0 & - bc \\ - bc & c^{2} & 0 \end{pmatrix} = \mathcal{Y}_{3}\left( \left\{ \begin{array}{r} d = ab \\ e = ab \\ f = c^{2} \end{array} \right.\ \right) - \mathcal{Y}_{2}\ \left( \left\{ \begin{array}{r} d = a^{2} \\ e = bc \\ f = bc \end{array} \right.\ \right), \end{array}
(1.5.11)

for which a pat­tern emerges, where the axes of ro­ta­tion for the re­spec­tive el­e­ments 𝒳j\mathcal{X}_{j} appear as squared values in the column rep­re­sent­ing the jthj^{th} column number not present in the brack­et op­er­a­tion. The axis in the left-po­si­tion of the Lie brack­et op­er­a­tion is pos­i­tive, while the axis in the right-po­si­tion of the Lie brack­et is neg­a­tive. To fur­ther clar­i­fy, we take the [𝒳2,𝒳3]\left\lbrack \mathcal{X}_{2},\mathcal{X}_{3} \right\rbrack brack­et op­er­a­tion, where 𝒳1\mathcal{X}_{1} is not present in the op­er­a­tion's cal­cu­la­tion, having 𝒳2\mathcal{X}_{2} on the left in the Lie brack­et and 𝒳3\mathcal{X}_{3} on the right. We then expect to see the squared en­tries of the 𝒳2,𝒳3\mathcal{X}_{2},\mathcal{X}_{3}-axes (b2,−c2)\left( b^{2}, - c^{2} \right) in the 1st axis column of the matrix result from the re­spec­tive X2,X3X_{2},X_{3}, rows. In bold font the squared axis num­bers are shown as they appear in the X1X_{1} column, from the Lie brack­et op­er­a­tion [𝒳2,𝒳3]\left\lbrack \mathcal{X}_{2},\mathcal{X}_{3} \right\rbrack

[𝒳2,𝒳3]=(0−abac𝐛𝟐0−ab−𝐜𝟐ac0).\begin{array}{r} \left\lbrack \mathcal{X}_{2},\mathcal{X}_{3} \right\rbrack = \begin{pmatrix} 0 & - ab & ac \\ \mathbf{b}^{\mathbf{2}} & 0 & - ab \\ \mathbf{-}\mathbf{c}^{\mathbf{2}} & ac & 0 \end{pmatrix}. \end{array}
(1.5.12)

The el­e­ments ∈𝒴2,𝒴3\in \mathcal{Y}_{2},\mathcal{Y}_{3} are trace­less, since no en­tries are present on the main di­ag­o­nal where they would typ­i­cal­ly appear as the axes of ro­ta­tion for the 𝒳j\mathcal{X}_{j} el­e­ments. To for­mal­ize this fur­ther, we can say that for the left-most el­e­ment ∈𝒳j\in \mathcal{X}_{j}, its axis of ro­ta­tion forms a prod­uct with the right-most el­e­ment ∈𝒳k\in \mathcal{X}_{k}, for its co­in­cid­ing 𝒳j\mathcal{X}_{j} row-com­po­nent of the 𝒳k\mathcal{X}_{k} el­e­ment. This prod­uct will appear in either of the re­sult­ing 𝒴2,𝒴3\mathcal{Y}_{2},\mathcal{Y}_{3}, el­e­ments in the cor­re­spond­ing 𝒳j\mathcal{X}_{j} row-com­po­nent of the column ∉{𝒳j,𝒳k}\notin \left\{ \mathcal{X}_{j},\mathcal{X}_{k} \right\}. Using the prod­ucts ex­am­ples 𝒳1∘𝒳2\mathcal{X}_{1} \circ \mathcal{X}_{2}, 𝒳1∘𝒳3\mathcal{X}_{1} \circ \mathcal{X}_{3},

𝒳1∘𝒳2=(00𝐚𝟐bc000bc0),\begin{array}{r} \mathcal{X}_{1} \circ \mathcal{X}_{2} = \begin{pmatrix} 0 & 0 & \mathbf{a}^{\mathbf{2}} \\ bc & 0 & 0 \\ 0 & bc & 0 \end{pmatrix}, \end{array}
(1.5.13)
𝒳1∘𝒳3=(0𝐚𝟐000bcbc00),\begin{array}{r} \mathcal{X}_{1} \circ \mathcal{X}_{3} = \begin{pmatrix} 0 & \mathbf{a}^{\mathbf{2}} & 0 \\ 0 & 0 & bc \\ bc & 0 & 0 \end{pmatrix}, \end{array}
(1.5.14)

we see the a2a^{2} en­tries in the first row (𝒳1\mathcal{X}_{1} com­po­nent row) in both re­sults are found in the column row of the el­e­ment ∉{𝒳j,𝒳k}\notin \left\{ \mathcal{X}_{j},\mathcal{X}_{k} \right\}. For el­e­ment prod­ucts ∈𝒳j\in \mathcal{X}_{j}, or­dered 𝒳j∘𝒳j+1\mathcal{X}_{j} \circ \mathcal{X}_{j + 1}, for jjmod 3, the result is an el­e­ment ∈𝒴3\in \mathcal{Y}_{3}, and the result for el­e­ment prod­ucts in the re­versed order 𝒳j∘𝒳j−1\mathcal{X}_{j} \circ \mathcal{X}_{j - 1}, yields an el­e­ment ∈𝒴2\in \mathcal{Y}_{2}. For any Lie brack­et with el­e­ments ∈Yj\in Y_{j}, the result will be an­oth­er el­e­ment ∈Yj\in Y_{j}, [𝒴j,𝒴k]=𝒴l\left\lbrack \mathcal{Y}_{j},\mathcal{Y}_{k} \right\rbrack = \mathcal{Y}_{l}, seen for the ex­am­ple [𝒴2,𝒴3]\left\lbrack \mathcal{Y}_{2},\mathcal{Y}_{3} \right\rbrack,

[𝒴2,𝒴3]=(de−df000ef−de000df−ef)=𝒴1({d=dee=eff=df )−𝒴1′ ({d=dfe=def=ef ).\begin{array}{r} \left\lbrack \mathcal{Y}_{2},\mathcal{Y}_{3} \right\rbrack = \begin{pmatrix} de - df & 0 & 0 \\ 0 & ef - de & 0 \\ 0 & 0 & df - ef \end{pmatrix} = \mathcal{Y}_{1}\left( \left\{ \begin{array}{r} d = de \\ e = ef \\ f = df \end{array} \right.\ \right) - \mathcal{Y}_{1}^{'}\ \left( \left\{ \begin{array}{r} d = df \\ e = de \\ f = ef \end{array} \right.\ \right). \end{array}
(1.5.15)

The [𝒴2,𝒴3]\left\lbrack \mathcal{Y}_{2},\mathcal{Y}_{3} \right\rbrack result takes the form 𝒴1−𝒴1′,\mathcal{Y}_{1} - \mathcal{Y}_{1}^{'},where the su­per­script in 𝒴1′\mathcal{Y}_{1}^{'}, de­notes the same Y1Y_{1} form, having po­ten­tial­ly dif­fer­ent entry values. For the Cartan sub­al­ge­bra cases for d=e=fd = e = f, the Lie brack­et result would be abelian and equal to zero. For the Lie brack­et [𝒴3,𝒴2]\left\lbrack \mathcal{Y}_{3},\mathcal{Y}_{2} \right\rbrack, we also find the neg­a­tive of the for­ward-ori­ent­ed Lie brack­et coun­ter­part [𝒴3,𝒴2]=−[𝒴2,𝒴3]\left\lbrack \mathcal{Y}_{3},\mathcal{Y}_{2} \right\rbrack = - \left\lbrack \mathcal{Y}_{2},\mathcal{Y}_{3} \right\rbrack, which is also abelian, im­ply­ing [𝒴3,𝒴2]+[𝒴2,𝒴3]=0.\left\lbrack \mathcal{Y}_{3},\mathcal{Y}_{2} \right\rbrack + \left\lbrack \mathcal{Y}_{2},\mathcal{Y}_{3} \right\rbrack = 0.Con­cern­ing the Lie brack­ets of el­e­ments with the same finite matrix entry lo­ca­tions [𝒴j,𝒴j′]\left\lbrack \mathcal{Y}_{j},\mathcal{Y}_{j}^{'} \right\rbrack, for 𝒴j=𝒴2\mathcal{Y}_{j} = \mathcal{Y}_{2}, and d′,e′,f′∈𝒴j′{d^{'},e}^{'},f' \in \mathcal{Y}_{j}^{'}, the result would equal

[𝒴2,𝒴2′]=(00de′−d′eef′−e′f000d′f−df′0)=𝒴3({d=de′e=ef′f=d′f )−𝒴3′ ({d′=d′ee′=e′ff′=df′ ).\begin{array}{r} \left\lbrack \mathcal{Y}_{2},\mathcal{Y}_{2}^{'} \right\rbrack = \begin{pmatrix} 0 & 0 & de^{'} - d^{'}e \\ ef^{'} - e^{'}f & 0 & 0 \\ 0 & d^{'}f - df^{'} & 0 \end{pmatrix} = \mathcal{Y}_{3}\left( \left\{ \begin{array}{r} d = de^{'} \\ e = ef^{'} \\ f = d^{'}f \end{array} \right.\ \right) - \mathcal{Y}_{3}^{'}\ \left( \left\{ \begin{array}{r} d^{'} = d^{'}e \\ e^{'} = e^{'}f \\ f^{'} = df^{'} \end{array} \right.\ \right). \end{array}
(1.5.16)

For d=d′,e=e′,f=f′d = d^{'},e = e^{'},f = f', the result above would com­mute as it does for the Cartan sub­al­ge­bra. Fi­nal­ly, for Lie brack­ets of the form [𝒳j,𝒴k]\left\lbrack \mathcal{X}_{j},\mathcal{Y}_{k} \right\rbrack, we find re­sults ∈𝒳\in \mathcal{X}, as [𝒳j,𝒴k]=−[𝒴k,𝒳j]\left\lbrack \mathcal{X}_{j},\mathcal{Y}_{k} \right\rbrack = - \left\lbrack \mathcal{Y}_{k},\mathcal{X}_{j} \right\rbrack, seen by the case ex­am­ple 𝒳j=𝒳1\mathcal{X}_{j} = \mathcal{X}_{1}, 𝒴k=𝒴2\mathcal{Y}_{k} = \mathcal{Y}_{2}

[𝒳1,𝒴2]=(0ad−bdbf−ce0−af0ce)=𝒳3({a=adb=bfc=ce )−𝒳2 ({d=bde=cef=af ).\begin{array}{r} \left\lbrack \mathcal{X}_{1},\mathcal{Y}_{2} \right\rbrack = \begin{pmatrix} 0 & ad & - bd \\ bf & - ce & 0 \\ - af & 0 & ce \end{pmatrix} = \mathcal{X}_{3}\left( \left\{ \begin{array}{r} a = ad \\ b = bf \\ c = ce \end{array} \right.\ \right) - \mathcal{X}_{2}\ \left( \left\{ \begin{array}{r} d = bd \\ e = ce \\ f = af \end{array} \right.\ \right). \end{array}
(1.5.17)

The result [𝒳1,𝒴2]=𝒳3−𝒳2\left\lbrack \mathcal{X}_{1},\mathcal{Y}_{2} \right\rbrack = \mathcal{X}_{3} - \mathcal{X}_{2}, makes it ap­par­ent that the com­po­si­tion 𝒳1∘𝒴2=𝒳3\mathcal{X}_{1} \circ \mathcal{Y}_{2} = \mathcal{X}_{3}, acts to right­ward shift the orig­i­nal el­e­ments ∈𝒳1\in \mathcal{X}_{1}, and the com­po­si­tion 𝒴2∘𝒳1=𝒳2\mathcal{Y}_{2} \circ \mathcal{X}_{1} = \mathcal{X}_{2} acts to upward shift the orig­i­nal el­e­ments ∈𝒳1\in \mathcal{X}_{1}.

  1. Ado, Igor D. – "Note on the rep­re­sen­ta­tion of finite con­tin­u­ous groups by means of linear sub­sti­tu­tions", Izv. Fiz.-Mat. Obsch. (Kazan'), 7: 1–43, 1935.
  2. *Still­well, John – "Naive Lie Theory", Springer 2008.*