Featured Post

Following the previous post on The Electromagnetic Field of Moving Charges, this is another non-trivial exercise, the derivation of 4-dimensional relativistic Lorentz transformations,  a problem of a 3rd-year undergraduate course on Special Relativity whose solution requires "tensor & matrix" manipulation techniques. At the end, there is a link to the Maple document, so that the computation below can be reproduced, and a link to a corresponding PDF file with all the sections open.

Deriving 4D relativistic Lorentz transformations

Freddy Baudine(1), Edgardo S. Cheb-Terrab(2)

(1) Retired, passionate about Mathematics and Physics

(2) Physics, Differential Equations and Mathematical Functions, Maplesoft

 

Lorentz transformations are a six-parameter family of linear transformations Lambda that relate the values of the coordinates x, y, z, t of an event in one inertial reference system to the coordinates diff(x, x), diff(y(x), x), diff(z(x), x), diff(t(x), x) of the same event in another inertial system that moves at a constant velocity relative to the former. An explicit form of Lambda can be derived from physics principles, or in a purely algebraic mathematical manner. A derivation from physics principles is done in an upcoming post about relativistic dynamics, while in this post we derive the form of Lambda mathematically, as rotations in a (pseudo) Euclidean 4 dimensional space. Most of the presentation below follows the one found in Jackson's book on Classical Electrodynamics [1].

 

The computations below in Maple 2022 make use of the Maplesoft Physics Updates v.1283 or newer.

Formulation of the problem and ansatz Lambda = exp(`𝕃`)

 

 

The problem is to find a group of linear transformations,

  "x^(' mu)=(Lambda^( mu))[nu]  x^(nu)" 

that represent rotations in a 4D (pseudo) Euclidean spacetime, and so they leave invariant the norm of the 4D position vector x^mu; that is,

"x^(' mu) (x')[mu]=x^( mu) (x^())[mu]"

For the purpose of deriving the form of `#msubsup(mi("Λ",fontstyle = "normal"),mi("ν",fontstyle = "normal"),mrow(mo("⁢"),mi("μ",fontstyle = "normal")))`, a relevant property for it can be inferred by rewriting the invariance of the norm in terms of `#msubsup(mi("Λ",fontstyle = "normal"),mi("ν",fontstyle = "normal"),mrow(mo("⁢"),mi("μ",fontstyle = "normal")))`. In steps, from the above,

"g[alpha,beta] x^(' alpha) (x^(' beta))[]=g[mu,nu] x^( mu) (x^( nu))[]"
 

g[alpha, beta]*`#msubsup(mi("Λ",fontstyle = "normal"),mi("μ",fontstyle = "normal"),mrow(mo("⁢"),mi("α",fontstyle = "normal")))`*x^mu*`#msubsup(mi("Λ",fontstyle = "normal"),mi("ν",fontstyle = "normal"),mrow(mo("⁢"),mi("β",fontstyle = "normal")))`*x^nu = g[mu, nu]*x^mu*`#msup(mi("x"),mrow(mo("⁢"),mi("ν",fontstyle = "normal")))`
 

g[alpha, beta]*`#msubsup(mi("Λ",fontstyle = "normal"),mi("μ",fontstyle = "normal"),mrow(mo("⁢"),mi("α",fontstyle = "normal")))`*x^mu*`#msubsup(mi("Λ",fontstyle = "normal"),mi("ν",fontstyle = "normal"),mrow(mo("⁢"),mi("β",fontstyle = "normal")))`*x^nu = g[mu, nu]*x^mu*`#msup(mi("x"),mrow(mo("⁢"),mi("ν",fontstyle = "normal")))`

from where,

g[alpha, beta]*`#msubsup(mi("Λ",fontstyle = "normal"),mi("μ",fontstyle = "normal"),mrow(mo("⁢"),mi("α",fontstyle = "normal")))`*`#msubsup(mi("Λ",fontstyle = "normal"),mi("ν",fontstyle = "normal"),mrow(mo("⁢"),mi("β",fontstyle = "normal")))` = g[mu, nu]``

or in matrix (4 x 4) form, `#mrow(msubsup(mi("Λ",fontstyle = "normal"),mi("μ",fontstyle = "normal"),mrow(mo("⁢"),mi("α",fontstyle = "normal"))),mo("⁢"),mo("≡"),mo("⁢"),mo("⁢"),mi("Λ",fontstyle = "normal"))`, `≡`(g[alpha, beta], g)

Lambda^T*g*Lambda = g

where Lambda^T is the transpose of Lambda. Taking the determinant of both sides of this equation, and recalling that det(Lambda^T) = det(Lambda), we get

 

det(Lambda) = `&+-`(1)

 

The determination of Lambda is analogous to the determination of the matrix R (3D tensor R[i, j]) representing rotations in the 3D space, where the same line of reasoning leads to det(R) = `&+-`(1). To exclude reflection transformations, that have det(Lambda) = -1 and cannot be obtained through any sequence of rotations, because they do not preserve the relative orientation of the axes, the sign that represents our problem is +. To explicitly construct the transformation matrix Lambda, Jackson proposes the ansatz

  Lambda = exp(`𝕃`)   

Summarizing: the determination of `#msubsup(mi("Λ",fontstyle = "normal"),mi("ν",fontstyle = "normal"),mrow(mo("⁢"),mi("μ",fontstyle = "normal")))` consists of determining `𝕃`[nu]^mu entering Lambda = exp(`𝕃`) such that det(Lambda) = 1followed by computing the exponential of the matrix `𝕃`.

Determination of `𝕃`[nu]^mu

 

In order to compare results with Jackson's book, we use the same signature he uses, "(+---)", and lowercase Latin letters to represent space tensor indices, while spacetime indices are represented using Greek letters, which is already Physics' default.

 

restart; with(Physics)

Setup(signature = "+---", spaceindices = lowercaselatin)

[signature = `+ - - -`, spaceindices = lowercaselatin]

(1)

Start by defining the tensor `𝕃`[nu]^mu whose components are to be determined. For practical purposes, define a macro LM = `𝕃` to represent the tensor and use L to represent its components

macro(LM = `𝕃`, %LM = `%𝕃`); Define(Lambda, LM, quiet)

LM[`~mu`, nu] = Matrix(4, symbol = L)

`𝕃`[`~mu`, nu] = Matrix(%id = 36893488153289603060)

(2)

"Define(?)"

{Lambda, `𝕃`[`~mu`, nu], Physics:-Dgamma[mu], Physics:-Psigma[mu], Physics:-d_[mu], Physics:-g_[mu, nu], Physics:-gamma_[a, b], Physics:-LeviCivita[alpha, beta, mu, nu]}

(3)

Next, from Lambda^T*g*Lambda = g (see above in Formulation of the problem) one can derive the form of `𝕃`. To work algebraically with `𝕃`, Lambda, g representing matrices, set these symbols as noncommutative

Setup(noncommutativeprefix = {LM, Lambda, g})

[noncommutativeprefix = {`𝕃`, Lambda, g}]

(4)

From

Lambda^T*g*Lambda = g

Physics:-`*`(Physics:-`^`(Lambda, T), g, Lambda) = g

(5)

it follows that

(1/g*(Physics[`*`](Physics[`^`](Lambda, T), g, Lambda) = g))/Lambda

Physics:-`*`(Physics:-`^`(g, -1), Physics:-`^`(Lambda, T), g) = Physics:-`^`(Lambda, -1)

(6)

eval(Physics[`*`](Physics[`^`](g, -1), Physics[`^`](Lambda, T), g) = Physics[`^`](Lambda, -1), Lambda = exp(LM))

Physics:-`*`(Physics:-`^`(g, -1), Physics:-`^`(exp(`𝕃`), T), g) = Physics:-`^`(exp(`𝕃`), -1)

(7)

Expanding the exponential using exp(`𝕃`) = Sum(`𝕃`^k/factorial(k), k = 0 .. infinity), and taking into account that the matrix product `𝕃`^k/g*g can be rewritten as(`𝕃`/g*g)^k, the left-hand side of (7) can be written as exp(`𝕃`^T/g*g)

exp(LM^T/g*g) = rhs(Physics[`*`](Physics[`^`](g, -1), Physics[`^`](exp(`𝕃`), T), g) = Physics[`^`](exp(`𝕃`), -1))

exp(Physics:-`*`(Physics:-`^`(g, -1), Physics:-`^`(`𝕃`, T), g)) = Physics:-`^`(exp(`𝕃`), -1)

(8)

Multiplying by exp(`𝕃`)

(exp(Physics[`*`](Physics[`^`](g, -1), Physics[`^`](`𝕃`, T), g)) = Physics[`^`](exp(`𝕃`), -1))*exp(LM)

Physics:-`*`(exp(Physics:-`*`(Physics:-`^`(g, -1), Physics:-`^`(`𝕃`, T), g)), exp(`𝕃`)) = 1

(9)

Recalling that  "g^(-1)=g[]^(mu,alpha)", g = g[beta, nu] and that for any matrix `𝕃`, "(`𝕃`^T)[alpha]^(   beta)= `𝕃`(( )^(beta))[alpha]",  

"g^(-1) `𝕃`^T g= 'g_[~mu,~alpha]*LM[~beta, alpha] g_[beta, nu] '"

Physics:-`*`(Physics:-`^`(g, -1), Physics:-`^`(`𝕃`, T), g) = Physics:-`*`(Physics:-g_[`~mu`, `~alpha`], `𝕃`[`~beta`, alpha], Physics:-g_[beta, nu])

(10)

subs([Physics[`*`](Physics[`^`](g, -1), Physics[`^`](`𝕃`, T), g) = Physics[`*`](Physics[g_][`~mu`, `~alpha`], `𝕃`[`~beta`, alpha], Physics[g_][beta, nu]), LM = LM[`~mu`, nu]], Physics[`*`](exp(Physics[`*`](Physics[`^`](g, -1), Physics[`^`](`𝕃`, T), g)), exp(`𝕃`)) = 1)

Physics:-`*`(exp(Physics:-g_[`~alpha`, `~mu`]*Physics:-g_[beta, nu]*`𝕃`[`~beta`, alpha]), exp(`𝕃`[`~mu`, nu])) = 1

(11)

To allow for the combination of the exponentials, now that everything is in tensor notation, remove the noncommutative character of `𝕃```

Setup(clear, noncommutativeprefix)

[noncommutativeprefix = none]

(12)

combine(Physics[`*`](exp(Physics[g_][`~alpha`, `~mu`]*Physics[g_][beta, nu]*`𝕃`[`~beta`, alpha]), exp(`𝕃`[`~mu`, nu])) = 1)

exp(`𝕃`[`~beta`, alpha]*Physics:-g_[beta, nu]*Physics:-g_[`~alpha`, `~mu`]+`𝕃`[`~mu`, nu]) = 1

(13)

Since every tensor component of this expression is real, taking the logarithm at both sides and simplifying tensor indices

`assuming`([map(ln, exp(`𝕃`[`~beta`, alpha]*Physics[g_][beta, nu]*Physics[g_][`~alpha`, `~mu`]+`𝕃`[`~mu`, nu]) = 1)], [real])

`𝕃`[`~beta`, alpha]*Physics:-g_[beta, nu]*Physics:-g_[`~alpha`, `~mu`]+`𝕃`[`~mu`, nu] = 0

(14)

Simplify(`𝕃`[`~beta`, alpha]*Physics[g_][beta, nu]*Physics[g_][`~alpha`, `~mu`]+`𝕃`[`~mu`, nu] = 0)

`𝕃`[nu, `~mu`]+`𝕃`[`~mu`, nu] = 0

(15)

So the components of `𝕃`[`~mu`, nu]

LM[`~μ`, nu, matrix]

`𝕃`[`~μ`, nu] = Matrix(%id = 36893488151939882148)

(16)

satisfy (15). Using TensorArray  the components of that tensorial equation are

TensorArray(`𝕃`[nu, `~mu`]+`𝕃`[`~mu`, nu] = 0, output = setofequations)

{2*L[1, 1] = 0, 2*L[2, 2] = 0, 2*L[3, 3] = 0, 2*L[4, 4] = 0, -L[1, 2]+L[2, 1] = 0, L[1, 2]-L[2, 1] = 0, -L[1, 3]+L[3, 1] = 0, L[1, 3]-L[3, 1] = 0, -L[1, 4]+L[4, 1] = 0, L[1, 4]-L[4, 1] = 0, L[3, 2]+L[2, 3] = 0, L[4, 2]+L[2, 4] = 0, L[4, 3]+L[3, 4] = 0}

(17)

Simplifying taking these equations into account results in the form of `𝕃`[`~mu`, nu] we were looking for

"simplify(?,{2*L[1,1] = 0, 2*L[2,2] = 0, 2*L[3,3] = 0, 2*L[4,4] = 0, -L[1,2]+L[2,1] = 0, L[1,2]-L[2,1] = 0, -L[1,3]+L[3,1] = 0, L[1,3]-L[3,1] = 0, -L[1,4]+L[4,1] = 0, L[1,4]-L[4,1] = 0, L[3,2]+L[2,3] = 0, L[4,2]+L[2,4] = 0, L[4,3]+L[3,4] = 0})"

`𝕃`[`~μ`, nu] = Matrix(%id = 36893488153606736460)

(18)

This is equation (11.90) in Jackson's book [1]. By eye we see there are only six independent parameters in `𝕃`[`~mu`, nu], or via

"indets(rhs(?), name)"

{L[1, 2], L[1, 3], L[1, 4], L[2, 3], L[2, 4], L[3, 4]}

(19)

nops({L[1, 2], L[1, 3], L[1, 4], L[2, 3], L[2, 4], L[3, 4]})

6

(20)

This number is expected: a rotation in 3D space can always be represented as the composition of three rotations, and so, characterized by 3 parameters: the rotation angles measured on each of the space planes x, y, y, z, z, x. Likewise, a rotation in 4D space is characterized by 6 parameters: rotations on each of the three space planes, parameters L[2, 3], L[2, 4] and L[3, 4],  and rotations on the spacetime planest, x, t, y, t, z, parameters L[1, j]. Define now `𝕃`[`~mu`, nu] using (18) for further computing with it in the next section

"Define(?)"

{Lambda, `𝕃`[`~mu`, nu], Physics:-Dgamma[mu], Physics:-Psigma[mu], Physics:-d_[mu], Physics:-g_[mu, nu], Physics:-gamma_[a, b], Physics:-LeviCivita[alpha, beta, mu, nu]}

(21)

Determination of Lambda[`~mu`, nu]

 

From the components of `𝕃`[`~mu`, nu] in (18), the components of Lambda[`~mu`, nu] = exp(`𝕃`[`~mu`, nu]) can be computed directly using the LinearAlgebra:-MatrixExponential command. Then, following Jackson's book, in what follows we also derive a general formula for `𝕃`[`~mu`, nu]in terms of beta = v/c and gamma = 1/sqrt(-beta^2+1) shown in [1] as equation (11.98), finally showing the form of Lambda[`~mu`, nu] as a function of the relative velocity of the two inertial systems of references.

 

An explicit form of Lambda[`~mu`, nu] in the case of a rotation on thet, x plane can be computed by taking equal to zero all the parameters in (19) but for L[1, 2] and substituting in "?≡`𝕃`[nu]^(mu)"  

`~`[`=`](`minus`({L[1, 2], L[1, 3], L[1, 4], L[2, 3], L[2, 4], L[3, 4]}, {L[1, 2]}), 0)

{L[1, 3] = 0, L[1, 4] = 0, L[2, 3] = 0, L[2, 4] = 0, L[3, 4] = 0}

(22)

"subs({L[1,3] = 0, L[1,4] = 0, L[2,3] = 0, L[2,4] = 0, L[3,4] = 0},?)"

`𝕃`[`~μ`, nu] = Matrix(%id = 36893488153606695500)

(23)

Computing the matrix exponential,

"Lambda[~mu,nu]=LinearAlgebra:-MatrixExponential(rhs(?))"

Lambda[`~μ`, nu] = Matrix(%id = 36893488151918824492)

(24)

"convert(?,trigh)"

Lambda[`~μ`, nu] = Matrix(%id = 36893488151918852684)

(25)

This is formula (4.2) in Landau & Lifshitz book [2]. An explicit form of Lambda[`~mu`, nu] in the case of a rotation on thex, y plane can be computed by taking equal to zero all the parameters in (19) but for L[2, 3]

`~`[`=`](`minus`({L[1, 2], L[1, 3], L[1, 4], L[2, 3], L[2, 4], L[3, 4]}, {L[2, 3]}), 0)

{L[1, 2] = 0, L[1, 3] = 0, L[1, 4] = 0, L[2, 4] = 0, L[3, 4] = 0}

(26)

"subs({L[1,2] = 0, L[1,3] = 0, L[1,4] = 0, L[2,4] = 0, L[3,4] = 0},?)"

`𝕃`[`~μ`, nu] = Matrix(%id = 36893488151918868828)

(27)

"Lambda[~mu, nu]=LinearAlgebra:-MatrixExponential(rhs(?))"

Lambda[`~μ`, nu] = Matrix(%id = 36893488153289306948)

(28)

NULL

Rewriting `%𝕃`[`~mu`, nu] = K[`~i`]*Zeta[i]+S[`~i`]*omega[i]

 

Following Jackson's notation, for readability, redefine the 6 parameters entering `𝕃`[`~mu`, nu] as

'{LM[1, 2] = `ζ__1`, LM[1, 3] = `ζ__2`, LM[1, 4] = `ζ__3`, LM[2, 3] = `ω__3`, LM[2, 4] = -`ω__2`, LM[3, 4] = `ω__1`}'

{`𝕃`[1, 2] = zeta__1, `𝕃`[1, 3] = zeta__2, `𝕃`[1, 4] = zeta__3, `𝕃`[2, 3] = omega__3, `𝕃`[2, 4] = -omega__2, `𝕃`[3, 4] = omega__1}

(29)

(Note in the above the surrounding backquotes '...' to prevent a premature evaluation of the left-hand sides; that is necessary when using the Library:-RedefineTensorComponent command.) With this redefinition, `𝕃`[`~mu`, nu] becomes

Library:-RedefineTensorComponent({`𝕃`[1, 2] = zeta__1, `𝕃`[1, 3] = zeta__2, `𝕃`[1, 4] = zeta__3, `𝕃`[2, 3] = omega__3, `𝕃`[2, 4] = -omega__2, `𝕃`[3, 4] = omega__1})

LM[`~μ`, nu, matrix]

`𝕃`[`~μ`, nu] = Matrix(%id = 36893488151939901668)

(30)

where each parameter is related to a rotation angle on one plane. Any Lorentz transformation (rotation in 4D pseudo-Euclidean space) can be represented as the composition of these six rotations, and to each rotation, corresponds the matrix that results from taking equal to zero all of the six parameters but one.

 

The set of six parameters can be split into two sets of three parameters each, one representing rotations on the t, x__j planes, parameters `ζ__j`, and the other representing rotations on the x__i, x__j planes, parameters `ω__j`. With that, following [1], (30) can be rewritten in terms of four 3D tensors, two of them with the parameters as components, the other two with matrix as components, as follows:

Zeta[i] = [`ζ__1`, `ζ__2`, `ζ__3`], omega[i] = [`ω__1`, `ω__2`, `ω__3`], K[i] = [K__1, K__2, K__3], S[i] = [S__1, S__2, S__3]

Zeta[i] = [zeta__1, zeta__2, zeta__3], omega[i] = [omega__1, omega__2, omega__3], K[i] = [K__1, K__2, K__3], S[i] = [S__1, S__2, S__3]

(31)

Define(Zeta[i] = [zeta__1, zeta__2, zeta__3], omega[i] = [omega__1, omega__2, omega__3], K[i] = [K__1, K__2, K__3], S[i] = [S__1, S__2, S__3])

{Lambda, `𝕃`[mu, nu], Physics:-Dgamma[mu], K[i], Physics:-Psigma[mu], S[i], Zeta[i], Physics:-d_[mu], Physics:-g_[mu, nu], Physics:-gamma_[a, b], omega[i], Physics:-LeviCivita[alpha, beta, mu, nu]}

(32)

The 3D tensors K[i] and S[i] satisfy the commutation relations

Setup(noncommutativeprefix = {K, S})

[noncommutativeprefix = {K, S}]

(33)

Commutator(S[i], S[j]) = LeviCivita[i, j, k]*S[k]

Physics:-Commutator(S[i], S[j]) = Physics:-LeviCivita[i, j, k]*S[`~k`]

(34)

Commutator(S[i], K[j]) = LeviCivita[i, j, k]*K[k]

Physics:-Commutator(S[i], K[j]) = Physics:-LeviCivita[i, j, k]*K[`~k`]

(35)

Commutator(K[i], K[j]) = -LeviCivita[i, j, k]*S[k]

Physics:-Commutator(K[i], K[j]) = -Physics:-LeviCivita[i, j, k]*S[`~k`]

(36)

The matrix components of the 3D tensor K__i, related to rotations on the t, x__j planes, are

K__1 := matrix([[0, 1, 0, 0], [1, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, 0]])

array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (1), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (1), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] )

(37)

K__2 := matrix([[0, 0, 1, 0], [0, 0, 0, 0], [1, 0, 0, 0], [0, 0, 0, 0]])

array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (1), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (1), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] )

(38)

K__3 := matrix([[0, 0, 0, 1], [0, 0, 0, 0], [0, 0, 0, 0], [1, 0, 0, 0]])

array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (1), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (1), ( 3, 4 ) = (0)  ] )

(39)

The matrix components of the 3D tensor S__i, related to rotations on the x__i, x__j 3D space planes, are

S__1 := matrix([[0, 0, 0, 0], [0, 0, 0, 0], [0, 0, 0, -1], [0, 0, 1, 0]])

array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (1), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (-1)  ] )

(40)

S__2 := matrix([[0, 0, 0, 0], [0, 0, 0, 1], [0, 0, 0, 0], [0, -1, 0, 0]])

array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (-1), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (1), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] )

(41)

S__3 := matrix([[0, 0, 0, 0], [0, 0, -1, 0], [0, 1, 0, 0], [0, 0, 0, 0]])

array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (1), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (-1), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] )

(42)

NULL

Verifying the commutation relations between S[i] and K[j]

   

The `𝕃`[`~mu`, nu] tensor is now expressed in terms of these objects as

%LM[`~μ`, nu] = omega[i].S[i]+Zeta[i].K[i]

`%𝕃`[`~μ`, nu] = K[`~i`]*Zeta[i]+S[`~i`]*omega[i]

(50)

where the right-hand side, without free indices, represents the matrix form of `%𝕃`[`~mu`, nu]. This notation makes explicit the fact that any Lorentz transformation can always be written as the composition of six rotations

SumOverRepeatedIndices(`%𝕃`[`~μ`, nu] = K[`~i`]*Zeta[i]+S[`~i`]*omega[i])

`%𝕃`[`~μ`, nu] = zeta__1*K[`~1`]+zeta__2*K[`~2`]+zeta__3*K[`~3`]+omega__1*S[`~1`]+omega__2*S[`~2`]+omega__3*S[`~3`]

(51)

Library:-RewriteInMatrixForm(`%𝕃`[`~μ`, nu] = zeta__1*K[`~1`]+zeta__2*K[`~2`]+zeta__3*K[`~3`]+omega__1*S[`~1`]+omega__2*S[`~2`]+omega__3*S[`~3`])

`%𝕃`[`~μ`, nu] = (array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (zeta__1), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (zeta__1), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] ))+(array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (zeta__2), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (zeta__2), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] ))+(array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (zeta__3), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (zeta__3), ( 3, 4 ) = (0)  ] ))+(array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (omega__1), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (-omega__1)  ] ))+(array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (-omega__2), ( 1, 2 ) = (0), ( 3, 2 ) = (0), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (omega__2), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] ))+(array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (0), ( 4, 2 ) = (0), ( 1, 2 ) = (0), ( 3, 2 ) = (omega__3), ( 1, 3 ) = (0), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (0), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (0), ( 2, 2 ) = (0), ( 2, 3 ) = (-omega__3), ( 4, 1 ) = (0), ( 3, 4 ) = (0)  ] ))

(52)

Library:-PerformMatrixOperations(`%𝕃`[`~μ`, nu] = zeta__1*K[`~1`]+zeta__2*K[`~2`]+zeta__3*K[`~3`]+omega__1*S[`~1`]+omega__2*S[`~2`]+omega__3*S[`~3`])

`%𝕃`[`~μ`, nu] = (array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (zeta__2), ( 4, 2 ) = (-omega__2), ( 1, 2 ) = (zeta__1), ( 3, 2 ) = (omega__3), ( 1, 3 ) = (zeta__2), ( 4, 3 ) = (omega__1), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (zeta__1), ( 3, 3 ) = (0), ( 2, 4 ) = (omega__2), ( 1, 4 ) = (zeta__3), ( 2, 2 ) = (0), ( 2, 3 ) = (-omega__3), ( 4, 1 ) = (zeta__3), ( 3, 4 ) = (-omega__1)  ] ))

(53)

NULL

which is the same as the starting point (30)NULL

The transformation Lambda[`~mu`, nu] = exp(`%𝕃`[`~mu`, nu]), where  `%𝕃`[`~mu`, nu] = K[`~i`]*Zeta[i], as a function of the relative velocity of two inertial systems

 

 

As seen in the previous subsection, in `𝕃`[`~mu`, nu] = K[`~i`]*Zeta[i]+S[`~i`]*omega[i], the second term, S[`~i`]*omega[i], corresponds to 3D rotations embedded in the general form of 4D Lorentz transformations, and K[`~i`]*Zeta[i] is the term that relates the coordinates of two inertial systems of reference that move with respect to each other at constant velocity v.  In this section, K[`~i`]*Zeta[i] is rewritten in terms of that velocity, arriving at equation (11.98)  of Jackson's book [1]. The key observation is that the 3D vector Zeta[i], can be rewritten in terms of arctanh(beta), where beta = v/c and c is the velocity of light (for the rationale of that relation, see [2], sec 4, discussion before formula (4.3)).

 

Use a macro - say ub - to represent the atomic variable `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))` (this variable can be entered as `#mover(mi("β"),mo("ˆ")`. In general, to create atomic variables, see the section on Atomic Variables of the page 2DMathDetails ).

 

macro(ub = `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`)

ub[j] = [ub[1], ub[2], ub[3]], Zeta[j] = ub[j]*arctanh(beta)

`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j] = [`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]], Zeta[j] = `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j]*arctanh(beta)

(54)

Define(`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j] = [`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]], Zeta[j] = `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j]*arctanh(beta))

{Lambda, `𝕃`[mu, nu], Physics:-Dgamma[mu], K[i], Physics:-Psigma[mu], S[i], Zeta[i], Physics:-d_[mu], Physics:-g_[mu, nu], Physics:-gamma_[a, b], omega[i], Physics:-LeviCivita[alpha, beta, mu, nu], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j]}

(55)

With these two definitions, and excluding the rotation term S[`~i`]*omega[i] we have

%LM[`~μ`, nu] = Zeta[j]*K[j]

`%𝕃`[`~μ`, nu] = Zeta[j]*K[`~j`]

(56)

SumOverRepeatedIndices(`%𝕃`[`~μ`, nu] = Zeta[j]*K[`~j`])

`%𝕃`[`~μ`, nu] = arctanh(beta)*(K[`~1`]*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]+K[`~2`]*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]+K[`~3`]*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3])

(57)

Library:-PerformMatrixOperations(`%𝕃`[`~μ`, nu] = arctanh(beta)*(K[`~1`]*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]+K[`~2`]*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]+K[`~3`]*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]))

`%𝕃`[`~μ`, nu] = (array( 1 .. 4, 1 .. 4, [( 3, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]*arctanh(beta)), ( 4, 2 ) = (0), ( 1, 2 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]*arctanh(beta)), ( 3, 2 ) = (0), ( 1, 3 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]*arctanh(beta)), ( 4, 3 ) = (0), ( 4, 4 ) = (0), ( 1, 1 ) = (0), ( 2, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]*arctanh(beta)), ( 3, 3 ) = (0), ( 2, 4 ) = (0), ( 1, 4 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]*arctanh(beta)), ( 2, 2 ) = (0), ( 2, 3 ) = (0), ( 4, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]*arctanh(beta)), ( 3, 4 ) = (0)  ] ))

(58)

 

From this expression, the form of "Lambda[nu]^(mu)" can be obtained as in (24) using LinearAlgebra:-MatrixExponential and simplifying the result taking into account that `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j] is a unit vector

SumOverRepeatedIndices(ub[j]^2) = 1

`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2 = 1

(59)

exp(lhs(`%𝕃`[`~μ`, nu] = (array( 1 .. 4, 1 .. 4, [( 3, 3 ) = (0), ( 2, 3 ) = (0), ( 4, 2 ) = (0), ( 1, 1 ) = (0), ( 1, 2 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]*arctanh(beta)), ( 4, 4 ) = (0), ( 4, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]*arctanh(beta)), ( 3, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]*arctanh(beta)), ( 3, 4 ) = (0), ( 4, 3 ) = (0), ( 1, 4 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]*arctanh(beta)), ( 3, 2 ) = (0), ( 1, 3 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]*arctanh(beta)), ( 2, 4 ) = (0), ( 2, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]*arctanh(beta)), ( 2, 2 ) = (0)  ] )))) = simplify(LinearAlgebra:-MatrixExponential(rhs(`%𝕃`[`~μ`, nu] = (array( 1 .. 4, 1 .. 4, [( 3, 3 ) = (0), ( 2, 3 ) = (0), ( 4, 2 ) = (0), ( 1, 1 ) = (0), ( 1, 2 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]*arctanh(beta)), ( 4, 4 ) = (0), ( 4, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]*arctanh(beta)), ( 3, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]*arctanh(beta)), ( 3, 4 ) = (0), ( 4, 3 ) = (0), ( 1, 4 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]*arctanh(beta)), ( 3, 2 ) = (0), ( 1, 3 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]*arctanh(beta)), ( 2, 4 ) = (0), ( 2, 1 ) = (`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]*arctanh(beta)), ( 2, 2 ) = (0)  ] )))), {`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2 = 1})

exp(`%𝕃`[`~μ`, nu]) = Matrix(%id = 36893488153234621252)

(60)

It is useful at this point to analyze the dependency on the components of `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j] of this matrix

"map(u -> indets(u,specindex(ub)), rhs(?))"

Matrix(%id = 36893488151918822812)

(61)

We see that the diagonal element [4, 4] depends on two instead of only one component of  `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j]. That is due to the simplification with respect to side relations , performed in (60), that constructs an elimination Groebner Basis that cannot reduce at once, using the single equation (59), the dependency of all of the elements [2, 2], [3, 3] and [4, 4] to a single component of  `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j]. So, to reduce further the dependency of the [4, 4] element, this component of (60) requires one more simplification step, using a different elimination strategy, explicitly requesting the elimination of "{(beta)[1],(beta)[2]}"

"rhs(?)[4,4]"

((`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2)*(-beta^2+1)^(1/2)-`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2-`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2+1)/(-beta^2+1)^(1/2)

(62)

 

simplify(((`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2)*(-beta^2+1)^(1/2)-`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2-`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2+1)/(-beta^2+1)^(1/2), {`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2 = 1}, {`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]})

(-(-beta^2+1)^(1/2)*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2+(-beta^2+1)^(1/2))/(-beta^2+1)^(1/2)

(63)

This result involves only `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3], and with it the form of Lambda[`~mu`, nu] = exp(`%𝕃`[`~mu`, nu]) becomes

"subs(1/(-beta^2+1)^(1/2)*((`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2)*(-beta^2+1)^(1/2)-`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1]^2-`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2]^2+1) = (-(-beta^2+1)^(1/2)*`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2+`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3]^2+(-beta^2+1)^(1/2))/(-beta^2+1)^(1/2),?)"

exp(`%𝕃`[`~μ`, nu]) = Matrix(%id = 36893488151918876660)

(64)

Replacing now the components of the unit vector `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j] by the components of the vector `#mover(mi("β",fontstyle = "normal"),mo("→"))` divided by its modulus beta

seq(ub[j] = beta[j]/beta, j = 1 .. 3)

`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1] = beta[1]/beta, `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[2] = beta[2]/beta, `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[3] = beta[3]/beta

(65)

and recalling that

exp(`%𝕃`[`~μ`, nu]) = Lambda[`~μ`, nu]

exp(`%𝕃`[`~μ`, nu]) = Lambda[`~μ`, nu]

(66)

to get equation (11.98) in Jackson's book it suffices to introduce (the customary notation)

1/sqrt(-beta^2+1) = gamma

1/(-beta^2+1)^(1/2) = gamma

(67)

"simplify(subs(`#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[1] = beta[1]/beta, exp(`%𝕃`[`~μ`,nu]) = Lambda[`~μ`,nu], 1/(-beta^2+1)^(1/2) = gamma,(1/(-beta^2+1)^(1/2) = gamma)^(-1),?))"

Lambda[`~μ`, nu] = Matrix(%id = 36893488151911556148)

(68)

 

This is equation (11.98) in Jackson's book.

 

Finally, to get the form of this general Lorentz transformation excluding 3D rotations, directly expressed in terms of the relative velocity v of the two inertial systems of references, introduce

v[i] = [v__x, v__y, v__z], beta[i] = v[i]/c

v[i] = [v__x, v__y, v__z], beta[i] = v[i]/c

(69)

At this point it suffices to Define (69) as tensors

Define(v[i] = [v__x, v__y, v__z], beta[i] = v[i]/c)

{Lambda, `𝕃`[mu, nu], Physics:-Dgamma[mu], K[i], Physics:-Psigma[mu], S[i], Zeta[i], beta[i], Physics:-d_[mu], Physics:-g_[mu, nu], Physics:-gamma_[a, b], omega[i], v[i], Physics:-LeviCivita[alpha, beta, mu, nu], `#mover(mi("β",fontstyle = "normal"),mo("ˆ"))`[j]}

(70)

and remove beta and gamma from the formulation using

(rhs = lhs)(1/(-beta^2+1)^(1/2) = gamma), beta = v/c

gamma = 1/(-beta^2+1)^(1/2), beta = v/c

(71)

"simplify(subs(gamma = 1/(-beta^2+1)^(1/2),simplify(?)),size) "

Lambda[`~μ`, nu] = Matrix(%id = 36893488153289646316)

(72)

NULL

``

NULL

References

 

[1] J.D. Jackson, "Classical Electrodynamics", third edition, 1999.

[2] L.D. Landau, E.M. Lifshitz, "The Classical Theory of Fields", Course of Theoretical Physics V.2, 4th revised English edition, 1975.

NULL

Download Deriving_the_mathematical_form_of_Lorentz_transformations.mw

Deriving_the_mathematical_form_of_Lorentz_transformations.pdf

Edgardo S. Cheb-Terrab
Physics, Differential Equations and Mathematical Functions, Maplesoft

Featured Post

We've just released Maple Flow 2022.1. We've squeezed in a few new features as requested by our users - I'll describe them below.

Before we get to that, I'd like to give everyone an open invitation to grab a Maple Flow trial - I'd love to know what you think. I'm fanatically devoted to making Flow better, but I can only do that if you give me your feedback.

You can specify if you want your results to be globally displayed using engineering, scientific, or fixed notation

Supporting images can be cut and pasted from another source directly into Maple Flow using standard clipboard operations.

You can now insert a time stamp in headers and footers. And you can optionally place a border around the header, footer or body of the page.

New content in the help system makes it easier to get started with advanced features, including techniques for optimization and signal processing.

Go here to learn more...and don't forget to grab a trial.

 



Read binary file

Maple asked by Bruno76 5 Yesterday

ellipse in a rectangle

Maple 2021 asked by JAMET 355 Yesterday

How to solve this integral?

Maple 2020 asked by WA573 10 August 15