TOPICS
Search

Christoffel Symbol of the Second Kind


Christoffel symbols of the second kind are the coefficients of the Levi-Civita connection of a Riemannian metric g in a chosen basis. They are variously denoted as {m; i  j} (Walton 1967) or Gamma^m_(ij) (Misner et al. 1973, Arfken 1985). They are also called connection coefficients (Misner et al. 1973, p. 210) or affine connection coefficients (Weinberg 1972, p. 71).

The coordinate-basis and noncoordinate-basis formulas below give coefficients of the same Levi-Civita connection in different bases. In a coordinate basis, the coefficients can be expressed in terms of the metric tensor alone.

For a coordinate basis, Arfken (1985, p. 161) gives

Gamma^m_(ij)=epsilon^m·(partialepsilon_i)/(partialq^j)
(1)
=g^(km)[ij,k]
(2)
=1/2g^(km)((partialg_(ik))/(partialq^j)+(partialg_(jk))/(partialq^i)-(partialg_(ij))/(partialq^k)),
(3)

where partial/partialq^j is a partial derivative, g_(ij) is the metric tensor, and g^(km) is the corresponding inverse metric,

 epsilon_i=(partialr)/(partialq^i),
(4)

where r is the radius vector, and

 epsilon^i=g^(ij)epsilon_j.
(5)

Therefore, for an orthogonal curvilinear coordinate system, by this definition,

 Gamma^m_(ij)=1/(g_(mm))epsilon_m·(partial^2r)/(partialq^jpartialq^i).
(6)

Because the coordinate basis vectors commute and the Levi-Civita connection is torsion-free,

 Gamma^k_(ij)=Gamma^k_(ji)
(7)

(Walton 1967).

This Christoffel symbol of the second kind is related to the Christoffel symbol of the first kind [bc,d] by

 Gamma^a_(bc)=g^(ad)[bc,d].
(8)

Walton (1967) lists Christoffel symbols of the second kind for the 12 basic orthogonal coordinate systems.

For an orthonormal basis u_i^^ that need not be a coordinate basis, Misner et al. (1973, p. 209) give the coefficients as

 Gamma^k_(ij)=u_k^^·(del _ju_i^^),
(9)

where del _j denotes the covariant derivative in the direction of u_j^^. These coefficients need not be symmetric in i and j because the basis vectors need not commute.

Christoffel symbols of the second kind are not tensors and do not transform as tensor components. For two coordinate bases related by a change from x_1,...,x_n to y_1,...,y_n, the transformation law is

 Gamma^k_(ij)^'=sum(partial^2x_l)/(partialy_ipartialy_j)(partialy_k)/(partialx_l)+sumGamma^t_(rs)(partialx_r)/(partialy_i)(partialx_s)/(partialy_j)(partialy_k)/(partialx_t).
(10)

Here the lowered-index convention is Gamma_(alphabetagamma)=g_(gammamu)Gamma^mu_(alphabeta). For the Levi-Civita connection in a general basis, these coefficients are

 Gamma_(alphabetagamma)=1/2(g_(alphagamma,beta)+g_(betagamma,alpha)-g_(alphabeta,gamma)+c_(gammaalphabeta)+c_(gammabetaalpha)-c_(alphabetagamma)),
(11)

(Misner et al. 1973, p. 210), where the gs are components of the metric tensor, c_(alphabetagamma)=g_(gammamu)c_(alphabeta)^mu for the commutation coefficients [e_alpha,e_beta]=c_(alphabeta)^mue_mu, and the commas indicate the comma derivative. In a coordinate basis, the commutation coefficients vanish and this reduces to a Christoffel symbol of the first kind. In an orthonormal basis, the metric components are constant, so g_(alphabeta,gamma)=0 and g_(mugamma)=delta_(mugamma). Hence

 Gamma_(alphabetagamma)=Gamma^mu_(alphabeta)g_(mugamma)=Gamma^gamma_(alphabeta)=1/2(c_(gammaalphabeta)+c_(gammabetaalpha)-c_(alphabetagamma)).
(12)

For an orthogonal coordinate basis, in contrast, the metric is diagonal but its diagonal components need not be constant. In this setting,

Gamma_(ijk)=0    for i!=j, j!=k, i!=k
(13)
Gamma_(iik)=-1/2(partialg_(ii))/(partialx^k)    for i!=k
(14)
Gamma_(iji)=Gamma_(jii)=1/2(partialg_(ii))/(partialx^j)
(15)
Gamma^k_(ij)=0    for i!=j, j!=k, i!=k
(16)
Gamma^k_(ii)=-1/(2g_(kk))(partialg_(ii))/(partialx^k)    for i!=k
(17)
Gamma^i_(ij)=Gamma^i_(ji)=1/(2g_(ii))(partialg_(ii))/(partialx^j)=1/2(partiallng_(ii))/(partialx^j).
(18)

In three dimensions, the Christoffel symbols of the second kind may be arranged in matrix form:

 Gamma^l=[Gamma^l_(ii) Gamma^l_(ij) Gamma^l_(ik); Gamma^l_(ji) Gamma^l_(jj) Gamma^l_(jk); Gamma^l_(ki) Gamma^l_(kj) Gamma^l_(kk)].
(19)

The Christoffel symbols are given in terms of the coefficients of the first fundamental form E, F, and G by

Gamma^1_(11)=(GE_u-2FF_u+FE_v)/(2(EG-F^2))
(20)
Gamma^1_(12)=(GE_v-FG_u)/(2(EG-F^2))
(21)
Gamma^1_(22)=(2GF_v-GG_u-FG_v)/(2(EG-F^2))
(22)
Gamma^2_(11)=(2EF_u-EE_v-FE_u)/(2(EG-F^2))
(23)
Gamma^2_(12)=(EG_u-FE_v)/(2(EG-F^2))
(24)
Gamma^2_(22)=(EG_v-2FF_v+FG_u)/(2(EG-F^2)),
(25)

and Gamma^1_(21)=Gamma^1_(12) and Gamma^2_(21)=Gamma^2_(12). If F=0, the Christoffel symbols of the second kind simplify to

Gamma^1_(11)=(E_u)/(2E)
(26)
Gamma^1_(12)=(E_v)/(2E)
(27)
Gamma^1_(22)=-(G_u)/(2E)
(28)
Gamma^2_(11)=-(E_v)/(2G)
(29)
Gamma^2_(12)=(G_u)/(2G)
(30)
Gamma^2_(22)=(G_v)/(2G)
(31)

(Gray 1997).

The following relationships hold between the Christoffel symbols of the second kind and coefficients of the first fundamental form,

Gamma^1_(11)E+Gamma^2_(11)F=1/2E_u
(32)
Gamma^1_(12)E+Gamma^2_(12)F=1/2E_v
(33)
Gamma^1_(22)E+Gamma^2_(22)F=F_v-1/2G_u
(34)
Gamma^1_(11)F+Gamma^2_(11)G=F_u-1/2E_v
(35)
Gamma^1_(12)F+Gamma^2_(12)G=1/2G_u
(36)
Gamma^1_(22)F+Gamma^2_(22)G=1/2G_v
(37)
Gamma^1_(11)+Gamma^2_(12)=(lnsqrt(EG-F^2))_u
(38)
Gamma^1_(12)+Gamma^2_(22)=(lnsqrt(EG-F^2))_v
(39)

(Gray 1997).

For a surface given in Monge's form z=F(x,y),

 Gamma^k_(ij)=(z_(ij)z_k)/(1+z_1^2+z_2^2).
(40)

Christoffel symbols of the second kind arise in the computation of geodesics. The geodesic equation of free motion is

 dtau^2=-eta_(alphabeta)dxi^alphadxi^beta,
(41)

or

 (d^2xi^alpha)/(dtau^2)=0.
(42)

Expanding,

 d/(dtau)((partialxi^alpha)/(partialx^mu)(dx^mu)/(dtau))=(partialxi^alpha)/(partialx^mu)(d^2x^mu)/(dtau^2)+(partial^2xi^alpha)/(partialx^mupartialx^nu)(dx^mu)/(dtau)(dx^nu)/(dtau)=0
(43)
 (partialxi^alpha)/(partialx^mu)(d^2x^mu)/(dtau^2)(partialx^lambda)/(partialxi^alpha)+(partial^2xi^alpha)/(partialx^mupartialx^nu)(dx^mu)/(dtau)(dx^nu)/(dtau)(partialx^lambda)/(partialxi^alpha)=0.
(44)

But

 (partialxi^alpha)/(partialx^nu)(partialx^lambda)/(partialxi^alpha)=delta_mu^lambda,
(45)

so

 delta_mu^lambda(d^2x^mu)/(dtau^2)+((partial^2xi^alpha)/(partialx^mupartialx^nu)(partialx^lambda)/(partialxi^alpha))(dx^mu)/(dtau)(dx^nu)/(dtau)=(d^2x^lambda)/(dtau^2)+Gamma^lambda_(munu)(dx^mu)/(dtau)(dx^nu)/(dtau),
(46)

where

 Gamma^lambda_(munu)=(partial^2xi^alpha)/(partialx^mupartialx^nu)(partialx^lambda)/(partialxi^alpha).
(47)

See also

Cartan Torsion Coefficient, Christoffel Symbol, Christoffel Symbol of the First Kind, Comma Derivative, Commutation Coefficient, Covariant Derivative, Gauss Equations, Levi-Civita Connection, Tensor

Explore with Wolfram|Alpha

References

Arfken, G. Mathematical Methods for Physicists, 3rd ed. Orlando, FL: Academic Press, pp. 160-167, 1985.Gray, A. "Christoffel Symbols." §22.3 in Modern Differential Geometry of Curves and Surfaces with Mathematica, 2nd ed. Boca Raton, FL: CRC Press, pp. 509-513, 1997.Misner, C. W.; Thorne, K. S.; and Wheeler, J. A. Gravitation. San Francisco, CA: W. H. Freeman, 1973.Morse, P. M. and Feshbach, H. Methods of Theoretical Physics, Part I. New York: McGraw-Hill, pp. 47-48, 1953.Sternberg, S. Differential Geometry. New York: Chelsea, p. 354, 1983.Walton, J. J. "Tensor Calculations on Computer: Appendix." Comm. ACM 10, 183-186, 1967.Weinberg, S. Gravitation and Cosmology: Principles and Applications of the General Theory of Relativity. New York: Wiley, 1972.

Referenced on Wolfram|Alpha

Christoffel Symbol of the Second Kind

Cite this as:

Weisstein, Eric W. "Christoffel Symbol of the Second Kind." From MathWorld--A Wolfram Resource. https://mathworld.wolfram.com/ChristoffelSymboloftheSecondKind.html

Subject classifications