Metric tensor

From Online Dictionary of Crystallography

Tenseur métrique (Fr). Metrischer Tensor (Ge). Tensore metrico (It). 計量テンソル (Ja). Метрический тензор (Ru). Tensor métrico (Sp).


A metric tensor is used to measure distances in a space. In crystallography the spaces considered are vector spaces with Euclidean metrics, i.e. ones for which the rules of Euclidean geometry apply. In that case, given a basis ei of a Euclidean space, En, the metric tensor is a rank 2 tensor the components of which are:

gij = ei . ej = ej.ei = gji.

It is a symmetrical tensor. Using the metric tensor, the scalar product of two vectors, x = xi ei and y = yj ej is written:

x . y = xi ei . yj ej = gij xi yj.

In a three-dimensional space with basis vectors a, b, c, the coefficients gij of the metric tensor are:

g11 = a2; g12 = a . b; g13 = a . c;
g21 = b . a; g22 = b2; g23 = b . c;
g31 = c . a; g32 = c . b; g33 = c2.

Because the metric tensor is symmetric, g12 = g21, g13 = g31, and g13 = g31. Thus there are only six unique elements, often presented as

g11 g22 g33
g23 g13 g12

or, multiplying the second row by 2, as a so-called G6 ("G" for Gruber) vector

( a2, b2, c2, 2b . c, 2a . c, 2a . b )

The inverse matrix of gij, gij, relates the dual basis, or reciprocal space vectors ei to the direct basis vectors ei, through the relations

ej = gij ej.

Note that gikgkj = δkj, where δkj is the Kronecker symbol, equal to 0 if ij, and equal to 1 if i = j.

In three-dimensional space, the dual basis vectors are identical to the reciprocal space vectors and the components of gij are:

g11 = a*2; g12 = a* . b*; g13 = a* . c*;
g21 = b* . a*; g22 = b*2; g23 = b* . c*;
g31 = c* . a*; g32 = c* . b*; g33 = c*2;


g11 = b2c2 sin2 α/ V2; g22 = c2a2 sin2 β/ V2; g33 = a2b2 sin2 γ/ V2;

g12 = g21 = (abc2/ V2)(cos α cos β - cos γ);

g23 = g32 = (a2bc/ V2)(cos β cos γ - cos α);

g31 = g13 = (ab2c/ V2)(cos γ cos α - cos β)

where V is the volume of the unit cell (a, b, c).

Change of basis

In a change of basis the direct basis vectors and coordinates transform like:

e'j = Aj i ei; x'j = Bi j x i,

where Aj i and Bi j are transformation matrices, transpose of one another. According to their definition, the components gij, of the metric tensor transform like products of basis vectors:

g'kl = AkiAljgij.

They are the doubly covariant components of the metric tensor.

The dual basis vectors and coordinates transform in the change of basis according to:

e'j = Bi j ei; x'j = Aj ixi,

and the components gij transform like products of dual basis vectors:

g'kl = Aik Ajl gij.

They are the doubly contravariant components of the metric tensor.

The mixed components, gij = δij, are once covariant and once contravariant and are invariant.

Properties of the metric tensor

  • The tensor nature of the metric tensor is demonstrated by the behaviour of its components in a change of basis. The components gij and gij are the components of a unique tensor.
  • The squares of the volumes V and V* of the direct space and reciprocal space unit cells are respectively equal to the determinants of the gij 's and the gij 's:

V 2 = Δ (gij) = abc(1 - cos 2 α - cos 2 β - cos2 γ + 2 cos α cos β cos γ)

V*2 = Δ (gij) = 1/ V 2.

  • One changes the variance of a tensor by taking the contracted tensor product of the tensor by the suitable form of the metric tensor. For instance:

gimt ij..kl.. = t j..klm..

Multiplying by the doubly covariant form of the metric tensor increases the covariance by one, multiplying by the doubly contravariant form increases the contravariance by one.

See also