InvariantTensors
InvariantTensors[alg, {rep1, rep2, …}]
returns a basis of invariant tensors (CGs) which can contract the indices of the representations (order preserved).
Details and Options
- The result is a list of tensors (SparseArray).
- The invariant tensors are invariant when contracted with objects in of given representations. In other words, the indices of the produced tensors are {CRep[rep1], CRep[rep2], …}.
- All representation bases are consistent also with the tensors from StructureConstants and Generators.
- The algebra argument must be a valid Alg[class, n] object.
- Representations are labeled with a list of integers corresponding to the Dynkin coefficients of the highest weight in the representation.
- The following options can be given:
-
AntisymmetricIndices {} restricts the tensors to be antisymmetric in the given indices Normalization Default the normalization of the norm square of the invariant tensors. SymmetricIndices {} restricts the tensors to be symmetric in the given indices - The options given to Symmetricindices and AntisymmetricIndices can be either a list of index numbers or a list of list of index numbers with indices to be symmetrized between.
- The default for value for the norm square normalization of the tensors is
, where
is the dimension of each representation.
Examples
open allclose allBasic Examples (1)
An invariant tensor between three fundamental representations of SU(3)/A2 (note the use of SU[3] as short-hand for the algebra):
The only such tensor is proportional to the rank-3 Levi-Civita tensor, though the default normalization differs.
Options (3)
AntisymmetricIndices (1)
Normalization (1)
SymmetricIndices (1)
In the product of an anti-fundamental, a fundamental, and two adjoint representations of SU(3), there is is a two-dimensional space of invariant tensors that are symmetric in the two adjoint indices. Typical basis vectors are
dcab and
δab. Note that the tensor generation does not now about this way of writing the basis
When working with the CGs methods, and an invariant tensor is known to be a contraction of other objects (as in this example), the user can consider using DefineCompositeCG and skip InvariantTensor all together.