Hi!
Sorry to ask a question again. I tried to implement a TensorFunction
instead of the Function and found that this object is not symmetrical
physically. I am working on a damage problem but as a more simple
example let us consider a 3D x-y spin model. This model is in the
three-dimensional space but its quantities (magnetization) are
two-dimensional. In other words, on the 3D space the local vector of
magnetization can rotate only in the x-y plane. This situation
prohibits me to use the TensorFunction to return both values and
gradients. Indeed,
typedef Tensor<rank,dim> TensorFunction< rank, dim >::value_type
means that to return a vector of 2-dimensional magnetization i need
rank=1 and dim=2. But then i need for the 2-component magnetization to
return for each component its 3-dimensional gradients. I cannot use
already
typedef Tensor<rank+1,dim> TensorFunction< rank, dim >::gradient_type
because i need a 2-component Vector of 3-dimensional Tensors.
Tensor<rank+1,dim> of rank=2 and dimensionality 2 will not do the
trick.
The reason of the question is that in the TensorFunction the
dimensionality of the function is restricted to the dimensionality of
the embedding space.
Best,
Sergey

_______________________________________________

Reply via email to