GitLab

#545 VariationalFormulation: remove the need for the DefineOperators within the CRTP. This operation is quite naturally part of SupplInit(). In the process, remove the template parameters for the argument of SupplInit in model instances: the input parameter is actually completely known within a model instance.

#511 Stokes: create a new target for Stokes with two operators; checked with both monolithic and Uzawa schemes.

#511 Stokes operator now used in Stokes model. Macro have been introduced to also allow the use of previous code (with two separate operators). A new target will be defined accordingly; compilation with the other macro actiavted has not yet been tested.

#511 Stokes operator: Put the minus sign in local variational operator rather than in variational formulation.

#511 Stokes model: replace ScalarDivVectorial by a still incomplete Stokes operator. Main difference is that Stokes oeperator relies on a LocalMatrix, not a LocalSymmetricMatrix.

#537 GradGrad operator: the local operator has been slightly rewritten. It is a tad more verbosy but also very likely a bit more efficient as well (there are less images through Seldon::SubMatrix used).

#536 ScalarDivVectorial now works with symmetric matrix (and rough patches - some cleaning-up is due!).

#535 XCode: change a compilation option that led to a linker warning when symmetric Seldon matrix are used.

#535 Modifications made to extend several functions and methods to a template determined matrix type. I tried to apply this to Stokes problem, but there is a problem somewhere: elementary matrix is correct but the result is the same as if there were no contribution at all from the operator. There is probably a bug in the assembling part.

#535 ElementaryData: now the elementary matrix or vector type is determined by a template parameter.

#535 LocalVariationalOperator now gets one or two additional template parameters: the type of matrix and/or vector to use.