Class that represent nxn sparse blocks to put into K global matrix, that is blocks that connect N 'variables' and build a matrix K in a sparse variational inequality VI(Z*x-d,K):
| M+K -Cq'|*|q|- | f|= |0| , l Y, C Ny, normal cone to Y | Cq -E | |l| |-b| |c|
Also Z symmetric by flipping sign of l_i: |M+K Cq'|*| q|-| f|=|0| |Cq E | |-l| |-b| |c|
- case linear problem: all Y_i = R, Ny=0, ex. all bilaterals
- case LCP: all Y_i = R+: c>=0, l>=0, l*c=0
- case CCP: Y_i are friction cones Note that K blocks often have a physical interpretation as stiffness, but not always, for example they can represent hessians. Note that all blocks in K, all masses and constraint jacobians Cq are not really assembled in large matrices, so to exploit sparsity.
|ChKblockGeneric (std::vector< ChVariables * > mvariables)|
|ChKblockGeneric (ChVariables *mvariableA, ChVariables *mvariableB)|
|ChKblockGeneric &||operator= (const ChKblockGeneric &other)|
|Assignment operator: copy from other object. |
|void||SetVariables (std::vector< ChVariables * > mvariables)|
|Set references to the constrained objects, each of ChVariables type, automatically creating/resizing K matrix if needed. More...|
|virtual size_t||GetNvars () const override|
|Returns the number of referenced ChVariables items. |
|ChVariables *||GetVariableN (unsigned int m_var) const|
|Access the m-th vector variable object. |
|virtual ChMatrix< double > *||Get_K () override|
|Access the K stiffness matrix as a single block, referring only to the referenced ChVariable objects. |
|virtual void||MultiplyAndAdd (ChMatrix< double > &result, const ChMatrix< double > &vect) const override|
|Computes the product of the corresponding blocks in the system matrix (ie. More...|
|virtual void||DiagonalAdd (ChMatrix< double > &result) override|
|Add the diagonal of the stiffness matrix block(s) as a column vector to 'result'. More...|
|virtual void||Build_K (ChSparseMatrix &storage, bool add) override|
|Writes the K matrix associated to these variables into a global 'storage' matrix, at the offsets of variables. More...|
Member Function Documentation
Writes the K matrix associated to these variables into a global 'storage' matrix, at the offsets of variables.
Most solvers do not need this: the sparse 'storage' matrix is used for testing, for direct solvers, for dumping full matrix to Matlab for checks, etc.
Add the diagonal of the stiffness matrix block(s) as a column vector to 'result'.
NOTE: the 'result' vector must already have the size of system unknowns, ie the size of the total variables&constraints in the system; the procedure will use the ChVariable offsets (that must be already updated) as index.
Computes the product of the corresponding blocks in the system matrix (ie.
the K matrix blocks) by 'vect', and add to 'result'. NOTE: the 'vect' and 'result' vectors must already have the size of the total variables&constraints in the system; the procedure will use the ChVariable offsets (that must be already updated) to know the indexes in result and vect.