Dohp aims to be a very efficient implementation of the hp-version of the finite element method. It exploits the tensor product structure of nodal bases on hexahedra to significantly reduce the memory requirements and computational cost compared to low-order elements. It typically forms a preconditioning matrix using an embedded Q1 discretization, which is much sparser than Q2 elements. Preliminary results show that memory and solver runtime for arbitrary order (2-10 or so) is half that required by a standard Q2 approximation. The methods are summarized in [1] and Dohp is used to solve nonlinear eigenproblems in [2].
Dohp currently depends on development versions of PETSc [3] and MOAB [4]. Additionally, some tests depend on SymPy [5] to manufacture the forcing terms for exact solutions.
If you are interested in Dohp, please let me know (jed at 59A2 dot org) and I'll fill you in on the design and how to get it installed (and update this README).
[1] http://dx.doi.org/10.1007/s10915-010-9396-8 [2] http://dx.doi.org/10.1007/s10915-011-9540-0 or http://arxiv.org/pdf/1011.3172v2 [3] http://mcs.anl.gov/petsc [4] http://trac.mcs.anl.gov/projects/ITAPS [5] http://sympy.org