An adaptive DAE solver implemented in C++ and based on variable stepsize variable order BDF methods for the fast and reliable error-controlled solution of stiff ODEs and linear-implicit index 1 DAEs, including sensitivity generation of arbitrarily high order.

Its main properties are:

• forward and adjoint sensitivity generation based on IND
• Generation of directional sensitivities of arbitrary(!) order by pure forward as well as combined forward/adjoint schemes
• Error control of forward sensitivities
• Error controlled continuous representation of the solution and the forward sensitivities
• Support of dense and sparse linear algebra solvers
• Step-size and order selection based on the actual variable integration grid
• Monitor strategy for the solution of the implicit systems leading to fewer matrix factorization and Jacobian evaluations of the model functions
• Computation of consistent initial values
• Support of a relaxed formulation of the algebraic equations for the use in optimization context
• Storage and reuse of the integration grid during later integrations and sensitivity generation runs
• Runge-Kutta starter for the BDF method
• Modular design in C++
• Global error control of nominal solution