Setting up a new problem¶
The easiest way to set up a new problem is typically to copy an existing problem directory that is similar.
Adapting a Clawpack problem setup to SharpClaw¶
It is relatively quick and straightforward to adapt a problem that is set up to run in Clawpack to run in SharpClaw. The following modifications need to be made:
- Modify setrun.py
- Set time_integrator, tfluct_solver, char_decomp, src_term, lim_type, mbc
- Set mbc to 3
- Modify Makefile
- For this, it is best to copy an existing SharpClaw Makefile
- Slight modifications to:
- rpnN.f
- bcN.f
- outN.f
Source terms: srcN.f should be converted from full-discrete to semi-discrete form.
Optionally, provide tfluctN.f and/or evec.f
It is expected that in Clawpack 5.0, the setup for using SharpClaw or Clawpack will be even less pronounced, as the interfaces to these latter functions will be unified.
User-provided functions¶
Riemann solver¶
The Riemann solver is the basic building block of any Godunov-type method, including the methods in SharpClaw. The routine required takes the form of a wave-propagation solver, just like those used in Clawpack. For many systems of hyperbolic equations, a wave propagation Riemann solver already exists within Clawpack or SharpClaw. For new systems of equations, this function must be written by the user.
The inputs to the Riemann solver are the left and right states appearing in the Riemann problem, as well as any auxiliary variables (such as spatially varying coefficients) that are required for the solution of the Riemann problem. The outputs of the Riemann solver routine are the waves, speeds, and fluctuations.
The following routines must be provided by the user, if necessary.
Source terms¶
If the problem includes source terms, these should be included in a routine src.f. Unlike the source term routine for Clawpack, this routine should not integrate the source term over a time step. Rather, it should compute the instantaneous value of the source term in the semi-discrete system.