PyLith Development Plans

Version 1.7 [by June workshop]

General

• Output of solution at arbitrary points via interpolation of solution. 
• Add 2-D plane strain version of the Drucker-Prager elastoplastic rheology. 
• Add 2-D plane strain version of the power-law viscoelastic rheology. 
• Elastic pre-step option for time-dependent problems. Remove assumed elastic pre-step at time step 0. 
• Add --initialize_only property to Problem  [0%]

  Permit users to stop simulation before time stepping in order to diagnose parameter settings.

Quasi-static

• Static Green's functions for fault slip. 

Dynamic

• None

Computational science

• Switch to new Sieve implementation.  [0%]

  C implementation better integrated with PETSc. Smaller memory usage. Facilitates multiphysics and higher order discretizations.

Version 1.8 (Fall or Winter 2012)

General

• Refactor initial fault tractions  [5%]

  Create Nucleation object with spatial and temporal perturbation of tractions from initial value.

Quasi-static

• Create strain hardening/softening 2-D and 3-D Drucker-Prager elastoplastic models.

Dynamic

• Compute stable time step for explicit time integration  [5%]
• Attenuation via generalized Maxwell model (bulk and shear relaxation)  [50%]

Computational science

• Accelerate FE integrations using GPUs  [25%]

  Will provide significant speedup to simulations run by many users because most are running on desktop machines that have GPUs.

Candidate Features for Version 2.0 (Spring 2013)

Major features

• Multiphysics
  1. Incompressible elasticity via a pressure field 
  2. Elasticity + heat flow 
  3. Elasticity + fluid flow 
• Earthquake cycle modeling
  1. Same mesh for dynamic and quasi-static parts 
     1. Dynamic -> quasi-static
     2. Quasi-static -> dynamic
     3. Complete cycle
  2. Different meshes for dynamic and quasi-static parts 

     Requires interpolation of fields between different meshes/discretizations and may require extrapolation of solutions when quasi-static problems span a larger domain than the dynamic problems.

• Use interpolated meshes (cells, faces, edges, vertices) to permit higher order basis functions 

Minor features

• Moment tensor point sources 

  Moment tensor point sources provide a mesh independent deformation source that is better suited for Green's function calculations than slip on a fault surface via cohesive cells.

• Time-step based on strain rate 
• Pressure field for incompressible elasticity problems 
• Use KD tree search algorithm to allow output of time histories at an arbitrary location 
• Combined prescribed slip / spontaneous rupture fault condition 

  Use fault constitutive model to control slip on fault except during episodes of prescribed slip. Need some way to describe when to turn on/off prescribed slip.

• Use threading to accelerate integrations on multi-core machines.