Skip to content

Latest commit

 

History

History
51 lines (43 loc) · 2.25 KB

todo.md

File metadata and controls

51 lines (43 loc) · 2.25 KB
Raw

Next Steps

  • Reset Button

  • JSON Serialization/Deserialization of settings

  • Partio/VTK input/output

  • Fix APIC Stability issue (invertibility?)

  • Marching Cubes

  • More Initial Conditions

    • Report suggested rest density for initial conditions

  • Refactor Constiutive Models into a separate file

  • Better Camera Controller

Longer Term

PhysBAM Stuff

What classes in PhysBAM would you actually need to port to get most of it?

Not necessary

  • Vectors and Matricies (provided by Nalgebra)
  • Arrays not necessary (Rust's Vec is probably good enough)
  • Math_Tools not necessary (we may or may not want to re-implement RANGE)
  • LOG -- just use rust's log library, maybe slog or tracing (including scope timers)
  • Random Numbers -- just use rand

Necessary

  • GRID -- this is a really handy class

  • Iterators -- RANGE_ITERATOR, NODE_ITERATOR, CELL_ITERATOR, FACE_ITERATOR (w/ FACE_INDEX), these can probably be written pretty cleanly with Rust's iterator (better than separate Next and Valid functions)

  • Arrays Nd -- we can probably implement these ourselves better than ndarray

  • Particles -- we can definitely come up with a better way to do this than PhysBAM's dynamic particle arrays

  • Poisson Disk sampling

  • Grid_PDE

    • Advection Trait w/ Semi lagrangian (maybe generic over a particle tracing API?), WENO, ENO, etc.
    • There has to be a more consistent way to handle boundaries
    • Craig's new Interpolation API is very nice, porting that to rust would be cool (and not being tied down by backwards compatibility would be nice as well)

  • Meshes -- this is necessary

  • Output -- somehow use serde for this, if binary is necessary (which it probably is), serde can handle it -- maybe partio for particles

  • Level Sets -- probably fine

    • what about analytic implicit objects? i'm not sure...

  • Krylov Solvers

    • Don't know this area well enough, but I'm sure PhysBAM's API can be cleaned up -- LAPLACE, POISSON, PROJECTION, KRYLOV_SYSTEM_BASE

  • Constitutive Models -- the rest of it probably isn't strictly necessary immediately

  • and of course, a viewer

  • Formats

    • Bincode for data serialization
    • Tracing/Tracing-subscriber for logging