You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
docs: retract the OrdinaryDiffEq interop claim and document field-level apply! as the explicit-stepping idiom #87
Two documentation fixes that surfaced in #73 and #52, grouped because they touch the same page.
README.md:145 advertises "the OrdinaryDiffEq.jl interop" and docs/ contains no occurrence of ODE, OrdinaryDiffEq or SciML. There is no SciML weak dep or extension. Retract the claim, or point it at the SciML umbrella as planned work. It becomes true when feat(sciml): build the planned SciMLOperators jac_prototype adapter (DESIGN §7) #65 lands.
Steer explicit stepping away from the flat mul! path (perf(algebra): constant-fold affine time-step operators (α·I + β·Laplacian) into one stencil pass #52 item 3). The prepared-mul! leg pays 15–50% over field-level apply! purely in flat_to_interior! / interior_to_flat! copies, and the gap widens with size. That price is right for Krylov, but the documented ODE-RHS idiom f!(du, u, p, t) = mul!(du, A, u) (src/linalg.jl:86-88) points explicit integrators at the expensive path. Document field-level apply! as the explicit-stepping idiom and keep mul! for the solver boundary.
Note quarto render docs runs the full test suite (CLAUDE.md); edit the .qmd and README without rendering unless the page changes materially.
🤖 Beep boop — Claude read the README against the code; Kyle is not overpromising to himself.
Activity
feat(linalg): apply! on prepared operators for field-level explicit s…
feat(linalg): apply! on prepared operators for field-level explicit s…