Repository navigation
Conversation
…/LSIC, Windkessel) assembled with Gridap and HyperFEMKernels; test_cfd.jl Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015yYpiif3W74pG5gVuyGq4o
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Adds
examples/aorta_cfd/: a self-contained extract ofhemodynamics/examples/patient_aorta/flow.jl. It covers incompressible Navier–Stokes with Oseen–BDF2 time stepping, Taylor–Hood P2–P1 elements with SUPG/PSPG/LSIC stabilization, backflow stabilization and Windkessel (Robin) outlets, on the coarse patient mesh (99 141 DOFs). The same weak form is assembled with Gridap and withKernelOperator.aorta_flow.jlmodes:compare(both assemblies every step, with relative differences and timings),gridap,kerneltest/test_cfd.jl: the same weak form on a small tetrahedral channel, against Gridap at 1e-12. It covers backward Euler and BDF2, the linear system with lifted Dirichlet values, data overwritten in place between steps, andupdate(16 tests)dev/prof_cfd.jl: time breakdown per compiled group (auxiliary data vs element kernels)Our measurement (Mac, 18 threads): matrix and vector equal to Gridap's within ~2e-16, with no fallback. Assembly takes 4.48 s with Gridap, 0.37 s with the kernel on 18 threads (×12) and 3.0 s on 1 thread (×1.5). A full step is about ×3 faster, and LU now dominates. On 1 thread, the volume Jacobian takes the dual-number path because the SUPG terms are products of auxiliary inputs.
Generated by Claude Code