Skip to content

Enable Enzyme SCC test on Julia 1.11+ (Enzyme 0.13.148) #1449

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
ChrisRackauckas merged 1 commit into from
May 21, 2026
Merged

Enable Enzyme SCC test on Julia 1.11+ (Enzyme 0.13.148) #1449

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 2 additions & 13 deletions test/scc_nonlinearsolve.jl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,22 +70,11 @@ p_test = [4.0, 3.0]
@test isapprox(fwd, fd, rtol = 0.05)
end

# Enzyme through Tuple-based SCCNonlinearProblem now works reliably on
# Julia 1.10 (verified across 40+ consecutive runs with the current
# SCCNonlinearSolve, NonlinearSolveBase, and Enzyme releases). On Julia
# 1.11+, Enzyme still hits an `IllegalTypeAnalysisException` on
# `Base._typed_vcat!` inside SCCNonlinearSolve's solution assembly,
# which segfaults the worker process — gate the test on the LTS until
# the upstream Enzyme/NonlinearSolve issues are fixed.
# Vector-based SCC remains unsupported because heterogeneous function
# types get erased to Any. See Enzyme.jl#3021.
@testset "Enzyme" begin
if VERSION < v"1.11"
enz = Enzyme.gradient(Enzyme.Reverse, loss, copy(p_test))
@test isapprox(collect(enz[1]), fd, rtol = 0.05)
else
@test_skip true
end
enz = Enzyme.gradient(Enzyme.Reverse, loss, copy(p_test))
@test isapprox(collect(enz[1]), fd, rtol = 0.05)
end

@testset "Mooncake" begin
Expand Down
Loading