Fix FunctionMap DiscreteProblem import; correct Hard DAE test assertion

[ChrisRackauckas-Claude]

FunctionMap — DiscreteProblem not in scope

v7's OrdinaryDiffEq no longer re-exports DiscreteProblem. Import from SciMLBase.

Hard DAE — test assertion was wrong, not solver

The test asserted abs(sol(2 - 2^-10, idxs=1)) <= 1e-4, expecting u[1] to decay from 5 toward 0. But the DAE has:

M * du = f(u)
[1 0 0] [du1]   [-100*(u1-u2)]
[0 0 0] [du2] = [u1 - u2     ]
[0 0 1] [du3]   [-u3 + u1 + p2]

The algebraic constraint (row 2) is u2 = u1. Substituting: du1 = -100*(u1 - u1) = 0. So u1 = 5 (constant) — this is the correct solution.

The old solver had imperfect algebraic constraint enforcement: u2 didn't snap to u1 immediately, leaving a small u1 - u2 residual that drove u1 down over 2000 steps. v7's implicit Euler correctly enforces u2 = u1 from step 1, confirmed locally:

u at t=0:   [5.0, 0.0, 0.0]   (inconsistent u0)
u at t≈1:   [5.0, 5.0, 9.48]  (u2 corrected in first step)
u at t≈2-ε: [5.0, 5.0, 12.97] (u1 constant, correct)

Fix: assert sol(2-2^-10, idxs=1) ≈ 5.0 (the correct DAE solution). Also add initializealg = BrownFullBasicInit() since v7 defaults to CheckInit which rejects inconsistent u0 = [5, 0, 0].

🤖 Generated with Claude Code