Benchmark problems look good

python/sdist/amici/de_model.py

@@ -1707,6 +1707,7 @@ def _compute_equation(self, name: str) -> None:
                     tmp_eq += smart_multiply(
                         self.eq("ddeltaxdx")[ie], tmp_dxdp
                     )
+
                 else:
                     tmp_eq = smart_multiply(
                         self.sym("xdot") - self.sym("xdot_old"),

tests/benchmark-models/test_petab_benchmark.py

@@ -76,7 +76,7 @@ class GradientCheckSettings:
 settings = defaultdict(GradientCheckSettings)
 # NOTE: Newton method fails badly with ASA for Blasi_CellSystems2016
 settings["Blasi_CellSystems2016"] = GradientCheckSettings(
-    atol_check=1e-12,
+    atol_check=1e-3,
     rtol_check=1e-4,
     ss_sensitivity_mode=amici.SteadyStateSensitivityMode.integrationOnly,
 )
@@ -87,6 +87,8 @@ class GradientCheckSettings:
 )
 settings["Brannmark_JBC2010"] = GradientCheckSettings(
     ss_sensitivity_mode=amici.SteadyStateSensitivityMode.integrationOnly,
+    atol_check=1e-6,
+    rtol_check=1e-3,
 )
 settings["Giordano_Nature2020"] = GradientCheckSettings(
     atol_check=1e-6, rtol_check=1e-3, rng_seed=1
@@ -102,6 +104,8 @@ class GradientCheckSettings:
     # Avoid "root after reinitialization"
     atol_sim=1e-12,
     rtol_sim=1e-12,
+    rtol_check=1e-3,
+    atol_check=1e-12,
 )
 settings["Smith_BMCSystBiol2013"] = GradientCheckSettings(
     atol_sim=1e-10,
@@ -163,18 +167,6 @@ def test_benchmark_gradient(model, scale, sensitivity_method, request):
         # only fail on linear scale
         pytest.skip()
 
-    if (
-        model
-        in (
-            # events with parameter-dependent triggers
-            #  https://github.com/AMICI-dev/AMICI/issues/18
-            "Oliveira_NatCommun2021",
-        )
-        and sensitivity_method == SensitivityMethod.adjoint
-    ):
-        # FIXME
-        pytest.skip()
-
     petab_problem = benchmark_models_petab.get_problem(model)
     petab.flatten_timepoint_specific_output_overrides(petab_problem)