Temporary fix of problem generation

Author: albertomercurio

src/time_evolution/mesolve.jl

@@ -87,7 +87,13 @@ function mesolveProblem(
     kwargs3 = _generate_se_me_kwargs(e_ops, makeVal(progress_bar), tlist, kwargs2, SaveFuncMESolve)
 
     tspan = (tlist[1], tlist[end])
-    prob = ODEProblem{getVal(inplace),FullSpecialize}(L, ρ0, tspan, params; kwargs3...)
+
+    # TODO: Remove this when https://github.com/SciML/SciMLSensitivity.jl/issues/1181 is fixed
+    if haskey(kwargs3, :sensealg)
+        prob = ODEProblem{getVal(inplace)}(L, ρ0, tspan, params; kwargs3...)
+    else
+        prob = ODEProblem{getVal(inplace),FullSpecialize}(L, ρ0, tspan, params; kwargs3...)
+    end
 
     return TimeEvolutionProblem(prob, tlist, L_evo.dimensions, (isoperket = Val(isoperket(ψ0)),))
 end

src/time_evolution/sesolve.jl

@@ -73,7 +73,13 @@ function sesolveProblem(
     kwargs3 = _generate_se_me_kwargs(e_ops, makeVal(progress_bar), tlist, kwargs2, SaveFuncSESolve)
 
     tspan = (tlist[1], tlist[end])
-    prob = ODEProblem{getVal(inplace),FullSpecialize}(U, ψ0, tspan, params; kwargs3...)
+
+    # TODO: Remove this when https://github.com/SciML/SciMLSensitivity.jl/issues/1181 is fixed
+    if haskey(kwargs3, :sensealg)
+        prob = ODEProblem{getVal(inplace)}(U, ψ0, tspan, params; kwargs3...)
+    else
+        prob = ODEProblem{getVal(inplace),FullSpecialize}(U, ψ0, tspan, params; kwargs3...)
+    end
 
     return TimeEvolutionProblem(prob, tlist, H_evo.dimensions)
 end

test/ext-test/autodiff/zygote.jl

@@ -32,7 +32,7 @@
         my_f_sesolve(params)
 
 
-        grad_qt = Zygote.gradient(my_f_sesolve, params)[1] ./ 2
+        grad_qt = Zygote.gradient(my_f_sesolve, params)[1]
         grad_exact = [my_f_analytic_deriv(params[1])]
 
         @test grad_qt ≈ grad_exact atol=1e-6
@@ -76,7 +76,7 @@
         params = [Δ, F, γ]
 
         # The factor 2 is due to a bug
-        grad_qt = Zygote.gradient(my_f_mesolve, params)[1] ./ 2
+        grad_qt = Zygote.gradient(my_f_mesolve, params)[1]
 
         grad_exact = Zygote.gradient((p) -> n_ss(p[1], p[2], p[3]), params)[1]