Format for JuliaFormatter v2

docs/make.jl

@@ -48,7 +48,7 @@ makedocs(;
         "https://twitter.com/ChrisRackauckas/status/1544743542094020615",
         "https://link.springer.com/article/10.1007/s40096-020-00339-4",
         "https://dl.acm.org/doi/10.1145/210089.210111",
-        "https://www.sciencedirect.com/science/article/abs/pii/S0045782523007156",
+        "https://www.sciencedirect.com/science/article/abs/pii/S0045782523007156"
     ],
     checkdocs = :exports,
     warnonly = [:missing_docs],

docs/src/tutorials/large_systems.md

@@ -74,14 +74,19 @@ function brusselator_2d_loop(du, u, p)
     @inbounds for I in CartesianIndices((N, N))
         i, j = Tuple(I)
         x, y = xyd_brusselator[I[1]], xyd_brusselator[I[2]]
-        ip1, im1, jp1, jm1 = limit(i + 1, N), limit(i - 1, N), limit(j + 1, N),
+        ip1, im1, jp1,
+        jm1 = limit(i + 1, N), limit(i - 1, N), limit(j + 1, N),
         limit(j - 1, N)
-        du[i, j, 1] = alpha * (u[im1, j, 1] + u[ip1, j, 1] + u[i, jp1, 1] + u[i, jm1, 1] -
-                       4u[i, j, 1]) +
-                      B +
-                      u[i, j, 1]^2 * u[i, j, 2] - (A + 1) * u[i, j, 1] + brusselator_f(x, y)
-        du[i, j, 2] = alpha * (u[im1, j, 2] + u[ip1, j, 2] + u[i, jp1, 2] + u[i, jm1, 2] -
-                       4u[i, j, 2]) + A * u[i, j, 1] - u[i, j, 1]^2 * u[i, j, 2]
+        du[i,
+            j,
+            1] = alpha * (u[im1, j, 1] + u[ip1, j, 1] + u[i, jp1, 1] + u[i, jm1, 1] -
+                  4u[i, j, 1]) +
+                 B +
+                 u[i, j, 1]^2 * u[i, j, 2] - (A + 1) * u[i, j, 1] + brusselator_f(x, y)
+        du[i,
+            j,
+            2] = alpha * (u[im1, j, 2] + u[ip1, j, 2] + u[i, jp1, 2] + u[i, jm1, 2] -
+                  4u[i, j, 2]) + A * u[i, j, 1] - u[i, j, 1]^2 * u[i, j, 2]
     end
 end
 p = (3.4, 1.0, 10.0, step(xyd_brusselator))

docs/src/tutorials/nonlinear_solve_gpus.md

@@ -31,7 +31,7 @@ using the first form.
 In this tutorial we will highlight both use cases in separate parts.
 
 !!! note
-
+    
     If you're looking for GPU-accelerated neural networks inside of nonlinear solvers,
     check out [DeepEquilibriumNetworks.jl](https://docs.sciml.ai/DeepEquilibriumNetworks/stable/).
 
@@ -58,7 +58,7 @@ f(u, p) = u .* u .- p
 u0 = cu(ones(1000))
 p = cu(collect(1:1000))
 prob = NonlinearProblem(f, u0, p)
-sol = solve(prob, NewtonRaphson(), abstol=1f-4)
+sol = solve(prob, NewtonRaphson(), abstol = 1.0f-4)
 ```
 
 Notice a few things here. One, nothing is different except the input array types. But
@@ -106,7 +106,7 @@ is saying, "for the ith call, get the i'th parameter set and solve with these pa
 The ith result is then this solution".
 
 !!! note
-
+    
     Because kernel code needs to be able to be compiled to a GPU kernel, it has very strict
     specifications of what's allowed because GPU cores are not as flexible as CPU cores.
     In general, this means that you need to avoid any runtime operations in kernel code,
@@ -137,16 +137,16 @@ Now let's build a nonlinear system to test it on.
     out2 = sqrt(p[2]) * (x[3] - x[4])
     out3 = (x[2] - p[3] * x[3])^2
     out4 = sqrt(p[4]) * (x[1] - x[4]) * (x[1] - x[4])
-    SA[out1,out2,out3,out4]
+    SA[out1, out2, out3, out4]
 end
 
 p = @SVector [@SVector(rand(Float32, 4)) for _ in 1:1024]
-u0 = SA[1f0, 2f0, 3f0, 4f0]
+u0 = SA[1.0f0, 2.0f0, 3.0f0, 4.0f0]
 prob = NonlinearSolveBase.ImmutableNonlinearProblem{false}(p2_f, u0, p)
 ```
 
 !!! note
-
+    
     Because the custom kernel is going to need to embed the the code for our nonlinear
     problem into the kernel, it also must be written to be GPU compatible.
     In general, this means that you need to avoid any runtime operations in kernel code,
@@ -173,4 +173,5 @@ vectorized_solve(prob, SimpleNewtonRaphson(); backend = Metal.MetalBackend())
 ```
 
 !!! warn
+
     The GPU-based calls will only work on your machine if you have a compatible GPU!

ext/NonlinearSolveFastLevenbergMarquardtExt.jl

@@ -21,7 +21,8 @@ function SciMLBase.__solve(
         termination_condition, alg
     )
 
-    f_wrapped, u, resid = NonlinearSolveBase.construct_extension_function_wrapper(
+    f_wrapped, u,
+    resid = NonlinearSolveBase.construct_extension_function_wrapper(
         prob; alias_u0, can_handle_oop = Val(prob.u0 isa SArray)
     )
     f = if prob.u0 isa SArray
@@ -49,7 +50,8 @@ function SciMLBase.__solve(
     )
 
     if prob.u0 isa SArray
-        res, fx, info, iter, nfev, njev = FastLM.lmsolve(
+        res, fx, info, iter, nfev,
+        njev = FastLM.lmsolve(
             f, jac_fn, prob.u0; solver_kwargs...
         )
         LM, solver = nothing, nothing
@@ -68,7 +70,8 @@ function SciMLBase.__solve(
 
         LM = FastLM.LMWorkspace(u, resid, J)
 
-        res, fx, info, iter, nfev, njev, LM, solver = FastLM.lmsolve!(
+        res, fx, info, iter, nfev, njev,
+        LM, solver = FastLM.lmsolve!(
             f, jac_fn, LM; solver, solver_kwargs...
         )
     end

ext/NonlinearSolveFixedPointAccelerationExt.jl

@@ -15,7 +15,8 @@ function SciMLBase.__solve(
         termination_condition, alg
     )
 
-    f, u0, resid = NonlinearSolveBase.construct_extension_function_wrapper(
+    f, u0,
+    resid = NonlinearSolveBase.construct_extension_function_wrapper(
         prob; alias_u0, make_fixed_point = Val(true), force_oop = Val(true)
     )
 

ext/NonlinearSolveMINPACKExt.jl

@@ -17,7 +17,8 @@ function SciMLBase.__solve(
         termination_condition, alg
     )
 
-    f_wrapped!, u0, resid = NonlinearSolveBase.construct_extension_function_wrapper(
+    f_wrapped!, u0,
+    resid = NonlinearSolveBase.construct_extension_function_wrapper(
         prob; alias_u0
     )
     resid_size = size(resid)

ext/NonlinearSolveNLSolversExt.jl

@@ -33,7 +33,8 @@ function SciMLBase.__solve(
             prob.f, autodiff, prob.u0, Constant(prob.p)
         )
 
-        fj_scalar = @closure (Jx, x) -> begin
+        fj_scalar = @closure (Jx,
+            x) -> begin
             return DifferentiationInterface.value_and_derivative(
                 prob.f, prep, autodiff, x, Constant(prob.p)
             )

ext/NonlinearSolvePETScExt.jl

@@ -22,7 +22,8 @@ function SciMLBase.__solve(
         termination_condition, alg; abs_norm_supported = false
     )
 
-    f_wrapped!, u0, resid = NonlinearSolveBase.construct_extension_function_wrapper(
+    f_wrapped!, u0,
+    resid = NonlinearSolveBase.construct_extension_function_wrapper(
         prob; alias_u0
     )
     T = eltype(u0)
@@ -48,7 +49,9 @@ function SciMLBase.__solve(
 
     nf = Ref{Int}(0)
 
-    f! = @closure (cfx, cx, user_ctx) -> begin
+    f! = @closure (cfx,
+        cx,
+        user_ctx) -> begin
         nf[] += 1
         fx = cfx isa Ptr{Nothing} ? PETSc.unsafe_localarray(T, cfx; read = false) : cfx
         x = cx isa Ptr{Nothing} ? PETSc.unsafe_localarray(T, cx; write = false) : cx
@@ -76,7 +79,8 @@ function SciMLBase.__solve(
             )
             J_init = zeros(T, 1, 1)
         else
-            jac!, J_init = NonlinearSolveBase.construct_extension_jac(
+            jac!,
+            J_init = NonlinearSolveBase.construct_extension_jac(
                 prob, alg, u0, resid; autodiff, initial_jacobian = Val(true)
             )
         end
@@ -85,7 +89,10 @@ function SciMLBase.__solve(
 
         if J_init isa AbstractSparseMatrix
             PJ = PETSc.MatSeqAIJ(J_init)
-            jac_fn! = @closure (cx, J, _, user_ctx) -> begin
+            jac_fn! = @closure (cx,
+                J,
+                _,
+                user_ctx) -> begin
                 njac[] += 1
                 x = cx isa Ptr{Nothing} ? PETSc.unsafe_localarray(T, cx; write = false) : cx
                 if J isa PETSc.AbstractMat
@@ -102,7 +109,10 @@ function SciMLBase.__solve(
             snes.user_ctx = (; jacobian = J_init)
         else
             PJ = PETSc.MatSeqDense(J_init)
-            jac_fn! = @closure (cx, J, _, user_ctx) -> begin
+            jac_fn! = @closure (cx,
+                J,
+                _,
+                user_ctx) -> begin
                 njac[] += 1
                 x = cx isa Ptr{Nothing} ? PETSc.unsafe_localarray(T, cx; write = false) : cx
                 jac!(J, x)

ext/NonlinearSolveSIAMFANLEquationsExt.jl

@@ -64,7 +64,8 @@ function SciMLBase.__solve(
         elseif method == :secant
             sol = secant(f, prob.u0; maxit = maxiters, atol, rtol, printerr)
         elseif method == :anderson
-            f_aa, u, _ = NonlinearSolveBase.construct_extension_function_wrapper(
+            f_aa, u,
+            _ = NonlinearSolveBase.construct_extension_function_wrapper(
                 prob; alias_u0, make_fixed_point = Val(true)
             )
             sol = aasol(
@@ -73,7 +74,8 @@ function SciMLBase.__solve(
             )
         end
     else
-        f, u, resid = NonlinearSolveBase.construct_extension_function_wrapper(
+        f, u,
+        resid = NonlinearSolveBase.construct_extension_function_wrapper(
             prob; alias_u0, make_fixed_point = Val(method == :anderson)
         )
         N = length(u)

ext/NonlinearSolveSpeedMappingExt.jl

@@ -16,7 +16,8 @@ function SciMLBase.__solve(
         termination_condition, alg
     )
 
-    m!, u, resid = NonlinearSolveBase.construct_extension_function_wrapper(
+    m!, u,
+    resid = NonlinearSolveBase.construct_extension_function_wrapper(
         prob; alias_u0, make_fixed_point = Val(true)
     )
     tol = NonlinearSolveBase.get_tolerance(abstol, eltype(u))

lib/BracketingNonlinearSolve/ext/BracketingNonlinearSolveForwardDiffExt.jl

@@ -7,7 +7,9 @@ using SciMLBase: SciMLBase, IntervalNonlinearProblem
 
 using BracketingNonlinearSolve: Bisection, Brent, Alefeld, Falsi, ITP, Ridder
 
-const DualIntervalNonlinearProblem{T, V, P} = IntervalNonlinearProblem{
+const DualIntervalNonlinearProblem{T,
+    V,
+    P} = IntervalNonlinearProblem{
     uType, iip, <:Union{<:Dual{T, V, P}, <:AbstractArray{<:Dual{T, V, P}}}
 } where {uType, iip}
 

lib/BracketingNonlinearSolve/src/bisection.jl

@@ -86,7 +86,8 @@ function CommonSolve.solve(
         i += 1
     end
 
-    sol, i, left, right, fl, fr = Impl.bisection(
+    sol, i, left, right,
+    fl, fr = Impl.bisection(
         left, right, fl, fr, f, abstol, maxiters - i, prob, alg
     )
 

lib/BracketingNonlinearSolve/src/brent.jl

@@ -118,7 +118,8 @@ function CommonSolve.solve(
         i += 1
     end
 
-    sol, i, left, right, fl, fr = Impl.bisection(
+    sol, i, left, right,
+    fl, fr = Impl.bisection(
         left, right, fl, fr, f, abstol, maxiters - i, prob, alg
     )
 

lib/BracketingNonlinearSolve/src/falsi.jl

@@ -76,7 +76,8 @@ function CommonSolve.solve(
         i += 1
     end
 
-    sol, i, left, right, fl, fr = Impl.bisection(
+    sol, i, left, right,
+    fl, fr = Impl.bisection(
         left, right, fl, fr, f, abstol, maxiters - i, prob, alg
     )
 

lib/BracketingNonlinearSolve/src/ridder.jl

@@ -90,7 +90,8 @@ function CommonSolve.solve(
         i += 1
     end
 
-    sol, i, left, right, fl, fr = Impl.bisection(
+    sol, i, left, right,
+    fl, fr = Impl.bisection(
         left, right, fl, fr, f, abstol, maxiters - i, prob, alg
     )
 

lib/BracketingNonlinearSolve/test/rootfind_tests.jl

@@ -18,7 +18,7 @@ end
 
         @testset for p in 1.1:0.1:100.0
             @test g(p)≈sqrt(p) atol=1e-3 rtol=1e-3
-            @test ForwardDiff.derivative(g, p)≈1 / (2 * sqrt(p)) atol=1e-3 rtol=1e-3
+            @test ForwardDiff.derivative(g, p)≈1/(2*sqrt(p)) atol=1e-3 rtol=1e-3
         end
 
         t = (p) -> [sqrt(p[2] / p[1])]
@@ -30,7 +30,7 @@ end
             return [sol.u]
         end
 
-        @test g2(p)≈[sqrt(p[2] / p[1])] atol=1e-3 rtol=1e-3
+        @test g2(p)≈[sqrt(p[2]/p[1])] atol=1e-3 rtol=1e-3
         @test ForwardDiff.jacobian(g2, p)≈ForwardDiff.jacobian(t, p) atol=1e-3 rtol=1e-3
 
         probB = IntervalNonlinearProblem{false}(quadratic_f, (1.0, 2.0), 2.0)
@@ -50,8 +50,8 @@ end
 end
 
 @testitem "Tolerance Tests Interval Methods" setup=[RootfindingTestSnippet] tags=[:core] begin
-    prob = IntervalNonlinearProblem(quadratic_f, (1.0, 20.0), 2.0)
-    ϵ = eps(Float64) # least possible tol for all methods
+    prob=IntervalNonlinearProblem(quadratic_f, (1.0, 20.0), 2.0)
+    ϵ=eps(Float64) # least possible tol for all methods
 
     @testset for alg in (Bisection(), Falsi(), ITP(), Muller(), nothing)
         @testset for abstol in [0.1, 0.01, 0.001, 0.0001, 1e-5, 1e-6]

lib/NonlinearSolveBase/ext/NonlinearSolveBaseForwardDiffExt.jl

@@ -124,7 +124,8 @@ for algType in GENERAL_SOLVER_TYPES
     @eval function SciMLBase.__solve(
             prob::DualAbstractNonlinearProblem, alg::$(algType), args...; kwargs...
     )
-        sol, partials = NonlinearSolveBase.nonlinearsolve_forwarddiff_solve(
+        sol,
+        partials = NonlinearSolveBase.nonlinearsolve_forwarddiff_solve(
             prob, alg, args...; kwargs...
         )
         dual_soln = NonlinearSolveBase.nonlinearsolve_dual_solution(sol.u, partials, prob.p)

lib/NonlinearSolveBase/src/NonlinearSolveBase.jl

@@ -73,7 +73,7 @@ include("forward_diff.jl")
 @compat(public, (construct_jacobian_cache,))
 @compat(public,
     (assert_extension_supported_termination_condition,
-        construct_extension_function_wrapper, construct_extension_jac))
+    construct_extension_function_wrapper, construct_extension_jac))
 
 export TraceMinimal, TraceWithJacobianConditionNumber, TraceAll
 

lib/NonlinearSolveBase/src/autodiff.jl

@@ -118,22 +118,25 @@ function nlls_generate_vjp_function(prob::NonlinearLeastSquaresProblem, sol, uu)
     # nested autodiff as the last resort
     if SciMLBase.has_vjp(prob.f)
         if SciMLBase.isinplace(prob)
-            return @closure (du, u, p) -> begin
+            return @closure (
+                du, u, p) -> begin
                 resid = Utils.safe_similar(du, length(sol.resid))
                 prob.f(resid, u, p)
                 prob.f.vjp(du, resid, u, p)
                 du .*= 2
                 return nothing
             end
         else
-            return @closure (u, p) -> begin
+            return @closure (
+                u, p) -> begin
                 resid = prob.f(u, p)
                 return reshape(2 .* prob.f.vjp(resid, u, p), size(u))
             end
         end
     elseif SciMLBase.has_jac(prob.f)
         if SciMLBase.isinplace(prob)
-            return @closure (du, u, p) -> begin
+            return @closure (
+                du, u, p) -> begin
                 J = Utils.safe_similar(du, length(sol.resid), length(u))
                 prob.f.jac(J, u, p)
                 resid = Utils.safe_similar(du, length(sol.resid))
@@ -142,7 +145,8 @@ function nlls_generate_vjp_function(prob::NonlinearLeastSquaresProblem, sol, uu)
                 return nothing
             end
         else
-            return @closure (u, p) -> begin
+            return @closure (u,
+                p) -> begin
                 return reshape(2 .* vec(prob.f(u, p))' * prob.f.jac(u, p), size(u))
             end
         end
@@ -152,7 +156,8 @@ function nlls_generate_vjp_function(prob::NonlinearLeastSquaresProblem, sol, uu)
                    select_reverse_mode_autodiff(prob, nothing) : AutoForwardDiff()
 
         if SciMLBase.isinplace(prob)
-            return @closure (du, u, p) -> begin
+            return @closure (
+                du, u, p) -> begin
                 resid = Utils.safe_similar(du, length(sol.resid))
                 prob.f(resid, u, p)
                 # Using `Constant` lead to dual ordering issues
@@ -163,7 +168,8 @@ function nlls_generate_vjp_function(prob::NonlinearLeastSquaresProblem, sol, uu)
                 return nothing
             end
         else
-            return @closure (u, p) -> begin
+            return @closure (u,
+                p) -> begin
                 v = prob.f(u, p)
                 # Using `Constant` lead to dual ordering issues
                 res = only(DI.pullback(Base.Fix2(prob.f, p), autodiff, u, (v,)))