Code Optimization for Differential Equations
See this FAQ for information on common pitfalls and how to improve performance.
Code Optimization in Julia
Before starting this tutorial, we recommend the reader to check out one of the many tutorials for optimization Julia code. The following is an incomplete list:
- The Julia Performance Tips
- MIT 18.337 Course Notes on Optimizing Serial Code
- What scientists must know about hardware to write fast code
User-side optimizations are important because, for sufficiently difficult problems, most time will be spent inside your f function, the function you are trying to solve. “Efficient” integrators are those that reduce the required number of f calls to hit the error tolerance. The main ideas for optimizing your DiffEq code, or any Julia function, are the following:
- Make it non-allocating
- Use StaticArrays for small arrays
- Use broadcast fusion
- Make it type-stable
- Reduce redundant calculations
- Make use of BLAS calls
- Optimize algorithm choice
We'll discuss these strategies in the context of differential equations. Let's start with small systems.
Example Accelerating a Non-Stiff Equation: The Lorenz Equation
Let's take the classic Lorenz system. Let's start by naively writing the system in its out-of-place form:
function lorenz(u, p, t)
dx = 10.0 * (u[2] - u[1])
dy = u[1] * (28.0 - u[3]) - u[2]
dz = u[1] * u[2] - (8 / 3) * u[3]
[dx, dy, dz]
endlorenz (generic function with 1 method)Here, lorenz returns an object, [dx,dy,dz], which is created within the body of lorenz.
This is a common code pattern from high-level languages like MATLAB, SciPy, or R's deSolve. However, the issue with this form is that it allocates a vector, [dx,dy,dz], at each step. Let's benchmark the solution process with this choice of function:
using DifferentialEquations, BenchmarkTools
u0 = [1.0; 0.0; 0.0]
tspan = (0.0, 100.0)
prob = ODEProblem(lorenz, u0, tspan)
@btime solve(prob, Tsit5()); 3.497 ms (192104 allocations: 7.44 MiB)The BenchmarkTools.jl package's @benchmark runs the code multiple times to get an accurate measurement. The minimum time is the time it takes when your OS and other background processes aren't getting in the way. Notice that in this case it takes about 5ms to solve and allocates around 11.11 MiB. However, if we were to use this inside of a real user code, we'd see a lot of time spent doing garbage collection (GC) to clean up all the arrays we made. Even if we turn off saving, we have these allocations.
@btime solve(prob, Tsit5(), save_everystep = false); 3.078 ms (169897 allocations: 6.49 MiB)The problem, of course, is that arrays are created every time our derivative function is called. This function is called multiple times per step and is thus the main source of memory usage. To fix this, we can use the in-place form to ***make our code non-allocating***:
function lorenz!(du, u, p, t)
du[1] = 10.0 * (u[2] - u[1])
du[2] = u[1] * (28.0 - u[3]) - u[2]
du[3] = u[1] * u[2] - (8 / 3) * u[3]
nothing
endlorenz! (generic function with 1 method)Here, instead of creating an array each time, we utilized the cache array du. When the in-place form is used, DifferentialEquations.jl takes a different internal route that minimizes the internal allocations as well.
Notice that nothing is returned. When in in-place form, the ODE solver ignores the return. Instead, make sure that the original du array is mutated instead of constructing a new array
When we benchmark this function, we will see quite a difference.
u0 = [1.0; 0.0; 0.0]
tspan = (0.0, 100.0)
prob = ODEProblem(lorenz!, u0, tspan)
@btime solve(prob, Tsit5()); 743.616 μs (23214 allocations: 1013.34 KiB)@btime solve(prob, Tsit5(), save_everystep = false); 353.478 μs (71 allocations: 3.89 KiB)There is a 16x time difference just from that change! Notice there are still some allocations and this is due to the construction of the integration cache. But this doesn't scale with the problem size:
tspan = (0.0, 500.0) # 5x longer than before
prob = ODEProblem(lorenz!, u0, tspan)
@btime solve(prob, Tsit5(), save_everystep = false); 1.830 ms (71 allocations: 3.89 KiB)Since that's all setup allocations, the user-side optimization is complete.
Further Optimizations of Small Non-Stiff ODEs with StaticArrays
Allocations are only expensive if they are “heap allocations”. For a more in-depth definition of heap allocations, there are many sources online. But a good working definition is that heap allocations are variable-sized slabs of memory which have to be pointed to, and this pointer indirection costs time. Additionally, the heap has to be managed, and the garbage controllers has to actively keep track of what's on the heap.
However, there's an alternative to heap allocations, known as stack allocations. The stack is statically-sized (known at compile time) and thus its accesses are quick. Additionally, the exact block of memory is known in advance by the compiler, and thus re-using the memory is cheap. This means that allocating on the stack has essentially no cost!
Arrays have to be heap allocated because their size (and thus the amount of memory they take up) is determined at runtime. But there are structures in Julia which are stack-allocated. structs for example are stack-allocated “value-type”s. Tuples are a stack-allocated collection. The most useful data structure for DiffEq though is the StaticArray from the package StaticArrays.jl. These arrays have their length determined at compile-time. They are created using macros attached to normal array expressions, for example:
using StaticArrays
A = SA[2.0, 3.0, 5.0]
typeof(A) # SVector{3, Float64} (alias for SArray{Tuple{3}, Float64, 1, 3})SVector{3, Float64} (alias for StaticArraysCore.SArray{Tuple{3}, Float64, 1, 3})Notice that the 3 after SVector gives the size of the SVector. It cannot be changed. Additionally, SVectors are immutable, so we have to create a new SVector to change values. But remember, we don't have to worry about allocations because this data structure is stack-allocated. SArrays have numerous extra optimizations as well: they have fast matrix multiplication, fast QR factorizations, etc. which directly make use of the information about the size of the array. Thus, when possible, they should be used.
Unfortunately, static arrays can only be used for sufficiently small arrays. After a certain size, they are forced to heap allocate after some instructions and their compile time balloons. Thus, static arrays shouldn't be used if your system has more than ~20 variables. Additionally, only the native Julia algorithms can fully utilize static arrays.
Let's ***optimize lorenz using static arrays***. Note that in this case, we want to use the out-of-place allocating form, but this time we want to output a static array:
function lorenz_static(u, p, t)
dx = 10.0 * (u[2] - u[1])
dy = u[1] * (28.0 - u[3]) - u[2]
dz = u[1] * u[2] - (8 / 3) * u[3]
SA[dx, dy, dz]
endlorenz_static (generic function with 1 method)To make the solver internally use static arrays, we simply give it a static array as the initial condition:
u0 = SA[1.0, 0.0, 0.0]
tspan = (0.0, 100.0)
prob = ODEProblem(lorenz_static, u0, tspan)
@btime solve(prob, Tsit5()); 264.578 μs (2615 allocations: 392.71 KiB)@btime solve(prob, Tsit5(), save_everystep = false); 191.159 μs (25 allocations: 2.03 KiB)And that's pretty much all there is to it. With static arrays, you don't have to worry about allocating, so use operations like * and don't worry about fusing operations (discussed in the next section). Do “the vectorized code” of R/MATLAB/Python and your code in this case will be fast, or directly use the numbers/values.
Example Accelerating a Stiff Equation: the Robertson Equation
For these next examples, let's solve the Robertson equations (also known as ROBER):
\[\begin{aligned} \frac{dy_1}{dt} &= -0.04y₁ + 10^4 y_2 y_3 \\ \frac{dy_2}{dt} &= 0.04 y_1 - 10^4 y_2 y_3 - 3*10^7 y_{2}^2 \\ \frac{dy_3}{dt} &= 3*10^7 y_{2}^2 \\ \end{aligned}\]
Given that these equations are stiff, non-stiff ODE solvers like Tsit5 or Vern9 will fail to solve these equations. The automatic algorithm will detect this and automatically switch to something more robust to handle these issues. For example:
using DifferentialEquations
using Plots
function rober!(du, u, p, t)
y₁, y₂, y₃ = u
k₁, k₂, k₃ = p
du[1] = -k₁ * y₁ + k₃ * y₂ * y₃
du[2] = k₁ * y₁ - k₂ * y₂^2 - k₃ * y₂ * y₃
du[3] = k₂ * y₂^2
nothing
end
prob = ODEProblem(rober!, [1.0, 0.0, 0.0], (0.0, 1e5), [0.04, 3e7, 1e4])
sol = solve(prob)
plot(sol, tspan = (1e-2, 1e5), xscale = :log10)
using BenchmarkTools
@btime solve(prob); 118.159 μs (1862 allocations: 134.56 KiB)Choosing a Good Solver
Choosing a good solver is required for getting top-notch speed. General recommendations can be found on the solver page (for example, the ODE Solver Recommendations). The current recommendations can be simplified to a Rosenbrock method (Rosenbrock23 or Rodas5) for smaller (<50 ODEs) problems, ESDIRK methods for slightly larger (TRBDF2 or KenCarp4 for <2000 ODEs), and QNDF for even larger problems. lsoda from LSODA.jl is sometimes worth a try for the medium-sized category.
More details on the solver to choose can be found by benchmarking. See the SciMLBenchmarks to compare many solvers on many problems.
From this, we try the recommendation of Rosenbrock23() for stiff ODEs at default tolerances:
@btime solve(prob, Rosenbrock23()); 81.209 μs (764 allocations: 36.53 KiB)Declaring Jacobian Functions
In order to reduce the Jacobian construction cost, one can describe a Jacobian function by using the jac argument for the ODEFunction. First we have to derive the Jacobian $\frac{df_i}{du_j}$ which is J[i,j]. From this, we get:
function rober_jac!(J, u, p, t)
y₁, y₂, y₃ = u
k₁, k₂, k₃ = p
J[1, 1] = k₁ * -1
J[2, 1] = k₁
J[3, 1] = 0
J[1, 2] = y₃ * k₃
J[2, 2] = y₂ * k₂ * -2 + y₃ * k₃ * -1
J[3, 2] = y₂ * 2 * k₂
J[1, 3] = k₃ * y₂
J[2, 3] = k₃ * y₂ * -1
J[3, 3] = 0
nothing
end
f! = ODEFunction(rober!, jac = rober_jac!)
prob_jac = ODEProblem(f!, [1.0, 0.0, 0.0], (0.0, 1e5), (0.04, 3e7, 1e4))ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
Non-trivial mass matrix: false
timespan: (0.0, 100000.0)
u0: 3-element Vector{Float64}:
1.0
0.0
0.0@btime solve(prob_jac, Rosenbrock23()); 54.720 μs (680 allocations: 33.72 KiB)Automatic Derivation of Jacobian Functions
But that was hard! If you want to take the symbolic Jacobian of numerical code, we can make use of ModelingToolkit.jl to symbolic-ify the numerical code and do the symbolic calculation and return the Julia code for this.
using ModelingToolkit
de = complete(modelingtoolkitize(prob))\[ \begin{align} \frac{\mathrm{d} \mathtt{x_1}\left( t \right)}{\mathrm{d}t} &= - \mathtt{x_1}\left( t \right) \mathtt{\alpha_1} + \mathtt{x_2}\left( t \right) \mathtt{x_3}\left( t \right) \mathtt{\alpha_3} \\ \frac{\mathrm{d} \mathtt{x_2}\left( t \right)}{\mathrm{d}t} &= \mathtt{x_1}\left( t \right) \mathtt{\alpha_1} - \left( \mathtt{x_2}\left( t \right) \right)^{2} \mathtt{\alpha_2} - \mathtt{x_2}\left( t \right) \mathtt{x_3}\left( t \right) \mathtt{\alpha_3} \\ \frac{\mathrm{d} \mathtt{x_3}\left( t \right)}{\mathrm{d}t} &= \left( \mathtt{x_2}\left( t \right) \right)^{2} \mathtt{\alpha_2} \end{align} \]
We can tell it to compute the Jacobian if we want to see the code:
ModelingToolkit.generate_jacobian(de)[2] # Second is in-place:(function (ˍ₋out, __mtk_arg_1, ___mtkparameters___, t)
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/Symbolics/285ea/src/build_function.jl:368 =# @inbounds begin
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/Symbolics/285ea/src/build_function.jl:368 =#
begin
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/SymbolicUtils/0GKgW/src/code.jl:409 =#
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/SymbolicUtils/0GKgW/src/code.jl:410 =#
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/SymbolicUtils/0GKgW/src/code.jl:411 =#
begin
__mtk_arg_2 = ___mtkparameters___[1]
__mtk_arg_3 = ___mtkparameters___[2]
begin
var"##cse#1" = (*)(-1, __mtk_arg_2[2])
var"##cse#2" = __mtk_arg_1[3]
var"##cse#3" = (*)(__mtk_arg_2[1], var"##cse#2")
var"##cse#4" = (*)((*)(-1, __mtk_arg_2[1]), var"##cse#2")
var"##cse#5" = __mtk_arg_1[2]
var"##cse#6" = (*)((*)(-2, var"##cse#5"), __mtk_arg_2[3])
var"##cse#7" = (+)(var"##cse#4", var"##cse#6")
var"##cse#8" = (*)((*)(2, var"##cse#5"), __mtk_arg_2[3])
var"##cse#9" = (*)(__mtk_arg_2[1], var"##cse#5")
var"##cse#10" = (*)((*)(-1, __mtk_arg_2[1]), var"##cse#5")
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/SymbolicUtils/0GKgW/src/code.jl:464 =# @inbounds begin
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/SymbolicUtils/0GKgW/src/code.jl:460 =#
ˍ₋out[1] = var"##cse#1"
ˍ₋out[2] = __mtk_arg_2[2]
ˍ₋out[3] = 0
ˍ₋out[4] = var"##cse#3"
ˍ₋out[5] = var"##cse#7"
ˍ₋out[6] = var"##cse#8"
ˍ₋out[7] = var"##cse#9"
ˍ₋out[8] = var"##cse#10"
ˍ₋out[9] = 0
#= /root/.cache/julia-buildkite-plugin/depots/0185fce3-4489-413a-a934-123dd653ef61/packages/SymbolicUtils/0GKgW/src/code.jl:462 =#
ˍ₋out
end
end
end
end
end
end)Now let's use that to give the analytical solution Jacobian:
prob_jac2 = ODEProblem(de, [], (0.0, 1e5), jac = true)ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
Initialization status: FULLY_DETERMINED
Non-trivial mass matrix: false
timespan: (0.0, 100000.0)
u0: 3-element Vector{Float64}:
1.0
0.0
0.0@btime solve(prob_jac2); 148.769 μs (1947 allocations: 139.62 KiB)See the ModelingToolkit.jl documentation for more details.
Accelerating Small ODE Solves with Static Arrays
If the ODE is sufficiently small (<20 ODEs or so), using StaticArrays.jl for the state variables can greatly enhance the performance. This is done by making u0 a StaticArray and writing an out-of-place non-mutating dispatch for static arrays, for the ROBER problem, this looks like:
using StaticArrays
function rober_static(u, p, t)
y₁, y₂, y₃ = u
k₁, k₂, k₃ = p
du1 = -k₁ * y₁ + k₃ * y₂ * y₃
du2 = k₁ * y₁ - k₂ * y₂^2 - k₃ * y₂ * y₃
du3 = k₂ * y₂^2
SA[du1, du2, du3]
end
prob = ODEProblem(rober_static, SA[1.0, 0.0, 0.0], (0.0, 1e5), SA[0.04, 3e7, 1e4])
sol = solve(prob, Rosenbrock23())retcode: Success
Interpolation: specialized 2nd order "free" stiffness-aware interpolation
t: 61-element Vector{Float64}:
0.0
3.196206628740808e-5
0.00014400709669791316
0.00025605212710841824
0.00048593872520561496
0.0007179482298405134
0.0010819240431281
0.0014801655696439716
0.0020679567767069723
0.0028435846274559506
⋮
25371.934242574636
30784.11997687391
37217.42637183451
44850.61378119757
53893.69155452613
64593.73799781129
77241.71960460565
92180.81944906656
100000.0
u: 61-element Vector{StaticArraysCore.SVector{3, Float64}}:
[1.0, 0.0, 0.0]
[0.9999987215181657, 1.2780900152625978e-6, 3.9181897521319503e-10]
[0.9999942397327672, 5.718510591908378e-6, 4.175664083814195e-8]
[0.9999897579685716, 9.992106860321925e-6, 2.4992456809854223e-7]
[0.99998056266788, 1.7833624284594367e-5, 1.6037078354141815e-6]
[0.9999712826600025, 2.403488607694387e-5, 4.6824539206438825e-6]
[0.9999567250106862, 3.0390689558961382e-5, 1.2884299754832173e-5]
[0.999940798607161, 3.388427373871301e-5, 2.531711910029394e-5]
[0.9999172960308617, 3.583508668595173e-5, 4.686888245237194e-5]
[0.9998862913719596, 3.641240165367128e-5, 7.729622638673522e-5]
⋮
[0.055635077310009065, 2.3546320422346706e-7, 0.944364687226786]
[0.0479253487640473, 2.012149504322489e-7, 0.9520744500210007]
[0.041233421490035026, 1.719278816689166e-7, 0.9587664065820823]
[0.03543699837030739, 1.4688361975260092e-7, 0.9645628547460712]
[0.030425371816163987, 1.2546809318472718e-7, 0.9695745027157405]
[0.026099132201150083, 1.0715608431717981e-7, 0.9739007606427643]
[0.02236969169455009, 9.149844876161108e-8, 0.9776302168069992]
[0.01915856331353317, 7.811096380138723e-8, 0.9808413585755008]
[0.01782789385075171, 7.258919982166322e-8, 0.9821720335600463]If we benchmark this, we see a really fast solution with really low allocation counts:
@btime sol = solve(prob, Rosenbrock23()); 17.660 μs (151 allocations: 18.90 KiB)This version is thus very amenable to multithreading and other forms of parallelism.
Example Accelerating Linear Algebra PDE Semi-Discretization
In this tutorial, we will optimize the right-hand side definition of a PDE semi-discretization.
We highly recommend looking at the Solving Large Stiff Equations tutorial for details on customizing DifferentialEquations.jl for more efficient large-scale stiff ODE solving. This section will only focus on the user-side code.
Let's optimize the solution of a Reaction-Diffusion PDE's discretization. In its discretized form, this is the ODE:
\[\begin{align} du &= D_1 (A_y u + u A_x) + \frac{au^2}{v} + \bar{u} - \alpha u\\ dv &= D_2 (A_y v + v A_x) + a u^2 + \beta v \end{align}\]
where $u$, $v$, and $A$ are matrices. Here, we will use the simplified version where $A$ is the tridiagonal stencil $[1,-2,1]$, i.e. it's the 2D discretization of the Laplacian. The native code would be something along the lines of:
using DifferentialEquations, LinearAlgebra, BenchmarkTools
# Generate the constants
p = (1.0, 1.0, 1.0, 10.0, 0.001, 100.0) # a,α,ubar,β,D1,D2
N = 100
Ax = Array(Tridiagonal([1.0 for i in 1:(N - 1)], [-2.0 for i in 1:N],
[1.0 for i in 1:(N - 1)]))
Ay = copy(Ax)
Ax[2, 1] = 2.0
Ax[end - 1, end] = 2.0
Ay[1, 2] = 2.0
Ay[end, end - 1] = 2.0
function basic_version!(dr, r, p, t)
a, α, ubar, β, D1, D2 = p
u = r[:, :, 1]
v = r[:, :, 2]
Du = D1 * (Ay * u + u * Ax)
Dv = D2 * (Ay * v + v * Ax)
dr[:, :, 1] = Du .+ a .* u .* u ./ v .+ ubar .- α * u
dr[:, :, 2] = Dv .+ a .* u .* u .- β * v
end
a, α, ubar, β, D1, D2 = p
uss = (ubar + β) / α
vss = (a / β) * uss^2
r0 = zeros(100, 100, 2)
r0[:, :, 1] .= uss .+ 0.1 .* rand.()
r0[:, :, 2] .= vss
prob = ODEProblem(basic_version!, r0, (0.0, 0.1), p)ODEProblem with uType Array{Float64, 3} and tType Float64. In-place: true
Non-trivial mass matrix: false
timespan: (0.0, 0.1)
u0: 100×100×2 Array{Float64, 3}:
[:, :, 1] =
11.0147 11.0615 11.0812 11.0281 … 11.0738 11.0938 11.0885 11.0481
11.0809 11.0751 11.0649 11.0792 11.0436 11.015 11.0612 11.0982
11.0375 11.0652 11.0239 11.0511 11.0078 11.0729 11.0692 11.0935
11.053 11.0136 11.013 11.0102 11.0797 11.0099 11.0073 11.0665
11.0737 11.0564 11.019 11.0614 11.0989 11.0 11.0987 11.0107
11.0274 11.0746 11.0616 11.0206 … 11.0334 11.0036 11.069 11.0868
11.0044 11.071 11.0474 11.0556 11.0862 11.0793 11.0365 11.0881
11.0368 11.0988 11.0392 11.0048 11.0905 11.097 11.065 11.0464
11.0191 11.0938 11.1 11.028 11.0802 11.0529 11.0792 11.0641
11.0718 11.0355 11.0016 11.0257 11.0571 11.0565 11.0965 11.0801
⋮ ⋱
11.0942 11.0658 11.037 11.0599 11.0591 11.0594 11.0615 11.052
11.0853 11.0163 11.0953 11.0961 11.0941 11.078 11.0935 11.061
11.0674 11.0041 11.0098 11.0143 11.0143 11.0462 11.0496 11.0633
11.0992 11.0952 11.066 11.005 11.0979 11.0627 11.0021 11.0809
11.0682 11.0137 11.0564 11.0248 … 11.0237 11.0367 11.0958 11.0025
11.048 11.0092 11.049 11.0024 11.0598 11.0564 11.0433 11.0293
11.0829 11.0731 11.05 11.0004 11.0522 11.0395 11.0207 11.0122
11.0457 11.0244 11.0822 11.0544 11.073 11.047 11.0698 11.0131
11.022 11.042 11.0193 11.0588 11.0695 11.0275 11.0275 11.0459
[:, :, 2] =
12.1 12.1 12.1 12.1 12.1 12.1 … 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 … 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
⋮ ⋮ ⋱ ⋮
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 … 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1
12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1 12.1In this version, we have encoded our initial condition to be a 3-dimensional array, with u[:,:,1] being the A part and u[:,:,2] being the B part.
@btime solve(prob, Tsit5()); 52.655 ms (8903 allocations: 186.88 MiB)While this version isn't very efficient,
We recommend writing the “high-level” code first, and iteratively optimizing it!
The first thing that we can do is get rid of the slicing allocations. The operation r[:,:,1] creates a temporary array instead of a “view”, i.e. a pointer to the already existing memory. To make it a view, add @view. Note that we have to be careful with views because they point to the same memory, and thus changing a view changes the original values:
A = rand(4)
@show A
B = @view A[1:3]
B[2] = 2
@show A4-element Vector{Float64}:
0.1738014283814302
2.0
0.9641256851445661
0.17146933637443873Notice that changing B changed A. This is something to be careful of, but at the same time we want to use this since we want to modify the output dr. Additionally, the last statement is a purely element-wise operation, and thus we can make use of broadcast fusion there. Let's rewrite basic_version! to ***avoid slicing allocations*** and to ***use broadcast fusion***:
function gm2!(dr, r, p, t)
a, α, ubar, β, D1, D2 = p
u = @view r[:, :, 1]
v = @view r[:, :, 2]
du = @view dr[:, :, 1]
dv = @view dr[:, :, 2]
Du = D1 * (Ay * u + u * Ax)
Dv = D2 * (Ay * v + v * Ax)
@. du = Du + a .* u .* u ./ v + ubar - α * u
@. dv = Dv + a .* u .* u - β * v
end
prob = ODEProblem(gm2!, r0, (0.0, 0.1), p)
@btime solve(prob, Tsit5()); 46.504 ms (6698 allocations: 119.66 MiB)Now, most of the allocations are taking place in Du = D1*(Ay*u + u*Ax) since those operations are vectorized and not mutating. We should instead replace the matrix multiplications with mul!. When doing so, we will need to have cache variables to write into. This looks like:
Ayu = zeros(N, N)
uAx = zeros(N, N)
Du = zeros(N, N)
Ayv = zeros(N, N)
vAx = zeros(N, N)
Dv = zeros(N, N)
function gm3!(dr, r, p, t)
a, α, ubar, β, D1, D2 = p
u = @view r[:, :, 1]
v = @view r[:, :, 2]
du = @view dr[:, :, 1]
dv = @view dr[:, :, 2]
mul!(Ayu, Ay, u)
mul!(uAx, u, Ax)
mul!(Ayv, Ay, v)
mul!(vAx, v, Ax)
@. Du = D1 * (Ayu + uAx)
@. Dv = D2 * (Ayv + vAx)
@. du = Du + a * u * u ./ v + ubar - α * u
@. dv = Dv + a * u * u - β * v
end
prob = ODEProblem(gm3!, r0, (0.0, 0.1), p)
@btime solve(prob, Tsit5()); 41.997 ms (3758 allocations: 29.86 MiB)But our temporary variables are global variables. We need to either declare the caches as const or localize them. We can localize them by adding them to the parameters, p. It's easier for the compiler to reason about local variables than global variables. ***Localizing variables helps to ensure type stability***.
p = (1.0, 1.0, 1.0, 10.0, 0.001, 100.0, Ayu, uAx, Du, Ayv, vAx, Dv) # a,α,ubar,β,D1,D2
function gm4!(dr, r, p, t)
a, α, ubar, β, D1, D2, Ayu, uAx, Du, Ayv, vAx, Dv = p
u = @view r[:, :, 1]
v = @view r[:, :, 2]
du = @view dr[:, :, 1]
dv = @view dr[:, :, 2]
mul!(Ayu, Ay, u)
mul!(uAx, u, Ax)
mul!(Ayv, Ay, v)
mul!(vAx, v, Ax)
@. Du = D1 * (Ayu + uAx)
@. Dv = D2 * (Ayv + vAx)
@. du = Du + a * u * u ./ v + ubar - α * u
@. dv = Dv + a * u * u - β * v
end
prob = ODEProblem(gm4!, r0, (0.0, 0.1), p)
@btime solve(prob, Tsit5()); 41.655 ms (1247 allocations: 29.66 MiB)We could then use the BLAS gemmv to optimize the matrix multiplications some more, but instead let's devectorize the stencil.
p = (1.0, 1.0, 1.0, 10.0, 0.001, 100.0, N)
function fast_gm!(du, u, p, t)
a, α, ubar, β, D1, D2, N = p
@inbounds for j in 2:(N - 1), i in 2:(N - 1)
du[i, j, 1] = D1 *
(u[i - 1, j, 1] + u[i + 1, j, 1] + u[i, j + 1, 1] + u[i, j - 1, 1] -
4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
end
@inbounds for j in 2:(N - 1), i in 2:(N - 1)
du[i, j, 2] = D2 *
(u[i - 1, j, 2] + u[i + 1, j, 2] + u[i, j + 1, 2] + u[i, j - 1, 2] -
4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
end
@inbounds for j in 2:(N - 1)
i = 1
du[1, j, 1] = D1 *
(2u[i + 1, j, 1] + u[i, j + 1, 1] + u[i, j - 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
end
@inbounds for j in 2:(N - 1)
i = 1
du[1, j, 2] = D2 *
(2u[i + 1, j, 2] + u[i, j + 1, 2] + u[i, j - 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
end
@inbounds for j in 2:(N - 1)
i = N
du[end, j, 1] = D1 *
(2u[i - 1, j, 1] + u[i, j + 1, 1] + u[i, j - 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
end
@inbounds for j in 2:(N - 1)
i = N
du[end, j, 2] = D2 *
(2u[i - 1, j, 2] + u[i, j + 1, 2] + u[i, j - 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
end
@inbounds for i in 2:(N - 1)
j = 1
du[i, 1, 1] = D1 *
(u[i - 1, j, 1] + u[i + 1, j, 1] + 2u[i, j + 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
end
@inbounds for i in 2:(N - 1)
j = 1
du[i, 1, 2] = D2 *
(u[i - 1, j, 2] + u[i + 1, j, 2] + 2u[i, j + 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
end
@inbounds for i in 2:(N - 1)
j = N
du[i, end, 1] = D1 *
(u[i - 1, j, 1] + u[i + 1, j, 1] + 2u[i, j - 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
end
@inbounds for i in 2:(N - 1)
j = N
du[i, end, 2] = D2 *
(u[i - 1, j, 2] + u[i + 1, j, 2] + 2u[i, j - 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
end
@inbounds begin
i = 1
j = 1
du[1, 1, 1] = D1 * (2u[i + 1, j, 1] + 2u[i, j + 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
du[1, 1, 2] = D2 * (2u[i + 1, j, 2] + 2u[i, j + 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
i = 1
j = N
du[1, N, 1] = D1 * (2u[i + 1, j, 1] + 2u[i, j - 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
du[1, N, 2] = D2 * (2u[i + 1, j, 2] + 2u[i, j - 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
i = N
j = 1
du[N, 1, 1] = D1 * (2u[i - 1, j, 1] + 2u[i, j + 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
du[N, 1, 2] = D2 * (2u[i - 1, j, 2] + 2u[i, j + 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
i = N
j = N
du[end, end, 1] = D1 * (2u[i - 1, j, 1] + 2u[i, j - 1, 1] - 4u[i, j, 1]) +
a * u[i, j, 1]^2 / u[i, j, 2] + ubar - α * u[i, j, 1]
du[end, end, 2] = D2 * (2u[i - 1, j, 2] + 2u[i, j - 1, 2] - 4u[i, j, 2]) +
a * u[i, j, 1]^2 - β * u[i, j, 2]
end
end
prob = ODEProblem(fast_gm!, r0, (0.0, 0.1), p)
@btime solve(prob, Tsit5()); 6.663 ms (659 allocations: 29.62 MiB)Notice that in this case fusing the loops and avoiding the linear operators is a major improvement of about 10x! That's an order of magnitude faster than our original MATLAB/SciPy/R vectorized style code!
Since this is tedious to do by hand, we note that ModelingToolkit.jl's symbolic code generation can do this automatically from the basic version:
using ModelingToolkit
function basic_version!(dr, r, p, t)
a, α, ubar, β, D1, D2 = p
u = r[:, :, 1]
v = r[:, :, 2]
Du = D1 * (Ay * u + u * Ax)
Dv = D2 * (Ay * v + v * Ax)
dr[:, :, 1] = Du .+ a .* u .* u ./ v .+ ubar .- α * u
dr[:, :, 2] = Dv .+ a .* u .* u .- β * v
end
a, α, ubar, β, D1, D2 = p
uss = (ubar + β) / α
vss = (a / β) * uss^2
r0 = zeros(100, 100, 2)
r0[:, :, 1] .= uss .+ 0.1 .* rand.()
r0[:, :, 2] .= vss
prob = ODEProblem(basic_version!, r0, (0.0, 0.1), p)
de = complete(modelingtoolkitize(prob))
# Note jac=true,sparse=true makes it automatically build sparse Jacobian code
# as well!
fastprob = ODEProblem(de, [], (0.0, 0.1), jac = true, sparse = true)ODEProblem with uType Vector{Float64} and tType Float64. In-place: true
Initialization status: FULLY_DETERMINED
Non-trivial mass matrix: false
timespan: (0.0, 0.1)
u0: 20000-element Vector{Float64}:
11.099802166242245
11.077758715427608
11.031733936639947
11.097036357617245
11.019942225284012
11.072204680343091
11.094605876269265
11.060392968824996
11.018406095697374
11.087504324706956
⋮
12.100000000000001
12.100000000000001
12.100000000000001
12.100000000000001
12.100000000000001
12.100000000000001
12.100000000000001
12.100000000000001
12.100000000000001Lastly, we can do other things like multithread the main loops. LoopVectorization.jl provides the @turbo macro for doing a lot of SIMD enhancements, and @tturbo is the multithreaded version.
Optimizing Algorithm Choices
The last thing to do is then ***optimize our algorithm choice***. We have been using Tsit5() as our test algorithm, but in reality this problem is a stiff PDE discretization and thus one recommendation is to use CVODE_BDF(). However, instead of using the default dense Jacobian, we should make use of the sparse Jacobian afforded by the problem. The Jacobian is the matrix $\frac{df_i}{dr_j}$, where $r$ is read by the linear index (i.e. down columns). But since the $u$ variables depend on the $v$, the band size here is large, and thus this will not do well with a Banded Jacobian solver. Instead, we utilize sparse Jacobian algorithms. CVODE_BDF allows us to use a sparse Newton-Krylov solver by setting linear_solver = :GMRES.
The Solving Large Stiff Equations tutorial goes through these details. This is simply to give a taste of how much optimization opportunity is left on the table!
Let's see how our fast right-hand side scales as we increase the integration time.
prob = ODEProblem(fast_gm!, r0, (0.0, 10.0), p)
@btime solve(prob, Tsit5()); 1.458 s (60116 allocations: 2.76 GiB)using Sundials
@btime solve(prob, CVODE_BDF(linear_solver = :GMRES)); 643.367 ms (17932 allocations: 118.74 MiB)prob = ODEProblem(fast_gm!, r0, (0.0, 100.0), p)
# Will go out of memory if we don't turn off `save_everystep`!
@btime solve(prob, Tsit5(), save_everystep = false); 4.015 s (90 allocations: 2.90 MiB)@btime solve(prob, CVODE_BDF(linear_solver = :GMRES), save_everystep = false); 1.815 s (42535 allocations: 2.43 MiB)prob = ODEProblem(fast_gm!, r0, (0.0, 500.0), p)
@btime solve(prob, CVODE_BDF(linear_solver = :GMRES), save_everystep = false); 2.646 s (61957 allocations: 3.12 MiB)Notice that we've eliminated almost all allocations, allowing the code to grow without hitting garbage collection and slowing down.
Why is CVODE_BDF doing well? What's happening is that, because the problem is stiff, the number of steps required by the explicit Runge-Kutta method grows rapidly, whereas CVODE_BDF is taking large steps. Additionally, the GMRES linear solver form is quite an efficient way to solve the implicit system in this case. This is problem-dependent, and in many cases using a Krylov method effectively requires a preconditioner, so you need to play around with testing other algorithms and linear solvers to find out what works best with your problem.
Now continue to the Solving Large Stiff Equations tutorial for more details on optimizing the algorithm choice for such codes.