Issue with demo code on the README.md for Training a Neural ODE

[wuphilipp]

Hi! I'm trying to run the Nerual ODE demo code and have run into this issue.

Specifically, when I run the demo code under the section Training a Neural Ordinary Differential Equation (exact code block below, directly copied from the README.md), I get the following error

MethodError: no method matching isless(::Tuple{Float32,RecursiveArrayTools.DiffEqArray{Float32,2,Array{Array{Float32,1},1},StepRangeLen{Float32,Float64,Float64}}}, ::Float32)
Closest candidates are:
  isless(!Matched::Float32, ::Float32) at float.jl:464
  isless(!Matched::Missing, ::Any) at missing.jl:87
  isless(::Tuple, !Matched::Tuple{}) at tuple.jl:358
  ...

Stacktrace:
 [1] <(::Tuple{Float32,RecursiveArrayTools.DiffEqArray{Float32,2,Array{Array{Float32,1},1},StepRangeLen{Float32,Float64,Float64}}}, ::Float32) at ./operators.jl:268
 [2] macro expansion at /home/phil/.julia/packages/DiffEqFlux/YKKwl/src/train.jl:113 [inlined]
 [3] macro expansion at /home/phil/.julia/packages/ProgressLogging/g8xnW/src/ProgressLogging.jl:328 [inlined]
 [4] (::DiffEqFlux.var"#23#28"{var"#59#61",Int64,Bool,Bool,typeof(loss_n_ode),Array{Float32,1},Zygote.Params})() at /home/phil/.julia/packages/DiffEqFlux/YKKwl/src/train.jl:43
 [5] with_logstate(::DiffEqFlux.var"#23#28"{var"#59#61",Int64,Bool,Bool,typeof(loss_n_ode),Array{Float32,1},Zygote.Params}, ::Base.CoreLogging.LogState) at ./logging.jl:398
 [6] with_logger at ./logging.jl:505 [inlined]
 [7] maybe_with_logger(::DiffEqFlux.var"#23#28"{var"#59#61",Int64,Bool,Bool,typeof(loss_n_ode),Array{Float32,1},Zygote.Params}, ::LoggingExtras.TeeLogger{Tuple{LoggingExtras.EarlyFilteredLogger{ConsoleProgressMonitor.ProgressLogger,DiffEqBase.var"#10#12"},LoggingExtras.EarlyFilteredLogger{Base.CoreLogging.SimpleLogger,DiffEqBase.var"#11#13"}}}) at /home/phil/.julia/packages/DiffEqBase/k3AhB/src/utils.jl:259
 [8] sciml_train(::Function, ::Array{Float32,1}, ::ADAM, ::Base.Iterators.Cycle{Tuple{DiffEqFlux.NullData}}; cb::Function, maxiters::Int64, progress::Bool, save_best::Bool) at /home/phil/.julia/packages/DiffEqFlux/YKKwl/src/train.jl:42
 [9] top-level scope at In[9]:45

I'm using the latest stable verion of julia (1.4.0), running on ubuntu 16.04.

Any advice?

Really cool work by the way!

---

Code block I'm running

using DiffEqFlux, OrdinaryDiffEq, Flux, Optim, Plots

u0 = Float32[2.; 0.]
datasize = 30
tspan = (0.0f0,1.5f0)

function trueODEfunc(du,u,p,t)
    true_A = [-0.1 2.0; -2.0 -0.1]
    du .= ((u.^3)'true_A)'
end
t = range(tspan[1],tspan[2],length=datasize)
prob = ODEProblem(trueODEfunc,u0,tspan)
ode_data = Array(solve(prob,Tsit5(),saveat=t))

dudt2 = FastChain((x,p) -> x.^3,
            FastDense(2,50,tanh),
            FastDense(50,2))
n_ode = NeuralODE(dudt2,tspan,Tsit5(),saveat=t)

function predict_n_ode(p)
  n_ode(u0,p)
end

function loss_n_ode(p)
    pred = predict_n_ode(p)
    loss = sum(abs2,ode_data .- pred)
    loss,pred
end

loss_n_ode(n_ode.p) # n_ode.p stores the initial parameters of the neural ODE

cb = function (p,l,pred;doplot=false) #callback function to observe training
  display(l)
  # plot current prediction against data
  if doplot
    pl = scatter(t,ode_data[1,:],label="data")
    scatter!(pl,t,pred[1,:],label="prediction")
    display(plot(pl))
  end
  return false
end

# Display the ODE with the initial parameter values.
cb(n_ode.p,loss_n_ode(n_ode.p)...)

res1 = DiffEqFlux.sciml_train(loss_n_ode, n_ode.p, ADAM(0.05), cb = cb, maxiters = 300)
cb(res1.minimizer,loss_n_ode(res1.minimizer)...;doplot=true)
res2 = DiffEqFlux.sciml_train(loss_n_ode, res1.minimizer, LBFGS(), cb = cb)
cb(res2.minimizer,loss_n_ode(res2.minimizer)...;doplot=true)

[ChrisRackauckas]

Just ran it and it ran fine. Possibly the issue is your package versions. If you do ]st what does it show you?

[wuphilipp]

Thanks for the reply. This is what I get from ]st.

(@v1.4) pkg> st
Status `~/.julia/environments/v1.4/Project.toml`
  [31a5f54b] Debugger v0.6.4
  [aae7a2af] DiffEqFlux v1.8.0
  [41bf760c] DiffEqSensitivity v6.10.0
  [0c46a032] DifferentialEquations v6.12.0
  [587475ba] Flux v0.10.3
  [7073ff75] IJulia v1.21.1
  [429524aa] Optim v0.20.5
  [1dea7af3] OrdinaryDiffEq v5.32.1
  [91a5bcdd] Plots v0.29.9
  [37e2e46d] LinearAlgebra

Does this look OK?

[ChrisRackauckas]

Looks like DiffEqFlux 1.8 had an issue, so I just put out a patch 1.8.1. If you update that should be good. Sorry for the hiccup.

[wuphilipp]

No worries, thanks for the quick patch!