using Random, Plots

using ForwardDiff

using OrdinaryDiffEq, DiffEqSensitivity

using LinearAlgebra

using Statistics

using ProgressBars, Printf

using Flux

using Flux.Optimise:update!

using Flux.Losses: mae, mse

using BSON: @save, @load

using DelimitedFiles

using YAML

using XLSX

using StatsBase

ENV["GKSwstype"] = "100"

conf = YAML.load_file("./config.yaml")

fuel_name = conf["fuel_name"]

expr_name = conf["expr_name"]

fig_path = string("./results/$fuel_name/$expr_name/figs")

ckpt_path = string("./results/$fuel_name/$expr_name/checkpoint")

config_path = "./results/$fuel_name/$expr_name/config.yaml"

is_restart = Bool(conf["is_restart"])

ns = Int64(conf["ns"])

nr = Int64(conf["nr"])

lb = Float64(conf["lb"])

n_epoch = Int64(conf["n_epoch"])

n_plot = Int64(conf["n_plot"])

grad_max = Float64(conf["grad_max"])

maxiters = Int64(conf["maxiters"])

n_train = Int64(conf["n_train"])

batchsize = Int64(conf["batchsize"])

lr_max = Float64(conf["lr_max"])

lr_min = Float64(conf["lr_min"])

lr_decay = Float64(conf["lr_decay"])

lr_decay_step = Int64(conf["lr_decay_step"])

w_decay = Float64(conf["w_decay"])

llb = lb;

global p_cutoff = -1.0

opt = Flux.Optimiser(

ExpDecay(lr_max, lr_decay, n_train * lr_decay_step, lr_min),

ADAMW(0.005, (0.9, 0.999), w_decay),

);

if !is_restart

if ispath(fig_path)

rm(fig_path, recursive=true)

end

if ispath(ckpt_path)

rm(ckpt_path, recursive=true)

end

if ispath("./results/$fuel_name") == false

mkdir("./results/$fuel_name")

if ispath("./results/$fuel_name/$expr_name") == false

mkdir("./results/$fuel_name/$expr_name")

if ispath(fig_path) == false

mkdir(fig_path)

mkdir(string(fig_path, "/conditions"))

if ispath(ckpt_path) == false

mkdir(ckpt_path)

cp("./config.yaml", config_path; force=true)