[hydromad] Is there any way to speed up the processing?

[Bo-Ram Lee,]

Hi, I'm researcher in South Korea.

First, Thank for developing and sharing hydromad.

Now, I'm working on SWMM parameter auto-calibration using R(Package SWMMR and hydromad). 

Code is attached in this massage.

The question is, like title in this massage, is there any way to speed up the processing?

If the number of variables applied to  auto-calibration  exceeds about 200, the calculation speed becomes very slow.

Is there any way to speed up like multicore processing or GPU processing.

------------------------------------------------------------------------------------------------------------------------------------

install.packages(c("zoo", "latticeExtra", "polynom", "car", "Hmisc","reshape"))
install.packages("hydromad", repos="http://hydromad.catchment.org")
install.packages("swmmr") #install packages
install.packages("readxl")
install.packages("xts")
install.packages("openxlsx")

library(package = "hydromad")
library(swmmr)
library(openxlsx)
library(readxl)
library(xts)

setwd("C:/Users/DS266/Downloads/optimization/test_sh") #Assign working directory

inp_file<-file.path("C:/Users/DS266/Downloads/optimization/test_sh/samho2.inp") #Read initial input file

swmm_file<-run_swmm(inp_file)

obs<-read_xlsx("obs2.xlsx")

obs2<-xts(obs$V1,order.by=obs$...1) #write.csv(as.data.frame(obs),"obs.csv")

inp<-read_inp(swmm_file$inp)

nse <- function(x){1-sum((x[,1]-x[,2])^2)/sum((x[,1]-mean(x[,1]))^2)}

obj_fun <- function(x, inp, obs2){
  inp$subcatchments <- within(inp$subcatchments,{
    Perc_Imperv<-x[1:107]
    Width<-x[108:214]
  })
  inp$subareas<-within(inp$subareas,{
    "N-Imperv"<-x[215]
    "N-Perv"<-x[216]
    "S-Imperv"<-x[217]
    "S-Perv"<-x[218]
    PctZero<-x[219]
  })
  inp$infiltration<-within(inp$infiltration,{
    MaxRate<-x[220]
    MinRate<-x[221]
    Decay<-x[222]
  })
  inp$conduits<-within(inp$conduits,{
    Roughness<-x[223]
  })
#  inp$infiltration<-within(inp$infiltration,{
#    DryTime<-x[224]
#  })
#  inp$subcatchments <- within(inp$subcatchments,{
#    Perc_Slope<-x[225]
#  })

   
#  tmp_inp<-file.path("C:/Users/DS266/Downloads/optimization/test_sh/inp/sh2.inp")
  tmp_inp<-tempfile()
  write_inp(inp, tmp_inp)
  swmm_file<-suppressMessages(run_swmm(tmp_inp, stdout = NULL))
  on.exit(file.remove(unlist(swmm_file)))
  sim<-read_out(swmm_file$out,
                iType = 2,
                object_name = "1036-7500",
                vIndex = 0)[["1036-7500"]]$flow_rate
  nse(merge(obs2,sim))*-1
}


set.seed(1234)

system.time({sce<-SCEoptim(
  obj_fun,
  par=c(inp[["subcatchments"]][["Perc_Imperv"]], #Perc_Imperv
        inp[["subcatchments"]][["Width"]], #Width
        0.03, #N-Imperv
        0.3, #N-Perv
        2.5, #S-Imperv
        5.0, #S-Perv
        10, #PctZero
        254, #Max_R
        25.4, #Min_R
        2, #Decay
        0.015),  #Manning
  lower=c(inp[["subcatchments"]][["Perc_Imperv"]]*0.5, #Perc_Imperv
          inp[["subcatchments"]][["Width"]]*0.5, #Width
          0.011, #N-Imperv
          0.1, #N-Perv
          1.6, #S-Imperv
          3.8, #S-Perv
          10, #PctZero
          76, #Max_R
          2.5, #Min_R
          1, #Decay
          0.011 ),  #Manning
  upper=c(inp[["subcatchments"]][["Perc_Imperv"]]*1.5, #Perc_Imperv
          inp[["subcatchments"]][["Width"]]*1.5, #Width
          0.02, #N-Imperv
          0.35, #N-Perv
          3.8, #S-Imperv
          6.4, #S-Perv
          30, #PctZero
          254, #Max_R
          25.4, #Min_R
          4, #Decay
          0.02),  #Manning
  control = list(
    trace=1,
    ncomplex=20,
    maxit=3),
  inp=inp,
  obs=obs2
  )})

summary(calibration_res)