Run multiple models with different parameter sets
This example shows how to run multiple RS Minerve models with different parametersets.
The tutorial is adapted from Visual Basic scripts provided by Javier Fluixa Sanmartin (CREALP).
Prerequisites
Install R, RTools, RStudio, RSMinerve and the R package rClr (see README for instructions and links).
Copy this vignette to your local computer and adapt the paths.
Open the model in and load & link the data base with the model objects. If you don’t know how to do this please follow the examples in the RS Minerve User Manual prior to continuing with this tutorial.
Loading Assemblies
# Adapt path in dir_RSM_install to your local installation of RSMinerve.
dir_RSM_install <- "C:/Program Files (x86)/RS MINERVE"
clrLoadAssembly(file.path(dir_RSM_install, 'log4net.dll'))
clrLoadAssembly(file.path(dir_RSM_install, 'Microsoft.Practices.Prism.Mvvm.dll'))
clrLoadAssembly(file.path(dir_RSM_install, 'RSMinerve.RS.dll'))
clrLoadAssembly(file.path(dir_RSM_install, 'RSMinerve.DB.dll'))
clrLoadAssembly(file.path(dir_RSM_install, 'RSMinerve.Base.dll'))Model settings
model_file <- if (file.exists("../inst/extdata/Tutorial_Model.rsm")) {
"../inst/extdata/Tutorial_Model.rsm"
} else {
"https://raw.githubusercontent.com/hydrosolutions/RSMinerveR/blob/main/inst/extdata/Tutorial_Model.rsm"
}
input_dataset <- if (file.exists("../inst/extdata/Tutorial_DataMeteo.dsx")) {
"../inst/extdata/Tutorial_DataMeteo.dsx"
} else {
"https://raw.githubusercontent.com/hydrosolutions/RSMinerveR/blob/main/inst/extdata/Tutorial_DataMeteo.dst"
}
parameter_file <- if (file.exists("../inst/extdata/Tutorial04_Parameters_1.txt")) {
"../inst/extdata/Tutorial04_Parameters_"
} else {
cat("Cannot find file Tutorial04_Parameters_1.txt. Please download from https://github.com/hydrosolutions/RSMinerveR/inst/extdata/. ")
}
ic_file <- if (file.exists("../inst/extdata/Tutorial_IC.txt")) {
"../inst/extdata/Tutorial_IC.txt"
} else {
"https://raw.githubusercontent.com/hydrosolutions/RSMinerveR/blob/main/inst/extdata/Tutorial_IC.txt"
}
saveDataInDstFile <- TRUE
# The RS Minerve documentation describes the date format to be "%d.%m.%Y"
start_date <- "02.09.2013 00:00:00" # format = "%d.%m.%Y %H:%M:%S"
end_date <- "09.09.2013 00:00:00" # format = "%d.%m.%Y %H:%M:%S"
simulationTimeStep <- "600"
recordingTimeStep <- "3600"
timeStepUnit <- "Seconds"The paths for the result files need to be full paths. Below are example paths. You’ll need to adapt these.
results_savePath <-
"C:/Users/<username>/Documents/scriping_rsm/tutorial04-results_"
presimreport_path <-
"C:/Users/<username>/Documents/scriping_rsm/tutorial04-preSimuReport.txt"
postsimreport_path <-
"C:/Users/<username>/Documents/scriping_rsm/tutorial04-postSimuReport.txt"
ic_save_folder <-
"C:/Users/<username>/Documents/scriping_rsm"Look at parameters
params <- NULL
for (f in c(1:5)) {
full_parameter_file <- base::paste0(parameter_file, f, ".txt")
temp <- readRSMParameters(full_parameter_file)
temp <- mutate(temp, "Parameter set" = f)
params <- rbind(params, temp)
}
params |>
filter(Object == "SOCONT") |>
ggplot() +
geom_point(aes(`Parameter set`, Values, colour = Name)) +
facet_wrap("Parameters", scales = "free_y") +
theme_bw()
Edit parameters and write to file
params <- params |>
mutate(Values = ifelse((`Parameter set` == "2" & Parameters == "Tcp1 [deg C]"),
0.01, Values))
params <- params |>
mutate(Values = ifelse((`Parameter set` == "3" & Parameters == "Tcp1 [deg C]"),
0.05, Values))
params <- params |>
mutate(Values = ifelse((`Parameter set` == "4" & Parameters == "Tcp1 [deg C]"),
0.1, Values))
params <- params |>
mutate(Values = ifelse((`Parameter set` == "5" & Parameters == "Kr [m1/3/s]"),
Values/2, Values))
params |>
filter(Object == "SOCONT") |>
ggplot() +
geom_point(aes(`Parameter set`, Values, colour = Name)) +
facet_wrap("Parameters", scales = "free_y") +
theme_bw()
writeRSMParameters(params, paste0(parameter_file_edit, ".txt"))
#> NULLApplying settings and running the model
The commands allow to call the Visual Basics commands documented in the RS Minerve Technical Manual. The calls return NULL when successful. Check your local paths to see if the output files have been created.
# Define a clr task
rsm <- clrNew("RSMinerve.RS.Task")
# Start-up the model
clrCall(rsm, "Start", model_file)
#> NULL
# Load data
clrCall(rsm, "LoadDatasetAndSetDates", input_dataset, TRUE, TRUE)
#> NULL
# Set simulation parameters
clrCall(rsm, "SetDates", start_date, end_date)
#> NULL
# SetSimulationTimeStep and SetRecordingTimeStep require characters
clrCall(rsm, "SetSimulationTimeStep", simulationTimeStep, timeStepUnit)
#> NULL
clrCall(rsm, "SetRecordingTimeStep", recordingTimeStep, timeStepUnit)
#> NULL
# Load initial conditions from file and copy file
clrCall(rsm, "LoadInitialConditionsFromFileAndCopyFile", ic_file, ic_save_folder)
#> NULL
# Load parameters from file & run the model
for (f in c(1:5)) {
full_parameter_file <- base::paste0(parameter_file_edit, "_", f, ".txt")
full_save_results_file <- base::paste0(results_savePath, f, ".dsx")
clrCall(rsm, "LoadParametersFromFile", full_parameter_file)
clrCall(rsm, "Simulate")
clrCall(rsm, "SaveFullResultsAs", full_save_results_file,
saveDataInDstFile)
}
# Stop task
clrCall(rsm, "Stop")
#> NULLRead results
# Calculate the number of time steps to read for each model component, i.e. the
# chunk size. Including the header of each chunk.
chunk_size <- getChunkSize(
lubridate::as_datetime(start_date,format = "%d.%m.%Y %H:%M:%S"),
lubridate::as_datetime(end_date, format = "%d.%m.%Y %H:%M:%S"),
recordingTimeStep
)
result <- NULL
for (i in c(1:5)) {
full_save_results_file <- base::paste0(results_savePath, i, ".dsx")
temp <- readResultDST(full_save_results_file, chunk_size)
temp <- temp |>
dplyr::mutate(`Parameter set` = as.character(i))
result <- rbind(result, temp)
}