New issue with calibrate_sim

[fanxian1998]

The following is part of the code of R language Lanmei :

calib_setup <- setup <- data.frame('pars' = as.character(c('wind_factor','lw_factor','ch','sed_temp_mean','sed_temp_mean',
'coef_mix_hyp','Kw')),
'lb' = c(0.7,0.7,5e-4,3,8,0.6,0.05),
'ub' = c(2,2,0.002,8,20,0.4,0.5),
'x0' = c(1,1,0.0013,5,15,0.5,0.1))
print(calib_setup)
glmcmd = NULL
first.attempt = TRUE
period = get_calib_periods(nml_file,ratio = 2)
period
scaling = TRUE
method = 'CMA-ES'
metric = 'RMSE'
target.fit = 1.5
target.iter = 100
plotting = TRUE
output = out_file
field_file = field_data
calibrate_sim(var, path, field_file, nml_file, calib_setup, glmcmd, first.attempt, period,
scaling, method, metric, target.fit, target.iter, plotting, output, verbose = FALSE)

When running to calibrate_sim, an error occurred, as shown below :
Calibration functions are under development, and are likley to change with future package updates.
[1] "RMSE 4.926"
[1] "RMSE 6.963"
Error in approx(x = elevs_re, y = temps_re, xout = elevs_out) :
need at least two non-NA values to interpolate
In addition: There were 33 warnings (use warnings() to see them)

I want to ask what is causing this and how to solve the problem, I hope someone can answer it, or contact me by email .my mail box :[email protected]

[fanxian1998]

I hope teachers will give suggestions .thank you all !

[jordansread]

@robertladwig can you respond to this ☝️

[fanxian1998]

@robertladwig Hello, do you have any good suggestions? I tried to modify it for a long time, but this problem still can't be solved .

[robertladwig]

that error means that a parameter combination of the calibration algorithm resulted in a GLM run with NaN's meaning that the simulation probably did crash. A cause could be that your lower boundary value (lb) of coef_mix_hyp is higher than your upper boundary value (ub), 0.6 to 0.4

[fanxian1998]

@robertladwig .After reading your reply, I adjusted it as soon as possible, but the same error still appeared. I reorganized it and found a warning. Could it be that it caused the error. The specific warnings are as follows:

period = get_calib_periods(nml_file, ratio = 2) # define a period for the calibration,
Warning messages:
1: In buildVal(textLine, lineNum = j, blckName):
NAs introduced by coercion
2: In buildVal(textLine, lineNum = j, blckName):
NAs introduced by coercion

[fanxian1998]

Hello, teachers. In my continuous attempts, I found that when running the model, this error seems to appear randomly. Whenever there is need at least two non-NA values to interpolate, I open the Lake.csv file obtained by running and it will There are many values such as -nan(ind), and then I run the model once. When there is no error, the Lake.csv file obtained by the model is correct again. However, adjusting the parameters requires the model to run repeatedly, and -nan may appear once. (ind) This situation leads to an error. I don’t know if the situation I saw is helpful to solve the error. I would like to hear your opinions. Looking forward to your reply!

[fanxian1998]

Two teachers, the problem has been solved, it turned out to be a version problem .thank you