search

DoseResponse / drc

Fitting dose-response models in R
https://doseresponse.github.io/drc/
21 stars 16 forks source link

Silence drm output #7

Open nlubock opened 5 years ago

nlubock commented 5 years ago

I'm trying to silence all drm output, but it appears to be ignoring the options set in drmc. For example:

dose <- c(
  1.69e-10,1.69e-10,1.69e-10,1.69e-10,
  5.08e-10,5.08e-10,5.08e-10,5.08e-10,
  1.52e-09,1.52e-09,1.52e-09,1.52e-09,
  4.57e-09,4.57e-09,4.57e-09,4.57e-09,
  1.37e-08,1.37e-08,1.37e-08,1.37e-08,
  4.12e-08,4.12e-08,4.12e-08,4.12e-08,
  1.23e-07,1.23e-07,1.23e-07,1.23e-07,
  3.7e-07,3.7e-07,3.7e-07,3.7e-07,
  1.11e-06,1.11e-06,1.11e-06,1.11e-06,
  3.33e-06,3.33e-06,3.33e-06,3.33e-06,
  1e-05,1e-05,1e-05,1e-05
)

response <- c(
  0.286,0.302,0.235,0.362,
  -0.08,-0.479,0.508,0.435,
  0.004,0.676,0.04,-0.751,
  -0.789,-1.053,0.389,-0.095,
  0.326,-0.226,0.038,0.137,
  0.406,-0.223,0.066,0.743,
  -0.113,0.061,-0.244,-0.237,
  0.658,0.146,-0.191,-0.106,
  -0.636,0.145,0.043,0.58,
  0.119,-0.219,0.262,-0.226,
  0.101,0.004,0.163,0.269
)

set.seed(42)
control <- drmc(errorm = FALSE, noMessage = TRUE, warnVal = -1)
foo <- drm(response ~ dose, fct=LL.4(), control = control)

results in the output:

Error in optim(startVec, opfct, hessian = TRUE, method = optMethod, control = list(maxit = maxIt,  : 
  non-finite finite-difference value [4]
In addition: Warning message:
In log(dose/parmMat[, 4]) : NaNs produced
Warning message:
Convergence failed. The model was not fitted!

Session Info

R version 3.5.2 (2018-12-20)
Platform: x86_64-pc-linux-gnu (64-bit)
Running under: Arch Linux

Matrix products: default
BLAS: /usr/lib/libblas.so.3.8.0
LAPACK: /usr/lib/liblapack.so.3.8.0

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C              
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8    
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8   
 [7] LC_PAPER=en_US.UTF-8       LC_NAME=C                 
 [9] LC_ADDRESS=C               LC_TELEPHONE=C            
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C       

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] bindrcpp_0.2.2  forcats_0.4.0   stringr_1.3.1   dplyr_0.7.8    
 [5] purrr_0.2.5     readr_1.3.1     tidyr_0.8.2     tibble_2.0.1   
 [9] ggplot2_3.1.0   tidyverse_1.2.1 magrittr_1.5    furrr_0.1.0    
[13] future_1.11.1.1 broom_0.5.1     drc_3.0-6       drcData_1.0-0  
[17] sandwich_2.5-0  glmnet_2.0-16   foreach_1.4.4   Matrix_1.2-15  
[21] MASS_7.3-51.1   edgeR_3.24.3    limma_3.38.3   

loaded via a namespace (and not attached):
 [1] httr_1.4.0        jsonlite_1.6      splines_3.5.2     carData_3.0-2    
 [5] modelr_0.1.2      gtools_3.8.1      assertthat_0.2.0  cellranger_1.1.0 
 [9] yaml_2.2.0        globals_0.12.4    pillar_1.3.1      backports_1.1.3  
[13] lattice_0.20-38   glue_1.3.0        digest_0.6.18     rvest_0.3.2      
[17] colorspace_1.4-0  plyr_1.8.4        pkgconfig_2.0.2   listenv_0.7.0    
[21] haven_2.1.0       mvtnorm_1.0-8     scales_1.0.0      openxlsx_4.1.0   
[25] rio_0.5.16        generics_0.0.2    car_3.0-2         withr_2.1.2      
[29] TH.data_1.0-10    lazyeval_0.2.1    cli_1.0.1         survival_2.43-3  
[33] crayon_1.3.4      readxl_1.3.0      fansi_0.4.0       nlme_3.1-137     
[37] xml2_1.2.0        foreign_0.8-71    tools_3.5.2       data.table_1.12.0
[41] hms_0.4.2         multcomp_1.4-8    munsell_0.5.0     locfit_1.5-9.1   
[45] plotrix_3.7-4     zip_1.0.0         compiler_3.5.2    rlang_0.3.1      
[49] grid_3.5.2        iterators_1.0.10  rstudioapi_0.8    gtable_0.2.0     
[53] codetools_0.2-15  abind_1.4-5       curl_3.3          R6_2.4.0         
[57] zoo_1.8-4         lubridate_1.7.4   utf8_1.1.4        bindr_0.1.1      
[61] stringi_1.2.4     parallel_3.5.2    Rcpp_1.0.0        tidyselect_0.2.5
jpquast commented 4 years ago

I have the same problem. Is there any update on this? I iterate over drm with purrr::map and use tryCatch to continue after an error occurs. This works. However, the error message is still printed, which messes up my progress bar.

OnofriAndreaPG commented 4 years ago

Hello, the error message comes from ‘optim()’ that is wrapped inside drc. I do not think it can be silenced from drc.

jpquast commented 4 years ago

Thank you for your answer. I already noticed that the problem is optim(). As far as I could see, it is called at least in drmOpt() which is called in drm(). However, it looks like the error of optim() is intended to be silenced since it is wrapped in try() with the option silent = TRUE. So this seems to be a bug. The only intended error message here seems to be "Convergence failed" which is part of drmOpt(). 

drmOpt <- function (opfct, opdfct1, startVec, optMethod, constrained, warnVal, 
    upperLimits, lowerLimits, errorMessage, maxIt, relTol, opdfct2 = NULL, 
    parmVec, traceVal, silentVal = TRUE, matchCall) 
{
    options(warn = warnVal)
    if (is.null(opdfct2)) {
        hes <- TRUE
    }
    else {
        hes <- FALSE
    }
    psVec <- abs(startVec)
    psVec[psVec < 1e-04] <- 1
    {
        if (!is.null(opdfct1)) {
            if (constrained) {
                nlsObj <- try(optim(startVec, opfct, opdfct1, 
                  hessian = hes, method = "L-BFGS-B", lower = lowerLimits, 
                  upper = upperLimits, control = list(maxit = maxIt, 
                    reltol = relTol, parscale = psVec)), silent = silentVal)
            }
            else {
                nlsObj <- try(optim(startVec, opfct, opdfct1, 
                  hessian = hes, method = optMethod, control = list(maxit = maxIt, 
                    reltol = relTol, parscale = psVec)), silent = silentVal)
            }

.....
OnofriAndreaPG commented 4 years ago

Hello,

I do not think that the error in ‘optim()’ can be silenced by using the try() construct, as it originates from an external C-based routine and it is directly printed to the console (as far as I know). So I am incline to think that this is not a bug in drm. However, if you find a solution to silence it (perhaps from the authors of drc?), I’ll be happy to learn that.

All the best

Andrea

On Jul 9, 2020, at 8:52 AM, jpquast notifications@github.com wrote:

Thank you for your answer. I already noticed that the problem is optim(). As far as I could see, it is called at least in drmOpt() which is called in drm(). However, it looks like the error of optim() is intended to be silenced since it is wrapped in try() with the option silent = TRUE. So this seems to be a bug. The only intended error message here seems to be "Convergence failed" which is part of drmOpt().

drmOpt <- function (opfct, opdfct1, startVec, optMethod, constrained, warnVal, upperLimits, lowerLimits, errorMessage, maxIt, relTol, opdfct2 = NULL, parmVec, traceVal, silentVal = TRUE, matchCall) { options(warn = warnVal) if (is.null(opdfct2)) { hes <- TRUE } else { hes <- FALSE } psVec <- abs(startVec) psVec[psVec < 1e-04] <- 1 { if (!is.null(opdfct1)) { if (constrained) { nlsObj <- try(optim(startVec, opfct, opdfct1, hessian = hes, method = "L-BFGS-B", lower = lowerLimits, upper = upperLimits, control = list(maxit = maxIt, reltol = relTol, parscale = psVec)), silent = silentVal) } else { nlsObj <- try(optim(startVec, opfct, opdfct1, hessian = hes, method = optMethod, control = list(maxit = maxIt, reltol = relTol, parscale = psVec)), silent = silentVal) }

..... — You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/DoseResponse/drc/issues/7#issuecomment-655937508, or unsubscribe https://github.com/notifications/unsubscribe-auth/AIXJNIUV75XOWZH4N6VJEFTR2VSK3ANCNFSM4GYZHRZQ.