Select Page

Monolix2016R1 error messages and known bugs list

Bugs have been reported for which there are workarounds

Structural model

  • When you have a structural model using a delay differential equation, if the initial condition of the delayed variable is time varying, then the information is not taken into account. Workaround: It is possible to code the initial condition in a different way as can be seen here.
  • If the initial value of a differential equation is defined by a regressor, it is not well taken into account. Workaround: In the Mlxtran model file, remove the regressor definition, and use instead a “PK:” block with a “depot” macro to add the initial value of your ODE at the desired time. If you do not have a AMT column, you just tag your regressor column X into AMT. Else wise, you need to modify your AMT column to include the regressor value and affect it to another administration route (using the ADM column).
  • For the indirect response model from the PD/PDe libraries (all the turn_input and turn_output model files), the initialization is incorrect. Workaround: You have to add the initial condition E_0 = Rin/kout to your model. You have to change it directly in your library (in monolixSuiteInstallationDir/factory/library/pd and monolixSuiteInstallationDir/factory/library/pde).
  • Error “The model could not be compiled. Error message: Cannot load the library …”: this error usually appears when the keyword inftDose is used. Workaround: use a regressor column containing the infusion duration information

Project definition

  • Error “id:lixoft:Exception … message:assert(son) failed …” when saving: the error appears if you change the name of a categorical covariate without applying any transformation. Note that the first save may work but if the project is loaded and saved again, the error appears. Workaround: Use the button “ADD” in the covariate transformation window. Using that, you can define any transformed covariates, possibly just rename even if no transformation is applied.
  • When selecting the option “mixed method” for parameter without variability in Settings > Pop. parameters > Advanced settings, the choice of the method is not properly saved when saving the project. As a consequence, when reloading a saved project, the  default method “no variability” will be used. Workaround: manually set the method to “mixed method” after having loaded the project
  • Error “A message identifier must be followed by another input argument, of type char, representing the message text” when saving:  Reason 1: this message appears when a covariate transformation has been defined using centering by mean or median, and the data set contains individuals without any observation for some YTYPEs. Workaround: use option “other” and define your covariate transformation, for instance “log(cov/70)”. Reason 2: the same error appears if you have set an ADM column, but not AMT column. Workaround: ignore the ADM column.
  • Error “The (lexicographical) ordering of the data observation type ’10’ does not match the ordering of its associated observation” when saving: this error appears when a project has more than 10 outputs (and 10 YTYPE). There is no workaround for the 2016R1 version, but you can use the 4.3.3 version which does not have this bug.
  • Error “Error when estimating the individual parameters: Attempted to access X(6); index out of bounds because numel(X)=1.” in the console when estimating individual parameters: this error appears for projects with a latent covariate, and parameters that have no influence on some individuals (for instance ka for individuals which receive a bolus dose). Workaround: estimate the individual parameters using the “conditional mean” option.
  • Error “Error: matrix must be definite positive” at the end of the first SAEM phase: this error may appear if you have a latent covariate and have chosen one the methods “add decreasing variability”, “variability in the first stage” or “mixed” for parameters without variability. Workaround: use method “no variability”.
  • Error “Execution error : CAT arguments dimensions are not consistent” at the end of Fisher Information Matrix calculation: error appears for categorical data with IOV. No workaround for 2016R1 version.
  • Error “Execution error: subscripted assigment dimension mismatch” when generating the graphics: this error may happen if some individuals do not have observations for all outputs/YTYPEs. Workaround: reorder your data set such that the last individual of the data set has observations for all outputs
  • Error “Bad lexical cast: source type value could not be interpreted as target” at project load: this error happens when bayesian priors with negative typical value have been defined in the project. Workaround: manually remove the minus signs in the section [POPULATION] DEFINITION of the .mlxtran project file, load and put the minus signs back in the Monolix GUI before running the project. 
  • Error “Execution error: All of the input arguments must be of the same size and shape. Previous inputs had size 1 in dimension 1. Input #3 has size 0.“. This error appears for TTE models, when calculating the individual parameters via mean. Workaround: use the calculation via mode instead


  • Stratification by latent covariate of a mixture model: the stratification works only if the individual parameters have been estimated via “cond. mode” but not via “cond. mean and sd”. In the latter case, there is no error message but all individuals are assigned to the same category.

Output files

  • When using “Export Graphics Data” for prediction-corrected (uppsala correction) VPCs, the outputted observation data points are not prediction-corrected. As a workaround, an R script permits to generate the prediction-corrected output file for most cases (no BLQ data, no stratification). Download here.

Export to Mlxplore

  • column RATE, TINF and ADDL of the data set are not properly exported. The infusions will be replaced by boluses and the additional doses not taken into account. Workaround: manually add “tinf=” or “rate=” and the additional doses in the lines of the [ADMINISTRATION] block. Mlxplore documentation here
  • export of custom parameter distributions is not supported. No error is generated at the time of export but generated Mlxplore script will not compile. It is not possible to use custom distributions with Mlxplore.

Meaning of error messages

  • Error “Cannot merge a regression update, as it would erase a meaningful value” at start of SAEM: this error appear when different regressor values are defined at the time for the same individual. Solution: check our data set; or if you use occasions, and would like to have different regressor values at the same time for different occasions, add an EVID column and set EVID=4 at the beginning of each occasion.
  • Error “There was a problem verifying the data file” when loading the data: the error might come from special characters (such as quotes, etc) in your data file. Solution: Check here for the allowed characters and modify your data set accordingly
  • Error “A model’s first parameter must have one column for each individual parameter. Model’s buffer could be out of date” : this error may appear if you modify the structural model file in an external editor and run tasks without having recompiled the model. Solution: recompile the model
  • Error “Execution error: Vectors must be the same lengths” during graphics generation: for some graphics predictions are calculated for time points not present in the original data set, and possibly for negative times. If your model returns NaNs for some time points (for instance if you output log(Cc) and Cc(t=0)=0 ), this error appears. Solution: use a saturation in your model file, for instance Ccsat=max(1e-6,Cc) and then use log(Ccsat).
  • Error “Execution error: Input #2 expected to be a cell array, was double instead” or  “index exceeds matrix dimension” when generating the graphics: this error appears if you have selected the table “covariate summary” in the “outputs to save” and you have set as CAT a column that have only one covariate value (all individuals have ‘0’ for male for instance). Solution: ignore this column in the data set loading window.
  • Error “Error while stratifying: the information about the covariates is not consistent. Impossible to load stratify, using default” when generating the graphics: This error may appear if you have used “attach current settings to project” and have changed the data set afterwards. Solution: manually remove the file stratify.mat from the result folder and the *_graphics.xmlx file, and load the project again.
  • Error “Parse not ended” when loading a project: If the data set contains header or categorical covariate with modalities containing special characters “.”, “-“, “*”,… (see here for the full list), the project runs fine and can be saved. However, there is an issue when reloading the project. Solution: To be able to reload the saved project, change the special character by a letter (“X” for example). Then, the modified project can be loaded. The categories will be automatically recomputed.
modal close image