8. Versions and Error Messages

Error Messages

The code has a number of errors that might be thrown at runtime. Some of them are listed below. In particular, the errors that occur during command line parsing (arg_parser) are not listed here.

• (sc_likelihood) ERROR: Log-likelihood calculation failed: Log-likelihood is Nan: The likelihood of the data is maximized starting from the initial set of parameters. However, for initial parameters far away from the optimum the evaluation of the likelihood becomes numerically inaccessible. Thus, if Nan occurs in during the maximization of the likelihood, the code is stopped and it is necessary to set new (and ideally more accurate) initial parameters. The issue might also occur if the initial guess is fine, but the first step (given by the step in the parameter file) kicks the calculation out of the range in which the likelihood can be evaluated. In this case, the Nans occur in the later entries in the iteration file and one needs to set new initial steps. Note, that also the sign of the step matters.
• (minimize_wrapper) ERROR: Log likelihood optimization failed: Something went wrong during the optimisation of the likelihood. Just let me know.
• (build_cell_genealogy) ERROR: Both daughter pointers are set: Each parent cell is expected to have 0, 1, or 2 daughter cells, in case there are more daughter cells associated with a parent cell, the code is stopped. The setting of parent_tags and cell_tags in the csv_config file might not be set correctly, meaning the combination of the tags is still ambiguous pointing to different cells. Revise the parent_tags and cell_tags.
• (get_data) ERROR: (...) is not a column in the input file: The keyword marking the column that is used for getting the time/log_length or fp from the input file is not found in the header. Make sure the keyword matches the name of the column in the input file.
• (get_data) ERROR: Line ... cannnot be processed: A line in the input file contains an entry (in time/log_length or fp) that is not a number. Note, that Nans are also rejected.
• (set_paramter) ERROR: Parameter settings of ... cannot be processed: The processing of the parameter file failed. The formatting of the parameter file is explained above. In particular, all arguments of the parameter setting need to be numbers. Note, that Nans are also rejected.
• (check_if_complete) ERROR: Parameter ... not found in parameter file: Each parameter needs to be set in the parameter file. The code is stopped if one (or more) is missing.

Changed in 0.2.0

• rescale_time default is 1
• time in the prediction file is also rescaled making the prediction file consistent
• tol default is 1e-3
• prior for the first time point is set according to the parameters making stationery and use_beta obsolete
• joint calculation and output of joints
• Use LN_NELDERMEAD

Changed in 0.3.0

• introduction of a scaled noise model

Changes in 0.4.0

• introduction of binomial cell division model
• cell is growing and producing between n and n+1 even if it is dividing, division happend just before n+1
• joint matrix has as identical row and column indices

Changes in 0.4.1

• changed tolerance defaults to 1e-10

Changes in 0.4.2

• taking care of prior correctly

Changes in 0.4.3

• only returns full posterior

Changes in 0.4.4

• use absolute value of scaled error variance to cope with negative input

Changes in 0.4.5

• do not remove .0 type ends of cell ids

Changes in 0.4.6 (current version)

• removed 1d parameter scans from options
• introduction of sampling of trajectories