Logging

logging

Custom Logging.

This module provides extensions of Pythons's logging capabilities for run and debug logs within MLDA runs. The more elaborate logging routines take Statistics objects, which makes their evaluation and formatted output more convenient.

Classes:

Name	Description
LoggerSettings	Data class storing settings for the logger
Statistic	Basic statistics object containing information for logging
RunningStatistic	Statistic for computing batch averages
AccumulativeStatistic	Statistic for computing overall averages
MTMLDALogger	Logger for the MLDA sampler
DebugFileHandler	Custom file handler for debug logging

mtmlda.core.logging.LoggerSettings dataclass

Data class storing settings for the logger.

Attributes:

Name	Type	Description
do_printing	bool	Decides if the default logger prints to the console, default is True
logfile_path	str	Directory to log run statistics to, default is None
debugfile_path	str	Directory to log debug statistics to, default is None
write_mode	str	Write mode for the log files (append, overwrite), default is 'w'

mtmlda.core.logging.Statistic

Basic statistics object containing information for logging.

Every statistics object has a string identifier and a format string for the value it stores. The value attribute is not accessed directly, but through getter and setter methods. The idea behind is that more sophisticated logic can be implemented in subclasses, still adhering to the generic interface.

Attributes:

Name	Type	Description
str_id	str	Identifier for the statistic
str_format	str	Format string for the value

Methods:

Name	Description
set_value	Set the value of the statistic
get_value	Get the value of the statistic

__init__

__init__(str_id: str, str_format: str) -> None

Constructor of the statistic.

Parameters:

Name	Type	Description	Default
str_id	_type_	Identifier for the statistic	required
str_format	_type_	Format string for the value	required

set_value

set_value(value: float | np.ndarray) -> None

Set the value of the statistic.

get_value

get_value() -> int | float | np.ndarray

Get the value of the statistic.

mtmlda.core.logging.RunningStatistic

Bases: Statistic

Statistic for computing batch averages.

The running statistics stores all values provided by the set_value method and computes their average when the get_value method is called.

__init__

__init__(str_id: str, str_format: str) -> None

Constructor, see base class for details.

set_value

set_value(new_value: float | np.ndarray) -> None

Set the value of the statistic.

get_value

get_value() -> np.ndarray

Get the value of the statistic.

mtmlda.core.logging.AccumulativeStatistic

Bases: Statistic

Statistic for computing overall averages.

The mean value is computed from all provided values.

__init__

__init__(str_id: str, str_format: str) -> None

Constructor, see base class for details.

set_value

set_value(new_value: float | np.ndarray) -> None

Set the value of the statistic.

get_value

get_value() -> np.ndarray

Get the value of the statistic's running average.

mtmlda.core.logging.MTMLDALogger

Logger for the MLDA sampler.

This custom logger wraps the standard Python logger, adding some convenience methods for logging on different levels to different files and the console.

Methods:

Name	Description
log_run_statistics	Log run statistics
log_debug_statistics	Log debug statistics
log_header	Log the header of the run statistics table
log_debug_new_samples	Log the start of a new chain segment in the debug log
log_debug_tree_export	Log the export of a tree in the debug log
info	Log an info message to the run logger
debug	Log a debug message to the debug logger
exception	Log an exception to all loggers

__init__

__init__(logger_settings: LoggerSettings) -> None

Constructor of the logger.

Initializes the run and debug log handles, depending on the user settings.

Parameters:

Name	Type	Description	Default
logger_settings	LoggerSettings	User settings	required

log_run_statistics

log_run_statistics(statistics: dict[str, Statistic]) -> None

Log statistics into run log table.

Parameters:

Name	Type	Description	Default
statistics	dict[str, Statistic]	Run statistics object to log	required

log_debug_statistics

log_debug_statistics(info: str, statistics: dict[str, Statistic]) -> None

Log statistics into debug file.

Parameters:

Name	Type	Description	Default
info	str	Event to log statistics for	required
statistics	dict[str, Statistic]	Statistics to log	required

log_header

log_header(statistics: dict[str, Statistic]) -> None

Print our the header for the run statistics table.

Parameters:

Name	Type	Description	Default
statistics	dict[str, Statistic]	Statistics to print header for	required

log_debug_new_samples

log_debug_new_samples(sample: int) -> None

Log divider into debug file, indicating new fine-level sample.

Parameters:

Name	Type	Description	Default
sample	int	Number of the new sample	required

log_debug_tree_export

log_debug_tree_export(tree_id: int) -> None

Note in debug file that Markov tree with given id has been exported.

Parameters:

Name	Type	Description	Default
tree_id	int	ID of the tree that has been exported	required

info

info(message: str) -> None

Wrapper for Python logger info call.

Parameters:

Name	Type	Description	Default
message	str	Info message to log	required

debug

debug(message: str) -> None

Wrapper for Python logger debug call.

Parameters:

Name	Type	Description	Default
message	str	Debug message to log	required

exception

exception(message: str) -> None

Wrapper for Python logger exception call.

Parameters:

Name	Type	Description	Default
message	str	Exception message to log	required

_process_value_str

_process_value_str(value: Any, str_format: str) -> str

Format a numerical value as string, given a suitable format.

if the provided value is None, it is formatted as np.nan. If the value is iterable, all values are concatenated as comma-separated list, each with the provided format.

Parameters:

Name	Type	Description	Default
value	Any	Value to format as string	required
str_format	str	format to use	required

Raises:

Type	Description
TypeError	If the value is of an unsupported type

Returns:

Name	Type	Description
str	str	Value as formatted string

mtmlda.core.logging.DebugFileHandler

Bases: logging.FileHandler

Custom file handler for Logger.

This file handler only transfers messages on the DEBUGlevel of python logging.

__init__

__init__(filename: Path, mode: str = 'a', encoding: str = None, delay: bool = False)

Constructor of the file handler.

Parameters:

Name	Type	Description	Default
filename	Path	File to log to	required
mode	str	Write mode for log messages. Defaults to "a".	'a'
encoding	str	Special encoding for messages. Defaults to None.	None
delay	bool	Determines if file opening is deferred until first emit call. Defaults to False.	False

emit

emit(record: logging.LogRecord) -> None

Transfer a log message.

Parameters:

Name	Type	Description	Default
record	logging.LogRecord	Log message object	required