gmisclib.mcmc

read_currentlist(fname, trigger=None)

Starting from a file name, read in a log file.

Parameters:

fname (str) - name of a (possibly compressed) log file.
trigger - where to start in the log file. See readiter.

Returns: tuple(list(dict(str:str)), dict(str:str))

a list of data lines and header information. A data line is a dictionary, containing parameters and other information from one iteration.

Note: this is very similar to newstem2.newlogger.read_currentlist except that returns a more complicated data structure that has one more level of indirection.

read_multisample(fname, Nsamp=10, tail=0.0, trigger=None, reader=<function read_currentlist at 0x465eed8>)

source code

Read in a log file: it gives you multiple samples from the log. In other words, it provides a time-series of the changes to the parameters.

Parameters:

tail - Where to start in the file? tail=0 means that you start at the trigger or beginning. tail=1-epsilon means you take just the last tiny bit of the file.
reader (function(str, trigger) -> (list(dict(str:str)), dict(str:str))) - A function that sucks in the data from the log file, yielding a list of iterations and an overall header.

Returns: (dict(str:str), list(indexer), list(float))

Raises:

NoDatError - when the log file is empty.
BadFormatError - when a parameter appears mid-way through the optimization or an iteration cannot be read.

Classes
	WildNumber
	logger_c This class is called from the Markov-Chain stepper and writes the log file.
	logger_A
	logger_N
	lti_c
	NoDataError

Variables
	__package__ = `'gmisclib'`

Module mcmc_logger

read_currentlist(fname, trigger=None)

read_multisample(fname, Nsamp=10, tail=0.0, trigger=None, reader=<function read_currentlist at 0x465eed8>)