triqs::stat::accumulator::constructor
#include <triqs/stat/accumulator.hpp>
Synopsis
Documentation
Template parameters
- T Type of object to be accumulated. Pre-requisites:
- T is a regular type
- T can be set to zero with: T x=0
- T has a multiplication operator (x * x) defined in an element-wise manner
- T can be made real using: nda::real(T)
Parameters
- data_instance An instance of the data type T that will be accumulated. This will be copied and zeroed to initialize the linear and logarithmic parts. Should be set-up so that further data can be correctly added to it (e.g. have the right dimensions and size for an array).
- n_log_bins_max The maximum number of bins to be kept in the logarithmic binning. Possible values are:
- n_log_bins_max == 0: turns off logarithmic binning.
- n_log_bins_max > 0: finite number of bins; the capacity of the largest bin is \(2^{\texttt{n_log_bins_max}}\).
- n_log_bins_max < 0: unbounded number of bins. A new bin of capacity \(2^m\) get created as soon as there are \(2^m\) measurements available.
- n_lin_bins_max The maximum number of data points to be kept by the linear bin. Possible values are:
- n_lin_bins_max == 0: turns off linear binning.
- n_lin_bins_max == 1: when there is only a single linear bin, the accumulator ignores lin_bin_capacity. This is so that all no data which is passed to the accumulator is ignored.
- n_lin_bins_max > 1: imposes a finite maximum bin number, causes automatic compression[REF] of the data when all bins are filled and additional data is being passed to the accumulator
- n_lin_bins_max < 0: unbounded number of bins. A new bin is created when all current bins have reached capacity.
- lin_bin_capacity The number of measurements the linear part will average together in a single bin, before starting a new bin.