Skip to content

Latest commit

 

History

History
78 lines (50 loc) · 2.65 KB

qtb_compare.md

File metadata and controls

78 lines (50 loc) · 2.65 KB

qtb_compare

Compares benchmarks of QT methods based on qtb_analyze raw outputs. The results are sorted according to the sample size benchmark.

← Back to contents

Usage

  • report = qtb_compare(results, tcode) returns the compare report based on a set of qtb_analyze raw outputs results for a specified test codename tcode
  • report = qtb_compare( ___ ,Name,Value) specifies additional arguments using one or more Name-Value pairs

Data type: cell array

A cell array of QT methods raw analysis results. Each element of the array is a path to analysis result file or the result structure array.

Data type: char

The codename of the test to compare.

Data type: cell array

A cell array of QT methods display names. If the array is empty display names are based on the name field of results structure arrays.

Default: {}

Data type: double

Error rate to compare QT methods bemchmarks.

Default: 1e-3

Data type: char

Plot type for infidelity vs sample size plot. Available values:

  • 'none' - do not make a plot
  • 'percentile' - percentile plot
  • 'stats' - median plot with 25-th and 75-th percentiles errorbar
  • 'mean' - mean value plot

Default: 'none'

Data type: double

The percentile value for QT method benchmarks comparison and percentile plot.

Default: 95

Data type: logical

Show theoretical lower bound in the compare table and plot.

Default: true

Data type: char

Path to file to export the compareison report as a table. Available formats: *xls, *xlsx, *csv.

Default: 'none'

Data type: structure array

Function output report is a structure array containing report data:

  • report.names - QT methods names
  • report.fields - cell array of benchmarks names (see benchmarks page)
  • report.data - array of benchmarks values: each row contains benchmark values for a given QT method (sorted by 'NSamples' field)
  • report.table - table array of benchmarks values
  • report.figure - figure object (if plot is not 'none')