コンテンツにスキップ

Chapter 1 Overview

1.1 Purpose of the App

AutoStatSpectra is an application for decomposing and analyzing spectral data based on Bayesian estimation.

  • From an observed spectrum, it estimates the number of peaks (model comparison) and obtains the posterior distributions of each peak parameter (position, width, intensity) in a single run

What you can do

  • Load spectral data in multiple formats, including CSV / TXT / VAMAS / PHI SOFH / Excel
  • Analyze a single spectrum, or run a group analysis that applies the same prior to multiple spectra
  • Compare models over a specified search range for the number of peaks (automatic selection of the optimal number)
  • Compare multiple candidate functions: Gaussian / Lorentzian / pseudo-Voigt
  • Estimate the background by Bayesian estimation (Constant / Linear / Shirley) or specify it manually
  • Estimate the background with arPLS
  • Set priors individually for each peak (Set per peak mode)
  • Export results as a ZIP archive, plot images, and PDF reports

What this manual does not cover

  • Details of the analysis theory (the mathematical background of Bayesian estimation and Free Energy)
  • The app's internal implementation and source code
  • Customization for site-specific workflows

1.2 Editions and Analysis Time

AutoStatSpectra is available in two editions: a Web edition and a Desktop edition.

Item Web edition Desktop edition
Usage environment Web browser Desktop app
Analysis engine GPU CPU
Where analysis runs GPU backend Local PC
Usage management Account, billing, GPU analysis time License key
Network Required Required for license verification

Analysis modes

  • Single analysis: analyze one spectrum by configuring Peak / Offset / Noise / Sampling
  • Group analysis: apply the same prior to multiple spectra, with results kept per dataset

Analysis speed

In the Web edition, the analysis time depends on GPU backend load, the number of data points, the number of peaks, and the number of samples. The Desktop edition depends on the CPU performance of your PC. When adopting the app, verify the analysis time with your own data.