Skip to content

Latest commit

 

History

History
33 lines (26 loc) · 2.17 KB

Version_History.md

File metadata and controls

33 lines (26 loc) · 2.17 KB

Version 1.0.0

  • Effective automation of the procedures described in Kingston and Millane (2022), IUCrJ, 9, 648-665.
  • Envelope and phase determination steps are parallelized.

Version 1.1.0

Major changes

  • Envelope comparison and clustering now allows for inversion symmetry.
  • The way the apodization function is specified has changed.
  • Checks on the magnitude of missing terms are now based entirely on the Wilson model.
  • Agreement of the data with the Wilson model is reported.
  • New procedures introduced for controlling the amplitudes of missing data during Fourier space projection onto the constraints
  • A much more flexible method of defining the update rules and constraints was introduced.
  • During DBSCAN clustering of envelopes, "k-distances" are now evaluated. Optionally use the k-distance distribution to set threshold value ε.
  • Internal handling of Fourier data was reworked. Fourier coefficients beyond the current "effective resolution limit" are now removed, speeding up the calculations with heavily apodized data.
  • Ouput from analysis of overall scale and B-factor (method of Rogers) is now reported in the summary log file for the job.
  • Parameterization of the envelope determination step has changed, significantly improving algorithm performance for some test cases.

Version 1.2.0

Major changes

  • Code restructuring to allow user to define the major steps to be perfomed and how the inputs and outputs of those steps are handled.
  • Updated parameter file syntax, allowing better control of what happens at each iteration.
  • Better procedures for detection and removal of outliers in the input diffraction data.
  • Addition of RRR and RAAR algorithms and their reversed variants.
  • Implementation of procedures to add error to an existing phase set using appropriate circular probability distributions.
  • Changes to the way the apodzation function is applied.
  • Better procedures for handling random number generation, facilitating algorithm comparisons.
  • Algorithms optimized to increase speed of execution.
  • Added ability to average phases over the algorithm trajectory, following convergence to the solution, and to calculate the phase retrival transfer function.