0.9.2

Changelog 0.9.1-0.9.2

Fixed

• In the toolbox-free version of DeerLab, when dealing with sub-optimal numerical conditions, the lmlsqnonlin solver would lead to (apparently) random crashes. This have been fixed and the solver no longer crashes midst optimization.

• Improved speed of the lmlsqnonlin solver making the toolbox-free version faster.

See all commits for further details.

0.9.1

Changelog 0.9.0-0.9.1

Added

• All fit functions now accept arbitrary signal scaling, i.e. the condition V(0)=1 is no longer required. The scale is internally optimized by the different fit functions and does not need to be pre-processed. The correctscale function is now partially obsolete and has been labeled as a legacy function in the documentation.
• fitsignal: added interface for lower and upper parameter bounds
• correctphase: added compatibility with 2D datasets, for e.g. phase-correction of multiple measurement scans with a phase drift

Changed

• fitparamodel: using the 'MultiStart' option with unconstrained parameters now prompts an error
• correctscale: a model can now be specified to fit the scale as well as a time interval
• fitsignal: optional key-value pairs can now be passed for all allowed input schemes. Before, the full input scheme had to be passed before being able to specify options.

Fixed

• fitparamodel: fixed error appearing when evaluating multiple confidence levels for the model fit confidence intervals
• deerload: now handles 2D-datasets correctly
• fitsignal: now returns the confidence intervals for the fitted distribution, background and signal as outputs
• Fixed errors in the 'User Guide' section of the documentation
• fitregmodel: degrees of freedom used for calculated chi2red values are now properly estimated

Removed

• fitsignal: removed internal pre-processing steps

See all commits for further details.

0.9.0

This pre-release present many major changes with respect to the last releases. See commits or documentation for more information.

0.8.beta

Major changes with respect to 0.6.beta

Functions

• fitparamodel
  the open-access solvers used to substitute the Optimization toolbox no longer default to 'nlsqbnd' to avoid compatibility issues with Linux and MacOS systems

• sensitivan
  re-factorization of the output variables. Instead of being returned individually, all outputs are collected in two structures.

• dipolarkernel
  preliminary implementation of the multipathway model (beta)

Parametric models

• rd_wormchain, rd_randcoil, rd_onerice,rd_tworice,rd_threerice
  models corrected from math errors in the implementation and bugs leading to errors

• rd_wormgauss
  added parametric model for worm-like chain with Gaussian convolution

Documentation

• major overhaul of the design and structure of the webpage

• examples have been substituted by tutorials, with in-depth descriptions of the code

• adapted missing or outdated information on the API documentation

Important bug fixes

• dipolarsignal & dipolarkernel
  single times can now be passed to both functions and time-axes are now longer required to be strictly increasing

• sensitivan
  fixed interpolation errors of the percentile function, which were leading to negative values of the distance distributions

0.6.beta

Major Changes with respect to 0.5.beta:

• MATLAB toolbox dependencies removed

  • Optimization toolbox no longer required to run the program

• Physical units system removed

  • Time-domain variables are given in microseconds
  • Distance-domain variables are given in nanometers

• Sensitivity analysis function improved, now...

  • called sensitivan
  • returns the main efffect and interaction of the individual factors
  • further general in terms of output types validated

0.5.beta

Major Changes with respect to 0.2.beta:

• Physical units system re-vamped
  - Heuristics now cover a most of all experimental cases.
  - If operated outside of those ranges, units can be specified via new specunits function.

• L-curve from selregparam:
  - The residual and penalty vectors computed during selregparam can be requested as output for constructing an L-curve.
  - Example added showcasing this functionality

0.1-beta

small edits

0.2.beta

added setup instructions to docs