Changes in version 1.4.2 (2025-09-28)                  

  - confint(), summary() and other methods no longer fail in case of a
    singular gradient
  - Fixed bug: missing confidence/prediction intervals in predict() for
    multi-start gsl_nls() call with fn defined as a function in
    combination with newdata.

                 Changes in version 1.4.1 (2025-01-17)                  

  - Fixed compatibility GSL versions < 2.5

                 Changes in version 1.4.0 (2025-01-16)                  

  - Robust loss optimization added in gsl_nls() via argument loss
  - weights in gsl_nls() accepts a matrix (in addition to a vector) in
    which case the objective function is generalized least squares
  - Added new function gsl_nls_loss()
  - Added new method cooks.distance()
  - Minor changes in predict() and hatvalues() for weighted NLS

                        Changes in version 1.3.3                        

  - Fix standard errors predict() when using newdata

                 Changes in version 1.3.2 (2024-05-01)                  

  - Reverted to static Makevars.win (supplied by T. Kalibera)
  - Added new method hatvalues()

                        Changes in version 1.3.1                        

  - Minor edits configure.ac to fix cran check results

                 Changes in version 1.3.0 (2024-04-24)                  

  - Missing starting values/ranges allowed in gsl_nls()
  - lower and upper parameter constraints included in gsl_nls()
  - Added 3 regression problems from Bates & Watts (1988)
  - Updated multi-start algorithm in gsl_nls()
  - Added configure.win, cleanup.win and Makevars.win.in
  - Removed old Makevars and Makevars.win
  - Several minor changes

                 Changes in version 1.2.0 (2023-12-11)                  

  - Added multi-start algorithm to gsl_nls()
  - Added 56 NLS regression and optimization test problems
  - Added unit tests in folder unit_tests
  - Several minor changes/fixes

                 Changes in version 1.1.1 (2021-12-13)                  

  - Clean exits gsl_nls() and gsl_nls_large() when interrupted
  - Default algorithm in gsl_nls_large() set to "lm"

                 Changes in version 1.1.0 (2021-11-26)                  

  - Added large-scale NLS regression with gsl_nls_large()

                 Changes in version 1.0.2 (2021-10-13)                  

  - Added Makevars.ucrt

                 Changes in version 1.0.1 (2021-10-11)                  

  - Initial release