====================== - lsqfit.nonlinear_fit has a new keyword: linear. This can be used to specify a list of fit variables that appear linearly in the fit function. These are removed from the fit using variable projection, thereby reducing the number of fit parameters; they are reintroduced after the fit. This parameter is optional and currently somewhat experimental.
- nonlinear_fit methods load_parameters and dump_p are deprecated; use gvar.load and gvar.dump instead. Similarly dump_pmean is deprecated.
9.4
====================== - p0=True generates a random p0 from the prior in nonlinear_fit. This can be useful for fits that are sensitive to the starting point.
- Updated time measurements for Python3.
- Improved makefile and pypi documentation.
- Fixed issue in tests.
9.3
====================== - Adds bootstrapped_fit_iter to MultiFitter (analogous to same for nonlinear_fit).
- Minor bug fixes in nonlinear_fit.bootstrapped_fit_iter.
- Method name bootstrap_iter replaced by bootstrapped_fit_iter, but old name still present, for legacy code.
- Improved setup.py.
9.2
====================== Minor update.
- MultiFitter.make_plot now has a choice of views (ratio, diff, log, ...).
- Minor fix in examples/nist.py to deal with changes in numpy.
- New setup.py to facilitate installations when numpy and cython are not installed. Also cython is now optional. scipy is now required (to guarantee that lsqfit will work even if the gsl library is absent).
- Slight improvement to MultiFitter's plotting capabilities.
- Fixed interface to scipy.optimize.least_squares so ignores if scipy version earlier than 0.17 (used to crash).