[SciPy-User] ANN: Lmfit 0.8.0

[Matt Newville]

Hi Folks,

Lmfit 0.8.0 has been released, and is available from PyPI and github:
    https://pypi.python.org/pypi/lmfit/
    http://lmfit.github.io/lmfit-py/

Lmfit provides a high level approach of least squares minimization and
curve fitting based on the routines from scipy.optimize.  The key idea is
to use Parameter objects that can be bounded, fixed, or algebraically
constrained in place of floating point variables.  Many additional features
to make minimization problems easier and better are included.  Lmfit is
MIT-licensed and a pure Python module.  It requires scipy version 0.13 or
later.

Lmfit version 0.8.0 includes several bug fixes and improvements.  The most
important new feature is a Model class for high level curve-fitting
problems, currently emphasizing 1-D functions.  The Model class, largely
the work of Daniel Allen with substantial input from Antonino Ingargiola,
wraps a model function that simulates some data.  It includes methods to
create parameters from function arguments, to fit to data, and to evaluate
models.  More than 20 pre-built models for line shapes such as Gaussian and
Exponential are included.  An important feature of Model is that they can
be added together, making it very easy to construct complex models.

Automated testing with nose and Travis-CI is greatly improved.  There are
over 100 tests, many of these checking the numerical results for
non-trivial fits.  All of the NIST StRD datasets are tested, requiring that
NIST certified values be found (with fairly loose precision) from at least
one of the NIST-provided starting values.

Daniel Allen also started an IPython GUI Fitter, providing a very nice tool
for fitting simple line-shapes to 1 dimensional data.

Finally, we've added a lmfit-py google group for questions about usage,
discussions of design, and announcements.  We're happy to have
conversations about ideas for improving lmfit, or minimization and
curve-fitting routines with scipy in general, on any appropriate forum.

Thanks,

--Matt Newville
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20140922/db2ba393/attachment.html>