Releases: CMA-ES/pycma
Releases · CMA-ES/pycma
r3.0.0
- constraints handling with an Augmented Lagrangian, see
cma.fmin_con
x0
andsigma0
cannot be a string anymore, butx0
may be acallable
(a function)- injections are excluded from negative covariance matrix updates addressing #124
- improved generic Logger, see
cma.logger.Logger
- a wrapper as scikit optimizer, use
import cma.wrapper
- plotting works now interactively also in a Jupyter notebook
logger.plot(addcols=1)
plots additionally correlation and precision matrices (experimental)- new
tolflatfitness
termination option (replaces warnings/errors) - new
tolfunrel
termination option (deactivated by default) - "signals" file is active by default but ignored in verbose=-9 setting
- various other fixes and small improvements
available at PyPI.
r2.7.0
CMADataLogger
code moved to its own logger module- output files go into a separate folder
outcmaes
by default - improved matplotlib 3 compatibility
contour_data
andsemilogy_signed
helper functionsfitness_transformations.EvalParallel2
class (different interface)fitness_models.LQModel
introduced
available at PyPI.
r2.6.1
Nothing special here, some fixes, some small things added.
r2.6.0
The initial solution x0
can be a callable now, which is useful e.g. for the restarted IPOP
version.