PyUp Safety actively tracks 266,461 Python packages for vulnerabilities and notifies you when to upgrade.
* Removed all code from __init__.py and instead used gp_base ,covar_base etc. as module * Moved linear algebra into seprate module * Added working gplvm demo * Moved the demo code in gpr to main, outside the extra function - gradX has changed (TODO: Max. Change the time shift code to reflect this). * Added bounds to optimization * Exponetiation of hyperparameters is now out of the code. The only placed that deside upon exp. are the covariance and the priors. * Changed lnpriors adding Exp prefix that indicates that parameters are in log space * Speedup for cholesky * Added Neils code for cholesky with Jitter * Changes to the covaraince interface * Renamed lml -> LML * Renamed dlml -> LMLgrad ** logtheta -> theta Parmaeters are now jus parameters of the covariance that decides internally how to handle them. ** renamed functions K(theta,x,[x2]) Kdiag (x) Kgrad_theta(x,i) Kgrad_x(x,[x2],d) Kgrad_xdiag(x,d) ** Removed _pointwise_ditance from Covariance Max: why do you need pointwise_distance? * 23.23.2011 (OS) Added an interace for likelihood models to get the noise out of the covariance For now both interfaces are supported. demo_gpr shows how to use the new interface. * 25.03.2011 Corrected use of interval_indices. Interval_indices are the indices on which the gp will predict.