Closed foolyc closed 6 years ago
if this bug not fixed, the function plot
in python/gps/gui/plotter_3d.py
will set sample in policy sample which out of the axis limits to NaN :
xs[np.any(np.c_[xs < xlim[0], xs > xlim[1]], axis=1)] = np.nan
ys[np.any(np.c_[ys < ylim[0], ys > ylim[1]], axis=1)] = np.nan
zs[np.any(np.c_[zs < zlim[0], zs > zlim[1]], axis=1)] = np.nan
This will lead to the failure of policy optimization without deepcopy of policy sample(the original code does not use deepcopy yet).
LGTM. Thanks!
there is an error in calculate the axis limits for GUI plotting which do not consider the policy sample