You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, a sample is drawn from the specified distribution and if it lies out of the specified bounds of desired values it is plainly minmaxed to the next bound. This creates singular points of concentrated probability weight, which is very ugly.
The correct way to draw a sample is to draw again from the original distribution until the drawn value lies in the specified range. This requires a check that the given interval of admissible values is not empty and perhaps a maximal number of tries until the simple fallback of cutting the value is applied, if no sample within the bounds is found.
The text was updated successfully, but these errors were encountered:
Currently, a sample is drawn from the specified distribution and if it lies out of the specified bounds of desired values it is plainly minmaxed to the next bound. This creates singular points of concentrated probability weight, which is very ugly.
The correct way to draw a sample is to draw again from the original distribution until the drawn value lies in the specified range. This requires a check that the given interval of admissible values is not empty and perhaps a maximal number of tries until the simple fallback of cutting the value is applied, if no sample within the bounds is found.
The text was updated successfully, but these errors were encountered: