|
|
|
|
|
|
by BlueTemplar
832 days ago
|
|
|
I had never realized the necessity of nearest even. I thought that 0.5 ought to be rounded to 1, mirroring 0.0 rounded to 0 ! But I can now see how this only works for reals, while anything non-continuous (even rationals ?) could end up with compounding errors... |
|
|