defining Game phases

[MahmoudUthman]

in my understanding:
*the transition from opening to middle game is well defined(using opening book).
*but when should the engine consider that the current phase is (End game).
*is there a standard way to define endgame phase, for example I've seen a program that decides that this is endgame if
a-there is no queens (seems wrong?!)
b-one side possess nothing but pawns

*is it be bifacial to define more phases (early mid-game, late mid-game,.. etc. )
*finally is there special consideration to using (Null Move Reductions
,with Null Move Pruning at the same time) and for NM reductions on chess programming wiki it says:

Thus, Null Move Reductions are therefor less vulnerable to Zugzwang and might even applied in (late) endings

does this(might) imply that it's better to turn it off in the endgame ?

[hgm]

Most engines turn NMP off for the side that has not at least one sliding piece. LMR does not make you vulnerable for zugzwang, but it seems counter-productive in end-gaes with only Pawns.

Most engines interpolate the score linearly between two extreme phases (called opening and end-game), based on some weighted sum of non-Pawn material (e.g. Q=6, R=3, B=N=1). If that sum (often referred to as 'game phase') equals zero it uses the pure end-game score, if it equals 40 it uses the pure opening score.

In Joker I distinguished 3 phases for determining King safety: the late end game where it did not care about safety, and was attracted by the center. The opening phase where it cowered on the back rank behind Pawns, and an in-between phase where it was OK to move forward, as long as it stayed behind a Pawn in the same file.

[MahmoudUthman]

Most engines turn NMP off for the side that has not at least one sliding piece. LMR does not make you vulnerable for zugzwang, but it seems counter-productive in end-gaes with only Pawns.

Most engines interpolate the score linearly between two extreme phases (called opening and end-game), based on some weighted sum of non-Pawn material (e.g. Q=6, R=3, B=N=1). If that sum (often referred to as 'game phase') equals zero it uses the pure end-game score, if it equals 40 it uses the pure opening score.

In Joker I distinguished 3 phases for determining King safety: the late end game where it did not care about safety, and was attracted by the center. The opening phase where it cowered on the back rank behind Pawns, and an in-between phase where it was OK to move forward, as long as it stayed behind a Pawn in the same file.

Thanks , which is it better to use (Null move reductions or Null move pruning) at different game phases?