Sergash: <32...Rxd1+ 33.Nxd1 
> It would have been preferable to avoid this trade by 32...Ra2 = or 32...Rb2 = Stockfish 7 - 64 bits Popcnt.<35...Bf8?!> Looks harmless, but better was 35...h5 / =. Stockfish 7 - 64 bits Popcnt.
<36.g3> Sorensen missed 36.Ke1! 
followed with Ke1-d2 and Nd3. Stockfish 7 - 64 bits Popcnt.
<37...g6?!> An energetic play was required here : 37...h5! 38.Rc1 h4! 39.gxh4 Bxh4 40.Kg2! = / Stockfish 7 - 64 bits Popcnt.
<38...Kd7?!>
A) 38...Rc8 39.gxf4 exf4 = / Stockfish 7 - 64 bits Popcnt.
B) 38...h5 39.Kf2! fxg3+ 40.Kxg3 Stockfish 7 - 64 bits Popcnt.
<39...Ke6?! 40.Nd3 > 39...g5! Stockfish 7 - 64 bits Popcnt.
<40...Bg5?!> Here we call this type of simple and evident threat that is not part of a logical plan a "cheapo"... 40...g5 Stockfish 7 - 64 bits Popcnt.
<41.g4> An easier win was obtained with 41.h4! Be7 42.gxf4 and now if 42...Bxd3 43.f5+! gxf5 44.Kxd3 
with the capture of the h4/pawn inviting Rc2-h2 and Rxh7 or, in some cases, Rh2-h6+. Stockfish 7 - 64 bits Popcnt.
<42.Nb4?! Rc8 > The win was obtained like this : 42.Rb2! (threat Rb2xb5! that would obtain 2 passed pawns) Bxd3 (42...Bc4 43.Kc3 Bxd3 44.Kxd3 would transpose into the main variation) 43.Kxd3 Rc8 (or 43...Bxc5 44.Bxc5 Rxc5 45.Rb6+ Rd7 46.Rxa6 Stockfish 7 - 64 bits Popcnt) 44.Kc4! Stockfish 7 - 64 bits Popcnt.
<49...Bf2?? 50.c6+ > Losing the bishop on c7. Moves like 49...Bg5 , 49...Bd8 or 49...Rc6 still had a chance to draw, despite that White has a big advantage.
Kibitzing
>
Stockfish 7 - 64 bits Popcnt.
