Bacot lost that because he went to the dumb square.
Question of how to trap the knight in R vs. N

Jesus de la Villa discusses the ending of this game in detail in "100 Endgames You Must Know", at page 43. Bacrot had a drawn ending but blundered his way into the position you posted. Along the way, Kamsky missed several wins and Bacrot missed as many draws.
Incidentally, the R vs N ending usually results from an important drawing mechanism in a R+P ending, from positions like this:

Jesus de la Villa discusses the ending of this game in detail in "100 Endgames You Must Know", at page 43. Bacrot had a drawn ending but blundered his way into the position you posted. Along the way, Kamsky missed several wins and Bacrot missed as many draws.
Incidentally, the R vs N ending usually results from an important drawing mechanism in a R+P ending, from positions like this:
Thanks OldPatzerMike. I have seen this endgame so many times in games by strong players and always heard that it is wrong to seperate the knight from the king. However, my question is whether there is any particular method by which to trap the knight.
Jesus de La Villa does not explain the moves after 12...Nd3. The game continued 13.Rh4. How do we know this move is correct in working towards trapping the knight without necessarily calculating all the tricky knight routes (e.g. under severe time-trouble)?
After 12...Nd3 13.Rh4 Ne5+ Kamsky apparently goes wrong by 14.Ke2:
Jesus de La Villa says: "Just as an example of how difficult this stage is".
Is this endgame practically difficult to play perfectly? Basically, is it enough to know that once as the attacker we get the knight separated from the king we have good winning chances by trying to trap the knight and simply hope for the best?

Jesus de la Villa discusses the ending of this game in detail in "100 Endgames You Must Know", at page 43.
I actually have a chessable version but regret not buying the book instead since the chessable version is probably quite different.

Basically, there is no shorthand calculation with these endgames (like grab opposition or run to the corner). That is why De Villa chose the Kamsky-Borcat game: they missed many opportunities to win and draw and they’re grandmasters. Absent the book and it’s instruction, wed likely make the same mistakes. But, once the knight and king are separated, you just need to position your king and rook to keep them separated while you harass the knight and remove safe squares for it to sit and move to.
i think.

Basically, there is no shorthand calculation with these endgames (like grab opposition or run to the corner). That is why De Villa chose the Kamsky-Borcat game: they missed many opportunities to win and draw and they’re grandmasters. Absent the book and it’s instruction, wed likely make the same mistakes. But, once the knight and king are separated, you just need to position your king and rook to keep them separated while you harass the knight and remove safe squares for it to sit and move to.
i think.
I suppose you are right. I wonder if someone with experience of playing this particular type of endgame (when the knight has been separated) knows whether it is easier for either the attacker or defender to go wrong, or whether the endgame is equally difficult to play correctly for both sides?
I am studying rook vs. knight and the following position. I wonder if there is some systematic procedure of how to trap the knight after 86...Nd3 in the game or is it a case of concrete calculation? Any tips or personal tricks of how to think and play this endgame is greatly appreciated.