For those who don't know, pv stands for Principal Variation.
When analyzing games, it is "probably" best to use multi-pv mode, and look at least two moves / variations.
Reason. Sometimes many moves "win" or are so close in strength it doesn't matter what move is chosen.
Say you're analyzing one of your own games, for example, and you see that Nf6+ gives a high score for you. Well, so does a completely different move and so does another move. You obviously have a strong position, with a bunch of different ways of converting this advantage and bringing home the point.
Now let's say your analyzing one of your own games, and the first PV shows Nf6+ with a high score (for you), and the NEXT best move shows a very low score for you. Here, Nf6+ is the ONLY move for you and the only move you should consider playing.
But you wouldn't know either of these things if your PV is only set to 1. Setting it to 2 or 3 can give you more information about the position.
However, it really depends upon what you want to do. Analyzing in multi-pv mode means the engine will take more time to come up with the best move. It will be using some of this time exploring the other variations, some of which will often be sub optimal.
Stockfish 4 (or any engine) shouldn't give "funny" lines in mulit PV mode. All it should do is give you the top x lines, for whatever you chose x to be.
Sure, if an engine doesn't display the best move for you in single PV mode, it might display it in multi PV mode because it's displaying more lines! But it still won't consider it best. It might display it but it will still consider its first PV as what it thinks is the best move. And yes, there are still lots of positions that engines don't fully "understand" yet, and will not give you the best move. This is why you should often analyze with more than one engine. I came across a position over at Talk Chess the other day that Komodo didn't understand at all, yet every other engine I tested this position with understood instantly, and gave the proper score.
lately, i have glanced upon a thread : positions which engines get wrong(something very similar)
in that thread, there is a game between a famous master ( i guess it was shirov) where the engines could not find the best move which wins immediately.
for this abnormality, the reason was that engines when allowed to work in SINGLE PV mode could not find the best move( sacrifice in the aforesaid game) but when the same engine was allowed to work in mutli PV MODE, it found the sacrifice which wins.( they were talking about engines like houdini and stockfish)
another which i have heard is that stockfish4 ( which is a free engine) gives some funny lines( wrong line) when it is allowed to work in multi pv mode or its strength decreases.
i have stockfish4 which i use it for analysing my game . i use it with arena(GUI)
so, my question is :
while analysing a game, should we allow engines to work in single PV mode or multi PV?
if we use single PV mode, then it may not be able to find the best move(sacrifices which wins)
if we use multi PV mode, then it will give some funny lines which may not justify the best move selected by the engine.
so, what should i have to do fix this problem ?
kindly advice me on this aspect.