We learned about “discovering a position”, a basic feature of Fritz13’s “Let’s Check” analysis system in the last blog post. This time around in USCFSales’ blog we’ll tackle another “Let’s Check” feature: analysis of complete games.
First we need to take a little side jaunt to explain another piece of “Let’s Check” terminology: the idea of “winning” or “conquering” a position (you’ll see both words used more or less interchangably). If you decide to have your engine analyze a position and, after you log on to the “Let’s Check” server, you see that someone else has already analyzed it, you can still get point credit for contributing your own analysis if you “win” the position. This requires you to have your engine analyze the position to a much greater depth than did the already-listed engine (and, honestly, if you’re running a single-core chess engine, this may well require that you run an analysis of several hours on just that one single position).
In addition to single position analysis, Fritz13’s “Let’s check” allows you to analyze complete games by fetching existing information stored by other users on the master server, or you can also combine this existing analysis with chess engine analysis you generate on your own computer locally. Let’s learn how to do it…
The first step is to load a game by double-clicking on it in the game list, in order to display it on the main chessboard screen. Next go to the Analysis menu and find the “Let’s Check Analysis” button displayed in the ribbon:
Clicking on that button displays the following dialogue:
This dialogue lists three types of game analysis which you can perform using the “Let’s Check” server. The first type is the simplest to understand: “Retrieval” simply fetches and displays any existing information from the server, and doesn’t use your own engine at all (ergo, you can ignore the “Time” settings in the dialogue). After selecting a “Retrieval” analysis, you’ll see the cursor move back and forth through the game notation for a couple of passes; after a moment or two, you’ll see something like this:
Every move which has already been analyzed by a user has analysis attached to it. The analysis displayed includes a three ply variation, the numerical evaluation (shown from White’s perspective, negative numbers therefore indicate a Black advantage), and the name of the chess engine which created the analysis. If you’ve checked the “User names” box in the initial dialogue, the names of the users who have discovered or won the positions will be displayed. In the illustration above, none of the positions show user names; this means either that the analysis was performed anonymously (which seems to be the case with a couple of these moves, as they display 20+ ply searches) or that the analysis wasn’t deep enough to allow the user to “discover” the position.
In the above illustration, few of the positions have analysis attached to them – yet. Over time, as more and more users contribute analysis to the “Let’s Check” server, the “Retrieval” form of analysis will become a more valuable tool and a huge time-saver; we’ll show a graphic illustration later in this article.
The second analysis form in the “Let’s Check analysis” dialogue is “Standard analysis”. Selecting this option means that your own chess engine will analyze each move in the game and will contribute its own analysis to the “Let’s Check” server. If no analysis of the position already exists on the server, you can possibly “discover” positions as the analysis progresses (although this can be very time consuming – a typical “discovery” analysis using a multi-processor engine on a dual core computer takes about seven to eight minutes per position on average; as always, your results may vary somewhat depending on the hardware used).
Here’s what the output looked like when I ran an analysis session with “Minimal” set to “30” seconds and “Maximal” set to “60” seconds:
I’ve drastically reduced the font size just to be able to show more of the game. We can see that this game has already been analyzed multiple times before, as existing engine analysis is attached to each move. My own engine’s analysis is also displayed with each move, but mine appears in half-tone since the sixty second maximum analysis time I specified is much too shallow by “Let’s Check” standards:
Notice, too, that the Playchess user names are in this case appended to each bit of analysis.
The top line of the game notation provides an interesting additional piece of information:
We’re provided with a comparison of how well the moves suggested by the chess engine match the players’ actual moves. In this case, the engine’s suggestions agree with 23% of White’s choices and with 59% of Black’s moves (which is not surprising, given that Black won the game).
Being as this game had already been fully analyzed by “Let’s Check” users, let’s run a “Retrieval” analysis on this game just to compare what that output looks like with the “Standard” analysis we’ve already seen for this game:
The software took about twenty seconds, give or take, to come up with what we see in the above illustration, graphically demonstrating what a time saver this will be after many thousands of positions are analyzed by “Let’s Check” users.
Here’s a typical “Let’s Check” analysis in “Standard” mode, showing the results for a game which had not yet been analyzed by other users:
Since the game’s positions had not previously been analyzed by other users, only my engines’ analysis is displayed.
The third “Let’s Check” analysis form is “Win Variations”. In this case, each position will be analyzed to a depth sufficient for you to either “discover” the position (if no one has yet analyzed it) or for you to “win” an already-analyzed position (assuming your “time” settings are high enough). Note that this can potentially be very time-consuming and is best suited for users who are running multi-processor chess engines on a multi-processor/multi-core computer. In this case you’d want to set a high “Minimal” value (perhaps “600” [ten minutes]) and a very high “Maximal” value (“900” [fifteen minutes] or more). If you’re the first to analyze a position, you’ll be credited as the position’s “discoverer”. If you’re not the first to analyze the position, you’ll score points for “winning” the position if your engine’s search depth is significantly deeper than the deepest existing search attached to that position. Note that when you set a really high “Maximal” time setting, your engine won’t necessarily analyze for that full time period – only long enough to either “discover” or “win” a position, after which the engine will move to the next position and begin to analyze it.
One last setting should be mentioned: the “Text comments” box. The “Let’s Check” server allows users to manually attach text notes to a position. If this box is checked, other users’ text notes will be displayed in the analysis output.
Have fun! – Steve Lopez
Chessplayers who have purchased their ChessBase brand chess computer software from USCFSales can receive free technical support and advice on their purchases straight from me; just shoot me an e-mail (email@example.com), but please remember to include the USCFSales order number from your ChessBase software purchase. – Steve
Copyright 2011, Steven A. Lopez. All rights reserved.