Last month I described several features in the new version of Rybka Aquarium of interest to the serious analyst. This month I’ll continue my coverage of improvements and look a little deeper into how users who prefer infinite analysis can take advantage of IDeA with minimal effort and without the need to study all the options available for automatic IDeA analysis.

HTML Export of iBooks and Games

The first free update of Aquarium 2011 has already been released and it comes with improved HTML export of both iBooks and games. There is a choice of standalone HTML pages that can be uploaded to a Web site or output suitable for inclusion in blog posts. Here is a screen shot of a web page generated from an iBook page with the new version.

You can play through the game by first clicking a move with the mouse and then using the arrow keys on the keyboard.

In order to create a web page from a single iBook page, right-click somewhere on the page to display the following menu.

Here you select “Save as HTML Page.” A standard Windows dialog box will be displayed for you to choose where to save the generated files and the name of the page. After that you can upload the files to a web server to make the page accessible to readers.

In addition to exporting a single page from an iBook, you can also export the whole book along with table of contents. While viewing the book, click the Book Options button in the Ribbon to display the window shown below.

Here you click Export to HTML and you will be given an option to select the output directory for the generated files. Afterward you can upload the whole directory (along with subdirectories) to the web server.

An iBook page can also be exported for use in a blog post. The first step is to right-click somewhere in the page and select “Export to Blog” as shown below.

The generated code and links to the required stylesheets will be displayed.

The stylesheets are only required the first time you export from Aquarium to a blog post. Click the highlighted “Copy links” to copy the stylesheet links. You need to insert these links into your blog template. As an example, here is how you do that on Blogger.com:

Click “Design” (the link is available on the Dashboard among other places).
• Click “Edit HTML” near the top of the page.
• Paste the stylesheet links in the head section (between the head tags).
• Click Save Template.

The stylesheet links are highlighted in the image and, as you can see, I pasted them right below the title tags.

You only need to do this once and now you can create a blog post for displaying the iBook page. This is done by copying the generated code by clicking the “Copy to Clipboard” button in the “HTML export to blog” dialog box shown above. After that create a new blog post and paste the generated code into the post.

The new HTML export features can also be applied directly to a database game or a game list. In that case, Aquarium converts the game(s) to a temporary iBook and then generates the HTML output in the same way as shown above. These new options are available in the Publish tab when you are viewing a game or a game list.

After switching to the Publish tab, click the Web Export button and select “iBook HTML Page” if you want to generate a web page or “iBook HTML for Blog” if you are going to add the output to a blog post. The rest works in exactly the same way as was described above when you export from an iBook.

Upgrade to Rybka 4.1

The latest version of Rybka, (Deep) Rybka 4.1 is included with (Deep) Rybka Aquarium 2011. Besides Rybka, you can use any UCI compatible engine in Aquarium.

ChessOK Playing Club

For the first time the ChessOK Playing Club (also known as Chess Planet) client is now included in the basic Aquarium installation package. Previous versions required a separate download.

ChessOK Playing Club is a chess server where you can play opponents all over the world for free. Besides offering the usual time-controls for shorter games, this site has great support for correspondence chess with new tournaments starting all the time. New in this version is support for Fischerandom or Chess960 games.

CAP

I’ve been looking at the new CAP analysis that comes with Aquarium 2011. For those who are not familiar with the CAP project, it is a chess tree (position database) with chess engine evaluation of fifty-five million positions. Each position is analyzed for one minute and the results are minimaxed. IDeA users will probably realize that the quality of such analysis can be quite good and in general the CAP evaluations after the early opening seem to reflect the possibilities of both sides quite well.

One of the new features of the Aquarium 2011 CAP database is the analysis of the first twenty-five moves of all high-rated correspondence games from the ICCF game archive. Therefore, it is worth keeping an eye on the CAP scores while browsing the games.

Naturally, a high rated correspondence player can’t rely entirely on CAP evaluations when selecting moves. Move selection in correspondence games requires much deeper analysis. However, a key factor in modern correspondence chess is to find positions that are initially evaluated favorably by chess engines, but can be shown to be dangerous after deep and careful analysis. CAP can be helpful in picking out such positions.

Hidden names: Auto invert

Many correspondence players use the “Hidden Names” option to hide their name in the sidebar. When viewing their ongoing games, they only want to see their opponents’ names and which color they are playing. The screenshot below shows an example, where a user is playing two games against the same opponent.

He has hidden his own name so he can clearly see the opponent’s name. The first game is labeled “Smith, John –” with the dash after the name indicating that the user has the black pieces. The second game is labeled “– Smith, John” and now the dash is on the left side indicating that the user has the white pieces.

This feature has an additional option in Aquarium 2011, as shown in the next image. First bring up the Hidden Names window (Aquarium Button > Options > Display Options > Hidden Names).

If you select “Invert board if Black,” the board will be automatically rotated when you open your games so you see it from the side that you are playing.

IDeA: Reduced Analysis Time for Blunders

Many users have noticed that IDeA spends the same amount of time on analyzing good moves and bad moves. If you let IDeA analyze each position deeply, it can be painful to watch a clear blunder analyzed for several minutes. In Aquarium 2011, analysis time for tasks with a bad evaluation is dynamically decreased, so less time is spent on such tasks and there is more analysis time available for good moves.

By default, the evaluation is first checked after three seconds, next after six seconds, then after twelve seconds, etc. The time is doubled with each check.

IDeA: Relative Project Score Bounds

Project score bounds can be used to limit IDeA analysis to positions having a score within a specified interval. If your project score bounds are -50 to 50, then IDeA will only analyze positions with a score from -50 to 50 centipawns. This would be fine while the root score was 0.00. However, if the root score jumps a few hours later to +0.60, then the score bounds would no longer server their purpose. This is solved in Aquarium 2011 by allowing the project score bounds to change automatically depending on the root score (see the IDeA tab of the “IDeA Project Properties”).

By selecting “Relative to root score” the project score bounds will follow changes to the root score. In the example above this means that when the root score changed from 0.00 to +0.60 the project score bounds would be automatically changed to 10 to 110 centipawns so they still cover all positions within 50 centipawns from the new root score.

Useful Keyboard Shortcuts

The following keyboard shortcuts are new in Aquarium 2011:

• Ctrl-F1 – minimizes/maximizes the Ribbon.
• Ctrl-K – is equivalent to clicking the Lock/Unlock button on the Analysis tab while infinite analysis is running.
• Ctrl-G – is equivalent to clicking the menu part of the Unlock button and selecting Go To Locked Position.
• Shift-F6 – displays the previous game (F6 displays the next game and was available in previous versions)

Infinite Analysis

The users that gain most from the improved analysis features in Aquarium 2011 are those who use infinite analysis. With the improvements in Aquarium 2011 you can automatically keep track of all your infinite analysis broken down by chess engine. If you want to check a previously analyzed position at a later time, you can see how it was evaluated by each of the chess engines used to analyze the position.

This image shows analysis that was performed over a month ago. Although I may have completely forgotten about this position, Aquarium hasn’t and shows me the evaluation of all the chess engines that analyzed this position.

Last month I showed how you can write the analysis results of a chess engine into a specific tree. If you want to display the tree in the tree window, you need to add it to a tree configuration (see Introduction to Tree Configurations and Creating Tree Configurations). You can use a special tree configuration for this purpose or you can add the columns with the engine evaluations to an existing tree configuration.

In this example I’m going to add a column with Stockfish to the standard WideBook tree configuration. While viewing a game (either in the Sandbox or Database mode), I select the Tree tab in the Ribbon. First I make sure that WideBook is selected and displayed in the leftmost button. If it isn’t, you can press the button and select WideBook from the drop-down list.

Next, click the Options button as shown in the image. The Tree configurations dialog box will be brought up.

Here you click the Add button displayed in the lower right corner.

There are three fields you need to fill out for the new column.

Tree path points to the tree where you store the Stockfish evaluations. It must be the same tree as you specified in the Engines list for Stockfish (Advanded Engine Options window). In this example I point to the tree Stockfish.hsh in the EngineIA subdirectory of ATrees.

Column type must be Computer evaluations as shown in the screen shot.

Column name can be anything you like. I have chosen “Stockfish,” but normally I let a three character abbreviation suffice, such as “Sto.”

After filling out these fields, click OK and the new column will be added to the WideBook tree configuration. You can repeat this procedure for as many engines as you like if you want to compare the analysis of several engines in the Tree window.

All of the above is strictly based on infinite analysis and doesn’t require any other analysis tools, such as IDeA. However, even if you are a die-hard infinite analysis user, you should consider a simple but effective way of combining IDeA with infinite analysis.

Let’s say you are analyzing a game, using your favorite analysis tool: infinite analysis. Your analysis methods have served you well and you are not interested in switching to something completely different. The following method allows you to analyze exactly like before, but still take advantage of IDeA. While viewing the game you are going to analyze go to the first position in the game that you are interested in (this will become the “root position” of the IDeA project), switch to the Analysis tab and click the downward pointing triangle on the IDeA button.

Select Analyze in Project from the menu and then click New Linked Project as shown above. The IDeA Project Properties will be displayed.

You only need to fill out the following fields:

Project description can be any text that describes the game you are analyzing (Project-1 in this example).

Analysis tree is the tree where IDeA stores its analysis results. I have selected Project-1 as the name of the project tree. It is stored in the IDeA folder of ATrees.

Note that Root positions will be automatically set to the position that you were viewing in the game, so nothing needs to be done there.

Analysis quality specifies for how long or how deeply you want IDeA to analyze positions that are automatically sent to the analysis queue from your infinite analysis. In this example I want each such position to be analyzed for at least ten seconds, but the analysis shouldn’t stop until it has reached at least depth eighteen.

You should select Insert infinite analysis from linked games. What this means is that when you run your normal infinite analysis, the positions you analyze will be sent automatically to the IDeA project. They can even be added automatically to the project tree along with the engine evaluation, depending on the settings you have chosen in the Advanced Engine Options in the Engines list. Min. time and Min. depth specify the minimum requirements for accepting positions from infinite analysis. In this example I have set Min. time to twenty seconds and Min. depth to fifteen. This means that the project will not accept infinite analysis from any engine unless it has analyzed the position for at least ten seconds and also reached depth of at least fifteen. Note that the depth is the calculated depth, which is influenced by “Change analysis depth by” in the Advanced Engine Options (see last month’s column for more information).

Finally, you need to deactivate all other IDeA options. Switch to the IDeA tab and disable Automatic tree expansion (IDeA) as shown below.

This means that IDeA will not generate tasks automatically. The tasks will come from your infinite analysis.

After this click OK and your project will be created by Aquarium.

Let’s first assume that you will only be using one engine for your analysis. In that case you probably want its analysis to be added directly to the IDeA tree. This means that you should switch to Engines mode, select the engine and then select “Add moves and evalutions to IDeA tree” in the Advanced Engine Options. You might even want to select “Add analysis to IDeA tree and generate alternative tasks,” but that would require you to start IDeA once in a while to analyze the alternatives.

Now you can start analyzing exactly as you have always done. You won’t notice any changes, except that the Stage Status window is now visible and the IDeA tabs have been added to the Ribbon. The Tree window also displays the IDeA tree configuration by default. While you analyze Aquarium is working behind the scenes adding positions and evaluations to the IDeA project tree. You will see the results in the tree window.

After analyzing for a while, you may want to minimax the IDeA project tree.

Click the Manage link in the Stage Status window and select Minimax Tree Now. Minimaxing has also been called “backsolving.” What it does is take the evaluation of the endpoint of every variation you have analyzed and propagate it up the tree. As a very simple example, let’s look at the following variation:

27.Kg2 +1.21 g5 +0.75 28.h4 +0.50 h5 0.00.

The engine evaluations are shown after every move. Although the variation starts off nicely for White, the evaluations drop quickly and after Black’s 28…h5 it becomes clear that it leads to an equal position. If the tree consisted only of this variation, the IDeA column in the tree window would show a score of 0.00 for all the moves after minimaxing and you will not have to go to the end of the variation to see that it leads to an equal position. In this case the minimaxed values give you a much more accurate picture than the engine evaluations of individual moves. Of course things get more complicated when you must take alternatives into account, but this example shows the basics of how minimaxing works.

This is one example of the value that simple IDeA usage can add to your infinite analysis process. By minimaxing the project tree once in a while you will get a better overview of your analysis than by looking at the point evaluations returned by infinite analysis.

If you use more than one chess engine for infinite analysis, you must decide which ones add analysis directly to the IDeA tree and which ones send tasks to the analysis queue. This is done in the Advanced Engine Options, which were explained in detail last month. If you let some of the engines send positions to the IDeA analysis queue, you will see them in the Stage Status window.

In this example, there are seven tasks in the analysis queue that were sent from an engine while running infinite analysis. Their color shows that these are “manual” tasks; i.e., tasks that were not created with normal IDeA task generation. These kind of tasks are expected from engines with “Send positions from analysis line to IDeA queue” selected in the Advanced Engine Options.

Besides the type of tasks we saw above, the queue in this image shows three “alternatives” tasks. They come from chess engines with a setting of “Add analysis to IDeA tree and generate alternative tasks.”

If you see tasks being added to the analysis queue, you need to start IDeA to analyze them. The first step is to decide which engine to use as the IDeA analysis engine. You select the engine by switching to the Home – IDeA tab and clicking the Engines button. The Engines Setup dialog box will be displayed.

In this example I’m using Rybka 4 as an analysis engine; two single core Rybka 4 engines and the remote engines on a slower hardware as the “Speed %” column shows. When I have selected the engines, I can start IDeA by clicking the “Start IDeA” button. The IDeA Projects window let’s me select which projects to analyze.

Make sure that only the project you created above is selected (Project-1 in this example). Once you click OK, the analysis starts. Since the project itself doesn’t generate any tasks, it will only analyze the tasks that were put there by the infinite analysis engines. When all the tasks have been analyzed, IDeA won’t continue and you can stop it by clicking the Stop IDeA button. This prevents IDeA from starting analysis of new tasks as they are sent to the queue. It is not desirable to let IDeA analyze at the same time as you are running infinite analysis unless you have enough free CPUs so IDeA isn’t competing with infinite analysis for CPU resources. If you do have enough CPUs, it may make sense to dedicate one (or more) of them to IDeA and have it running all the time. Note that you can use remote engines in IDeA so the IDeA engine(s) can be located on a different computer.

The advantage of this approach is that you can automatically have your favorite engine evaluate all the variations suggested by the other engines you use for analysis.

Finally, it’s worth reminding infinite analysis users of the Infinite button in IDeA.

Even if you are using IDeA in the simple way described here, you can still take advantage of the Interactive buttons, including Infinite. This button creates an infinite analysis task for the current position. You can even use this method to analyze several different positions in the game simultaneously. Each infinite analysis task will run until you stop it manually. For further information about this method of analysis see last month’s column.