CG Scheveningen and Simultaneous tournaments

This is still very preliminary, as I’m still crafty both the programming tools, and the terminology I want to use.

But yesterday, <Retireborn> reminded me that my effort to categorize tournaments should also include Simuls and Scheveningens. I had intended to get to Simuls, to look for missing games, but Scheveningens entirely slipped my mind – possibly because they’re so few of them, and as far as I can tell, the few are all complete.

But both Simuls and Scheveningens (hereafter abbreviated Schevens), share the property that their numerology is distinctive and fun. Well, maybe not as distinction as I’d like for Schevens, but fun nonetheless.

Simuls are very easy, as a missing game will only be the domain of a diligent biographer. For our purposes all Simuls are, almost by definition, complete. They are single round games, where one player is against all the others. So the numbers are very distinctive. We can make them uniquely so by assuming the no fewer than three opponents show up.

Hence:  N_games = N_players – 1, with no overlap between a simul and a complete Swiss or Scheveningen.

I can tighten up the requirement by also requiring the term “Simul” or “Exhibition” to show up in the tournament name. This yields the following list of Simuls in the <CG> TI:

  83432    16     15   Janowski Exhibition Series at MCC(*) (1899) // New York USA   (1899.01.23)
  39935     6      5                         Belzberg Simul (2003) // London ENG     (?)
  43198    18     17               Kasparov Sao Paulo Simul (2004) // Sao Paolo BRA  (2004.08.21)
  75344     5      4                    Topalov Clock Simul (2011) // Dublin IRL     (2011.10.03)
  77385     9      8             Topalov Vienna Clock Simul (2012) // Vienna AUT     (2012.04.27)

[(*) slight edit to avoid word-warp, MCC = Manhattan Chess Club]

So, the tid, N_players, and N_games are listed first, followed by “official” tournament name (TN), then the Site and EventDate tags (these tournaments have all been normalized on my end).

The reason Simuls are fun, is that they’re easy to spot. And rather uncomplicated. A simple nomenclature would be to designate a Simul by the number of games – Simul-17 for Kasparov’s simul. If a GM were to face teams for each board, like Capablanca’s 1931 world record setting simul in NYC, we could denote the number of opponents/team:  4-Simul-200. (I forget the exact number at the moment, but it was 4 members/team).

Scheveningens are not quite so easy, starting with the fact they’re harder to spell! But they have a distinctive numerology – the players divide up into two teams, and each member of a team plays every other member of the team at least once. Like RRn’s, Scheven’s can have multiple rounds.

What’s a little interesting is that square roots are involved. Consider,

< Armenia – The Rest of the World (2004)>     N_players = 12,   N_games = 36

The <CG> intro gives no information, but the tournament name suggests the meeting of two teams. And so, if we divide the players into two camps of 6 each, and have each player meet every player of the other team, we expect to have 6×6 = 36 games. Which is indeed the case.

Of course, the players could meet each member of the other team with both colors, for a total of 72 games, or any multiple of 36 actually. Just like for RR’s. So, to turn it around, we are looking for tournaments where the number of games divided by some integer yields an integer square root, equal to half of the total number of players.

Let’s look at the <CG> Tournament Index for such tournaments:

>>> for t in [ t for t in ncT if t in Scheven ]:
       print "%s     %s " \
          % ("%6d  %3d  %4d   %2d  Scheven-%d" % Scheven[t], TN[t] )

 79955   10    25    9  Scheven-1     Hastings (1927/28)
 79961   10    25    9  Scheven-1     Hastings (1929/30)
 41225   12    36    6  Scheven-1     Armenia - The Rest of the World (2004)
 44539    8    16    0  Scheven-1     Indochess Man - Machine (2005)
 53650   10    50   10  Scheven-2     Youth - Experience (2006)
 62641   10    50   10  Scheven-2     NH Chess Tournament (2007)
 62642   20   100   10  Scheven-1     Russia - China Match (2007)
 65065    4    20    6  Scheven-5     Chess Classic Mainz (2008)
 65385   10    50   10  Scheven-2     NH Chess (2008)
 66935    4    16    6  Scheven-4     Aker Chess Challenge (2009)
 69335   10    50   10  Scheven-2     Rising Stars - Experience (2009)
 72039   10    25    5  Scheven-1     China - Russia (Women) (2010)
 72038   10    25    5  Scheven-1     China - Russia (2010)
 72041   10    50   10  Scheven-2     Rising Stars - Experience (2010)
 72163    4    16    6  Scheven-4     Arctic Securities Chess Stars (2010)
 75170   10    25    5  Scheven-1     Kings - Queens (2011)
 76295    8    32    8  Scheven-2     Snowdrops and Old-hands (2011)
 78163   10    25    5  Scheven-1     Russia - China (2012)
 78164   10    25    5  Scheven-1     Russia - China (Women) (2012)
 78640   10    50   10  Scheven-2     CEG vs Legends (2012)
 81335   60   900    0  Scheven-1     FIDE World Blitz Championship (2013)

This is a list of tid, N_players, N_games, R_max, where R_max is found by scanning all the games to find the maximum Round number given in the PGN. Next comes the Scheven classification, and then the tournament name.

It appears that my classification algorithm needs refinement, as all the blued entries are unlikely candidates. The N_players / N_games just combined to make a coincidental hit.

The R_max provides additional information that can be used to determine if the tournament really was a Scheven. That is, if the info is provided (see below). When it is provided, it should match (N_players/2)*Scheven-number.

For example: <Kings–Queens (2011)> has 10-players, or 5/team. That means 25 games for a Scheven-1, and we’d expect the R_max to be 5. Which it is.

Another example is provided by <CEG vs Legends (2012)>. Here, there are again 10-players, or 5/team. Each match-up is 25 games, but it’s a Scheven-2, so each player-pair played twice, for 50-games. The number of rounds should be 5*2 = 10, which matches R_max.

Let’s look at a mismatch. These are likely Swiss or RR tournaments which were’s identified because of missing games. (Actually, the classification should probably pick out Scheven’s before Swiss – I warned you this is a work in progress). OK, then, <Hasting (1927/28)> which we know isn’t a Scheven. But suppose we didn’t, how do the numbers work out?

There are 10-players yet again, 5/team. That makes 25-games for a Scheven-1. Since there are 25-games (and not 50) we have a Scheven one. That should only have a R_max of 5. But the table entry shows R_max = 9. That doesn’t fit a Scheven-1. Of course, it does fit a 10-RR-1.

I “blued-out” all the entries above where R_max disallows the Scheven.

One would hope to add a test utilizing R_max into the classification algorithm. But care must be taken. Look at <Indochess Man — Machine (2005)>. There, R_max = 0. That is because none of those games have round information. Yet, it’s clearly a Scheven tournament.

R_max, if present and correct, most assuredly can be used. But if unreliable, it can really only be used to exclude, and only for those cases where a game is found with a disallowed round number.




Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s