Finally ?

[Tony Thomas]

I wasnt the only one who said that there is no longer any King safety margin code..

Your welcome Eduard! But sorry to disappoint you on the KingSafetyMargin Issue. Maybe Thomas is the right person to answer why he chose to terminate using this option in the new versions. The versions are still on beta stage so the code hasn'r been cleaned yet, even 1.4.1SE as well.

I don't see in the Toga 1.4.1 source anymore the king safety eval.cpp (shelter_square function) code even thou the variable is still taken from the interface.

So you must be seeing things of randomness. There should not be difference in what you put in the king safety.

Jarkko

Hello Eduard,

I don't understand how that could be, unless Fritz remembers different settings than what it says is using. The only two lines where KingSafetyMargin appears are:

static int KingSafetyMargin = 1600;

KingSafetyMargin = option_get_int("Toga King Safety Margin");

The sourcecode creates the variable and still reads a new value that you put in it, but it is not used anywhere else in the program. I think that if you were to play without book, the start of each game at least should be the same against a fixed opponent, but on Playchess that is difficult to arrange. I don't feel much for doing the experiment myself, and I don't have Fritz. I'm blaming Fritz at the moment if this really is true. Good luck with your games!

Regards, Eelco

I think that if you have "Toga King Safety" set to false (or unchecked) "Toga King Safety Margin" has no effect.

[Edward German]

If the Toga king safety option is not checked (not activated), and even more if there is no Toga king safety code anymore at all as a poster here mentioned (if I get him correctly), then the Toga king safety margin has of course no effect. Anything else is pure phantasy. It is not rational.

It makes no sense to try alternative settings, as long as we do not have a reliable rating for the default settings! With what do you compare if it's better or worse?!

Hello Mike!

I will anwer in german, Mike can then translate it if someone have an Question.

Ich habe alles verstanden, auch mit meinem schlechten English.

Zu "King Safety Margin" ein Wort:

Ich weiß sehr wohl dass diese Funktion keine Auswirkung hat wenn "Toga King Safety" nicht aktiviert ist. Leider habe ich dennoch trotz dem in vielen ENG-ENG Matches (in meinen Tests) beobachten können dass Toga anders spielt wenn man den Wert für Toga Safety Margin verringert. Ich kann sogar sagen in welche Richtung sich das Auswirkt. Verkleinert man den Wert, dann spielt Toga aktiver. Toga ist dann eher bereit auf den gegnerischen König zuzugehen. Und: So paradox es klingt, die eigene Königssicherheit steigt sogar.

Normal darf das ja nicht sein, aber ich habe wirklich schon viel getestet und dieses Phänomen entdeckt - ich kann es nicht erklären aber es ist so da wie ich schreibe.

Zu den anderen 2 Parametern:

a) Toga Lazy Eval Margin. Hier ist der Wert 200 sehr gut, aber er ist nur gut bei den Vorgängerversionen von Toga, nicht aber bei Toga 1.4.1SE! Bei der version 1.4.1SE ist der Wert 225 besser weil Toga 1.4.1SE taktisch viel schwächer ist als die vorherigen originalen Toga's von Thomas Gaksch! Mit dem Wert 225 aber kompensiert Toga 1.4.1 diese Schwäche zu 80%, wenngleich die Geschwindigkeit der Engine (Suchtiefe) etwas daran leidet. Der Vorteil jedoch überwiegt deutlich!

b) Verification Reduction. Toga 1.4.1SE ist extrem selektiv und übersieht oft taktische Einschläge dadurch. Auf Schach.de wurden die Default Settings buchstäblich zerschossen! Toga 1.4.1SE ist taktisch schwach in der Grundeinstellung! Verändert man nun Verification Reduction auf den Wert 4 (Default ist 5) dann übersieht Toga 1.4.1SE weniger! natürlich geht auch das ein wenig auf die Kosten der Geschwindigkeit, aber die ist speziell bei Toga 1.4.1SE extrem - Kurzum gesagt: Toga 1.4.1SE prunt zu stark und hinterlässt damit viele Löcher. Wenn man jedoch Verification Reduction auf 4 setzt, dann ist das nicht mehr so der Fall. Toga 1.4.1SE spielt dann sehr schön - und bei der Analyse ist sie auch findiger!

Bitte beachten: Alle diese Angaben gelten NUR für Toga 1.4.1SE und nicht für andere Togaversionen.

Das ist erstens meine Meinung, und zweitens habe ich das mit einigen Dutzend Spielen mit PB auf Schach.de erprobt. Ich weiß also was ich sage, auch wenn manches utopisch klingt!

Regards!

[Henrik Dinesen]

If the Toga king safety option is not checked (not activated), and even more if there is no Toga king safety code anymore at all as a poster here mentioned (if I get him correctly), then the Toga king safety margin has of course no effect. Anything else is pure phantasy. It is not rational.

If the stearwheel in the car isn't connected with the machanish that controls the wheels, you can't stear the car - but can still drive, and surely you can turn the stearwheel as much as you like... And if you turn your head enough during this process, you may get feeling you're actually changing directions

[Tony Thomas]

If the Toga king safety option is not checked (not activated), and even more if there is no Toga king safety code anymore at all as a poster here mentioned (if I get him correctly), then the Toga king safety margin has of course no effect. Anything else is pure phantasy. It is not rational.

If the stearwheel in the car isn't connected with the machanish that controls the wheels, you can't stear the car - but can still drive, and surely you can turn the stearwheel as much as you like... And if you turn your head enough during this process, you may get feeling you're actually changing directions

Nice one, haha. What the heck is Machanish?

[Ovyron]

It makes no sense to try alternative settings, as long as we do not have a reliable rating for the default settings! With what do you compare if it's better or worse?!

I don't need many games to see the engine's playing style, and I don't care much if the engine is 30 ELO weaker or stronger. Because the engine is already 1400 ELO rating points stronger than me so I can't tell the difference in strength, but the style just jumps to my eye.

If someone is like me and has a preference for style, and likes active/aggressive/offensive style better than subtle/solid/squeezing style then I suggest Toga II 3.1.2SE with Chekov settings.

That is:

http://computerchess.org.uk/Toga/TogaIIv3.1.2SE.zip

With:
[OPTIONS]
Verification Reduction=6
Futility Margin=125
Extended Futility Margin=325
Delta Margin=55
Quiescence Check Plies=2
Material=102
Piece Activity=103
King Safety=125
Pawn Structure=105
Passed Pawns=105
Toga Lazy Eval Margin=245
Toga Extended History Pruning=true

Thanks to Fabien Letouzey, Thomas Gaksch, Chris Formula, Denis Mendoza, and Eelco de Groot (And anybody else responsible) to make this possible.

For those that only care about ELO points, then you'll have to wait a while

I tested the 1.4.1SE version with Chekov settings and it didn't work as expected - But now I'll test Eduard Nemeth's settings, to see if the style is as lively

[Henrik Dinesen]

Nice one, haha. What the heck is Machanish?

ehh, typo of a kind... mechanics

[Sylwy]

In other words we are leaving you in suspense Ruxy
Eelco

I see !
I other words I must delay my " Monsters tournament " (100 games each vs. each, on quad,first 10 engines ,128 hash, long time control ).
No problem, I'll wait for !

Regards,
Sylwy

[Golem]

From my tests*, the best Toga so far (for blitz 2' 2") seems to be Toga1.4beta5 (Toga1.4.1SE seems to be about the same strength).

@Edward : I also confirm that the king safety code is not working in Toga1.4beta5. Please remove this line from your settings. It's just nonsense to tell users to modify this value.

*tested at blitz 2' 2" with rybka 1.0, naum 2.0, spike 1.2 and fruit 2.3.2. PS : I have to run some more games to confirm (I'll stop the tests when I'll have 300 games).