Hi all,
the interview with author of strelka is now online.
http://sdchess.ru/osipov_int1.htm
try to translate with google or another tool
Interview with Yuri Osipova
Moderator: Ras
-
Werner
- Posts: 2981
- Joined: Wed Mar 08, 2006 10:09 pm
- Location: Germany
- Full name: Werner Schüle
Interview with Yuri Osipova
Werner
-
Andre van Ark
- Posts: 91
- Joined: Fri Mar 10, 2006 9:32 am
- Location: Amersfoort
Re: Interview with Yuri Osipova
1. Yuri, probably you did not expect this method of your cursor. The repeated as incantations of word "clone, clone, clone" do not contain any serious facts and arguments and they are actually unsubstantiated. Moreover our compatriots play, perhaps, the first violins in this polecat.
It does not seem you that one of the reasons, on which this occurred, it became too purchasing the characteristic of your cursor, about which you they spoke that it usual alphabeta, which it realized in the form of the chess cursor Strelka, there is nothing original there, everything was already thousands of times realized and described by others - even and to add something, algorithm completely standard.
4 it did not actually expect this method, and from this I feel itself today is very comfortably. As far as the brief description of algorithm is concerned, algorithm actually is completely usual. I is ready to meet with any person, who examines chess algorithms, to bring the source texts - we them together will look, will compile and will verify on the tests. I is confident, that any specialist, after looking on iskhodniki, he will say: well and that in them such?
If we speak in greater detail about the algorithm of pointer, then in it is made essential support to the tactics. Algorithm was fixed on the collection of the test positions, in which there is the nontrivial tactical solution. In order to attain the presence of such solutions, it was necessary to introduce the sufficiently daring heuristics of an increase in the depth of sorting or cutting of unpromising versions into the program. In addition to this, were used the simple, but sufficiently effective methods of the optimization of the parameters of estimated function. In order to describe about this in more detail, much place and time is necessary; therefore give let us leave this on then.
2. Not could you in connection with this in more detail to describe to the amateurs of chess about the history of the creation of pointer, or your enthusiasm by chess programming? How I do understand, pointer - this the first public realization of your ideas in the programming and, probably the objective result of great work, which was made by you in this region and did not obtain, in your view, valuable realization in this version of pointer? Main domestic billets did remain outboard?
The history of the creation of pointer can be begun from a distance. Approximately in 1986 I for the first time interested itself in play algorithms. Not there were any special ideas whatever, simply it was desirable to try to make the playing program. I selected for the beginning the simple small cross- zeros (five in a number), and on this game it for the first time tested the basic minimax principles and alphabeta. In four years I decided to be swung back to chess, and for the beginning wrote something entirely simple. This program played very badly - it rather imitated game. Gradually the game of program was improved, but then 4 it threw.
But then, as long ago as 1996 happened the sequential splash of interest in computer chess. Modest Fritts, written for PC, on the sequential championship of peace unexpectedly for all became champion, after leaving behind this monster as Deep Blue. I understood that came the time, when practically any qualified programmer on the sufficiently modest technology can achieve substantial advances.
But me it no longer was desirable to return to the sufficiently dull alphabeta to algorithm. Arose the idea to realize something completely original, and the corresponding algorithm gradually was formed. The idea of algorithm is sufficiently simple - we bring the chess task to the task of the search for the extremum of function, assigned on the point set, and instead of the sorting we attempt to search for this extremum in the promising trend (something of the type of the method of descent). Misfortune proved to be (both for all such algorithms), in a constant dumping into the local extrema, from which 4 and I could not be selected.
I undertook the numerous attempts to correct something, but so nothing and it left.
And then I decided to return to the old good alphabeta to algorithm and to bring it to the mind. By this time was accumulated a very large quantity of diverse information about this in the Internet, including the open source texts. After studying all this, I revealed, to what extent it were improved in recent years alphabeta. And I villages to write, using in the essence that which was known for all. As a result the pointer, which completely unexpectedly proved to be sufficiently strong, although very imperfect, was borne for me.
3. Do not consider you that, takeing into account the practical force of pointer perhaps it does be worthwhile not to lay out the initial files of cursor, but to grant to their independent authoritative experts for the study and the analysis?
The second version (i.e. examination) completely arranges me. Moreover, 4 it is simple to grief by desire so that such experts would turn themselves to me, and 4 to them I will return everything that i have. I hope that this examination will make it possible to remove or at least to weaken charges in the cloning.
As far as the free facing of iskhodnikov is concerned, 4 and were assembled this to make. And most likely, in the course of time so I will make. However as long as I do not want with this to hurry. It is necessary to very thoroughly weigh the consequences of this step.
4. Your further plans? Do plan you to develop pointer, to supplement with its new functions and to make support UCI of protocol? It can be, in your plans, make a pointer commercial project?
Thus far he did not think about his further plans. For the beginning it is necessary to realize that new situation, in which 4 unexpectedly for itself it proved to be. I can only say that my interest in alphabeta to algorithm is until today satisfied, and 4 no longer I want to it to return. Me completely it would arrange, if the enthusiast (or several enthusiasts), would be located, by whom this, until now, is interesting, and to which I could transmit the work made by me.
To the commerce I relate with the large doubt - 4 I do not think that on chess it is possible to earn something essential.
5. One additional question is connected with your public appearance under the pseudonym. These are the unwillingness to become public in the world of chess programming and your internal tendency toward "cryptography"?
I would want to leave this question without the commentaries. Pseudonym has real prototype, and 4 it wanted nothing about this to speak.
6. Much changed in recent years in the world of chess programming. If earlier for a period of entire decade was observed the absolute domination of products ChessBase (Fritz, Junior, Hiarcs), then today we have somewhat different picture. Prevail cursors created by the talented authors, who do not have relation to ChessBase. 4 not I will call the names of programs, they even without that are well known to the amateurs of chess. Do consider you that future after the talented or even brilliant lone persons or today in the period of universal globalization even "genius" will have either to create I do be worthwhile command (organizational, financial and professional) or to go to the bow to the important producers of chess products?
It is very difficult to attempt to predict future. I think, future after those, to whom by sheer luck will come to mind interesting ideas. And it is completely unimportant, it does work to Chessbase or simply it sits house.
But this is what to make then - to create the command or to go to the bow to the important producers - depends on person himself. If we there is the desire earn on its ideas - throw to program and begin to be occupied by organization, but if money little interest you - continue to work.
7. In autumn this year our site plans conducting the sequential fourth championship OF THE CIS. In this of times we would want to change the regulations of this tournament, also, for purposes of larger transparency to propose to the authors to send to us the initial files of our cursors, which it will analyze independent authoritative ekspert(-y). Are prepared you to participate in our championship for such conditions?
4 by both hands for the total transmittance. And I will first be, who will send the source texts of his program to the championship. I fear only, that I will be not only first, but also only, who this will make.
8. Do not have you plans to participate in the sequential championship of peace among the chess programs or in any other authoritative international competition?
4 he already indicated that yet he did not think about further plans. However, I consider that in the international competitions to participate necessary. And the matter not only in the ambitions or the prestige. Only so it is possible to be developed and to advance.
The many thanks, for the interesting interview. We desire to you the new successes in the chess programming! Entire to you best, to the new encounters.
Babelfish regards,
André
It does not seem you that one of the reasons, on which this occurred, it became too purchasing the characteristic of your cursor, about which you they spoke that it usual alphabeta, which it realized in the form of the chess cursor Strelka, there is nothing original there, everything was already thousands of times realized and described by others - even and to add something, algorithm completely standard.
4 it did not actually expect this method, and from this I feel itself today is very comfortably. As far as the brief description of algorithm is concerned, algorithm actually is completely usual. I is ready to meet with any person, who examines chess algorithms, to bring the source texts - we them together will look, will compile and will verify on the tests. I is confident, that any specialist, after looking on iskhodniki, he will say: well and that in them such?
If we speak in greater detail about the algorithm of pointer, then in it is made essential support to the tactics. Algorithm was fixed on the collection of the test positions, in which there is the nontrivial tactical solution. In order to attain the presence of such solutions, it was necessary to introduce the sufficiently daring heuristics of an increase in the depth of sorting or cutting of unpromising versions into the program. In addition to this, were used the simple, but sufficiently effective methods of the optimization of the parameters of estimated function. In order to describe about this in more detail, much place and time is necessary; therefore give let us leave this on then.
2. Not could you in connection with this in more detail to describe to the amateurs of chess about the history of the creation of pointer, or your enthusiasm by chess programming? How I do understand, pointer - this the first public realization of your ideas in the programming and, probably the objective result of great work, which was made by you in this region and did not obtain, in your view, valuable realization in this version of pointer? Main domestic billets did remain outboard?
The history of the creation of pointer can be begun from a distance. Approximately in 1986 I for the first time interested itself in play algorithms. Not there were any special ideas whatever, simply it was desirable to try to make the playing program. I selected for the beginning the simple small cross- zeros (five in a number), and on this game it for the first time tested the basic minimax principles and alphabeta. In four years I decided to be swung back to chess, and for the beginning wrote something entirely simple. This program played very badly - it rather imitated game. Gradually the game of program was improved, but then 4 it threw.
But then, as long ago as 1996 happened the sequential splash of interest in computer chess. Modest Fritts, written for PC, on the sequential championship of peace unexpectedly for all became champion, after leaving behind this monster as Deep Blue. I understood that came the time, when practically any qualified programmer on the sufficiently modest technology can achieve substantial advances.
But me it no longer was desirable to return to the sufficiently dull alphabeta to algorithm. Arose the idea to realize something completely original, and the corresponding algorithm gradually was formed. The idea of algorithm is sufficiently simple - we bring the chess task to the task of the search for the extremum of function, assigned on the point set, and instead of the sorting we attempt to search for this extremum in the promising trend (something of the type of the method of descent). Misfortune proved to be (both for all such algorithms), in a constant dumping into the local extrema, from which 4 and I could not be selected.
I undertook the numerous attempts to correct something, but so nothing and it left.
And then I decided to return to the old good alphabeta to algorithm and to bring it to the mind. By this time was accumulated a very large quantity of diverse information about this in the Internet, including the open source texts. After studying all this, I revealed, to what extent it were improved in recent years alphabeta. And I villages to write, using in the essence that which was known for all. As a result the pointer, which completely unexpectedly proved to be sufficiently strong, although very imperfect, was borne for me.
3. Do not consider you that, takeing into account the practical force of pointer perhaps it does be worthwhile not to lay out the initial files of cursor, but to grant to their independent authoritative experts for the study and the analysis?
The second version (i.e. examination) completely arranges me. Moreover, 4 it is simple to grief by desire so that such experts would turn themselves to me, and 4 to them I will return everything that i have. I hope that this examination will make it possible to remove or at least to weaken charges in the cloning.
As far as the free facing of iskhodnikov is concerned, 4 and were assembled this to make. And most likely, in the course of time so I will make. However as long as I do not want with this to hurry. It is necessary to very thoroughly weigh the consequences of this step.
4. Your further plans? Do plan you to develop pointer, to supplement with its new functions and to make support UCI of protocol? It can be, in your plans, make a pointer commercial project?
Thus far he did not think about his further plans. For the beginning it is necessary to realize that new situation, in which 4 unexpectedly for itself it proved to be. I can only say that my interest in alphabeta to algorithm is until today satisfied, and 4 no longer I want to it to return. Me completely it would arrange, if the enthusiast (or several enthusiasts), would be located, by whom this, until now, is interesting, and to which I could transmit the work made by me.
To the commerce I relate with the large doubt - 4 I do not think that on chess it is possible to earn something essential.
5. One additional question is connected with your public appearance under the pseudonym. These are the unwillingness to become public in the world of chess programming and your internal tendency toward "cryptography"?
I would want to leave this question without the commentaries. Pseudonym has real prototype, and 4 it wanted nothing about this to speak.
6. Much changed in recent years in the world of chess programming. If earlier for a period of entire decade was observed the absolute domination of products ChessBase (Fritz, Junior, Hiarcs), then today we have somewhat different picture. Prevail cursors created by the talented authors, who do not have relation to ChessBase. 4 not I will call the names of programs, they even without that are well known to the amateurs of chess. Do consider you that future after the talented or even brilliant lone persons or today in the period of universal globalization even "genius" will have either to create I do be worthwhile command (organizational, financial and professional) or to go to the bow to the important producers of chess products?
It is very difficult to attempt to predict future. I think, future after those, to whom by sheer luck will come to mind interesting ideas. And it is completely unimportant, it does work to Chessbase or simply it sits house.
But this is what to make then - to create the command or to go to the bow to the important producers - depends on person himself. If we there is the desire earn on its ideas - throw to program and begin to be occupied by organization, but if money little interest you - continue to work.
7. In autumn this year our site plans conducting the sequential fourth championship OF THE CIS. In this of times we would want to change the regulations of this tournament, also, for purposes of larger transparency to propose to the authors to send to us the initial files of our cursors, which it will analyze independent authoritative ekspert(-y). Are prepared you to participate in our championship for such conditions?
4 by both hands for the total transmittance. And I will first be, who will send the source texts of his program to the championship. I fear only, that I will be not only first, but also only, who this will make.
8. Do not have you plans to participate in the sequential championship of peace among the chess programs or in any other authoritative international competition?
4 he already indicated that yet he did not think about further plans. However, I consider that in the international competitions to participate necessary. And the matter not only in the ambitions or the prestige. Only so it is possible to be developed and to advance.
The many thanks, for the interesting interview. We desire to you the new successes in the chess programming! Entire to you best, to the new encounters.
Babelfish regards,
André
-
Dave McClain
- Posts: 1018
- Joined: Fri Mar 10, 2006 12:56 am
- Location: Major, 45 Commando, Royal Marines, Condor Barracks, Arbroath, Scotland
- Full name: Dave MCClain
Re: Interview with Yuri Osipova
Well, the open invitation is there. Who is going to accept it for the world to see?André van Ark wrote: As far as the brief description of algorithm is concerned, algorithm actually is completely usual. I am ready to meet with any person, who examines chess algorithms, to bring the source texts - we then together will look, will compile and will verify on the tests. I am confident, that any specialist, after looking on iskhodniki, he will say: well and that in them such?
Major, 45 Commando, Royal Marines,
Condor Barracks, Arbroath, Scotland
Condor Barracks, Arbroath, Scotland