Algorithm Optimisation Championship. - page 101
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
You two are so funny... ) No offence. ))))
1. If the global maximum is known to all without exception, then think again, carefully, what happens?
Can a championship of optimization algorithms be held at all in such a case? Certainly, it cannot, because any participant can idly call the FF at 60-70% of the maximum allowed and give a result with 100% accuracy!
2. We can only compete on REAL FFs! The maximum value of one of the competitors is the best result.
1. I was saying that the global maximum must be known by the referee.
This is necessary so that after running the algorithms, at the stage of checking the results, we could compare not only the number of FF calls, but also the accuracy of calculating the max. If you don't know the max-function, you can't check the accuracy of algorithm's calculation. What is the point of an optimization algorithm whose accuracy (and hence effectiveness) has not been tested and proven?
2. In the same way, you can run the algorithm at "idle" and at the end give out the highest number possible as a result... And win?
If not, the contestant would have to save the values leading to the maximum to prove that the algorithm actually worked.
Does this mean that the algorithm is not ready yet? How much time do you expect to get to understand and prepare an algorithm for a known FF? What difference does it make what is contained in the FF? The algorithm has to be able to work with an FF it doesn't know, otherwise it's bullshit and not an algorithm.
How can my algorithm be ready if it hasn't yet been given the problem it's supposed to solve? Not yet all the components of the big picture to present.
However, don't worry, I solved the last task in 6 hours, I will solve this one in a bit more time...
Creating universal algorithms is my vocation.))
The essence of the task is to create an algorithm for effectively finding the maximum value of an analytic function. No more universality of the algorithm than in the formulation of the essence of the problem is required.
And what universality are you talking about if the algorithm always uses 4 basic parameters - range, step, number of parameters and value obtained from the FF?
How can my algorithm be ready, if in fact it hasn't yet provided the problem it's supposed to solve? Not yet all the components of the big picture to present.
However, don't worry, I solved the last problem in 6 hours, I will solve this one in a little more time...
Creating universal algorithms is my vocation)))
It all comes down to facing the Mad Hatter's Riddle -- and the whole thing called the Championship -- becomes like the Mad Tea Party --https://ru.wikipedia.org/wiki/Болванщик
The Hatter eventually admitted he didn't know the answer to his riddle.
1. What I was saying is that the global maximum should be known by the referee.
This is necessary so that after the algorithms run, at the stage of checking the results, you can compare not only the number of FF calls, but also the accuracy of the calculation of the max. If you don't know the max-function, you can't check the accuracy of algorithm's calculation. What is the point of an optimization algorithm whose accuracy (and hence effectiveness) has not been tested and proven?
2. in the same way, you could run the algorithm at "idle" and at the end produce the highest possible number as a result... And win?
3. if not, the contestant will need to save the values leading to the maximum to prove that the algorithm actually worked.
1. If the max will only be known by the referee (and it is the referee who has to compile the FF for this) then for what purpose do you insist on looking at the insides of the FF? No, as I said, no one will know the max, including the referee. The best result would be considered the maximum. If there were many more contestants than the 2nd, we could take the maximum result of one of the contestants as 100% accurate, and the others would be ranked from that value accordingly. But since there are only 2 of us, we can't measure accuracy, we can only choose the best result within the championship. But I repeat once again, we can use a known function with a known maximum outside the championship and see the real accuracy of the algorithms.
2. In case of unknown FF such "exactly the same" cannot be done, because at the end of all runs it is necessary to save the parameters into a file, which will certainly be checked by feeding them into FF and checking the FF result and the result of the contestant.
3. It does not matter with an unknown FF. What matters is the final result and the parameters for that result that will be checked. There is no way for anyone to cheat.
If you are looking for loopholes, you are in vain. If you are worried about mistakes in the FF, you are in vain. If you are worried about not being able to assess the real ability of algorithms to search - then in vain, you will have the opportunity.
It all comes down to you facing the Mad Hatter's Riddle -- and the whole thing called the Championship -- becomes like the Mad Tea Party --https://ru.wikipedia.org/wiki/Болванщик.
The funny thing is, the Hatter himself eventually admitted he didn't know the answer to his riddle.
Well Time stopped himself for the Hatter, and thereby did him a great honour). In our case, Time is unlikely to stop himself, so our "tea party" will inevitably come to an end...
And at this "tea party" the Enigma will not allow itself to confuse the Hatter enough to remain unsolved.
How can my algorithm be ready, if in fact it has not yet provided the task it must solve? Not yet all the components of the big picture to present.
However, don't worry, I solved the last task in 6 hours, I will solve this one in a little longer...
Creating universal algorithms is my vocation.))
The essence of the task is to create an algorithm for effectively finding the maximum value of an analytic function. No more universality of algorithm than in the formulation of the essence of the problem is required.
And what universality are you talking about if the algorithm always uses 4 basic parameters - range, step, number of parameters and value obtained from the FF?
That's what I thought. Everything I said earlier is like a wall of peas, and you don't even have an algorithm ready at all....
I'm sorry you still don't have the algorithm ready by the time the championship starts, which means you are far from the notion of a universal search algorithm.
I'm posting the FF today and following the steps I've voiced, with a break of 1-2 days. If by the time the algorithms are tested my algorithm turns out to be the only one, then I am a winner.
I have already advised you a million times to take a readily available algorithm, figure it out, modify it if necessary and use it in the Championship. Do you really think that you can create a universal algorithm in a few hours or a few days? At least take Alglib from the code base, set it up, although it won't be your algorithm, but at least I'll have the opportunity to compete with someone other than myself.
So - further on the steps of the championship without delay. It's all chewed up and chewed out in an impossibly detailed way. I'm sorry, I can't wait any longer - otherwise this dialogue between me and you can go on forever.
Well, Time has stopped itself for the Hatter, and thereby done him a great honour). In our case, Time is unlikely to stop itself, so our "tea party" will inevitably come to an end...
And at this "tea party" the Enigma will not allow itself to confuse the Hatter enough to remain unresolved.
1. If the max will only be known by the referee (and it is the referee who has to make up the FF for this to happen), what is the purpose of your insistence on looking at the insides of the FF? No, as I said, no one will know the max, including the referee. The best result would be considered the maximum. If there were many more contestants than the 2nd, we could take the maximum result of one of the contestants as 100% accurate, and the others would be ranked from that value accordingly. But since there are only 2 of us, we can't measure accuracy, we can only choose the best result within the championship. But I repeat once again, we can use a known function with a known maximum outside the championship and see the real accuracy of the algorithms.
2. In case of unknown FF such "exactly the same" cannot be done, because at the end of all runs it is necessary to save the parameters into a file, which will certainly be checked by feeding them into FF and verifying the FF result and the result of the participant.
3. It does not matter with an unknown FF. What matters is the final result and the parameters for that result that will be checked. There is no way for anyone to cheat.
If you are looking for loopholes, you are in vain. If you are worried about mistakes in the FF, you are in vain. If you are worried about not being able to evaluate the real ability of algorithms to search - then in vain, you will have the opportunity.
Conclusion - whatever I do - all in vain... ))
"for what purpose do you insist on looking at the insides of the FF?" - I've just never seen it.
"If you're looking for loopholes, it's for nothing." - How can I, having never once solved a similar problem and not having any algorithm, look for any loopholes? I'm trying to reason. Just like before.
I don't think you're discussing anything, you're just making rules.
That's what I thought. Everything I said before is just peas in a pod, and you don't even have an algorithm ready at all....
I'm sorry you still don't have the algorithm ready by the time the championship starts, which means you are far from the notion of a universal search algorithm.
I'm posting the FF today and following the steps I've voiced, with a break of 1-2 days. If by the time the algorithms are tested my algorithm turns out to be the only one, then I am a winner.
I have already advised you a million times to take a readily available algorithm, figure it out, modify it if necessary and use it in the Championship. Do you really think that you can create a universal algorithm in a few hours or a few days? At least take Alglib from the code base, set it up, although it won't be your algorithm, but at least I'll have the opportunity to compete with someone other than myself.
So - further on the steps of the championship without delay. It's all chewed up and chewed out in great detail. I'm sorry, I can't wait any longer - otherwise this dialogue with you and I can go on forever.
I see. You've decided to tighten the screws. No time to prepare, no compromise, no joint decision-making.
If I never write my algorithm and don't fight you, will you still consider yourself the winner?
Doesn't such a victory look pathetic?
Conclusion - whatever I do, it's all for nothing... ))
"what's the purpose of your insistence on looking at the insides of the FF?" - I've just never seen it.
"If you're looking for loopholes, it's for nothing." - How can I, having never once solved a similar problem and not having any algorithm, look for any loopholes? I'm trying to reason. Just like before.
I don't think you're discussing anything, you're just making rules.
By now you should have at least some algorithm, but you don't. I told you not to stick to functions, no analytical constructs and solutions will help you, because you have no knowledge of the contents of the FF. I cited a thousand times the example of a staff optimizer, he knows nothing about users' tasks, nothing, and yet successfully solves the optimization tasks.
Create a universal algorithm, find it on the Internet, use random search, whatever, but you should already have a ready-made algorithm. How are you going to compete in something you don't already have?
SZZ Anyone who has any algorithm at all is still welcome to join the competition, since apparently I'm really the only one left. But in any case it doesn't change anything, my algorithm and test bench will be shown within a couple of days, no matter what.