Log in

Genetic quantum algorythm

Abstract- This paper proposes a novel evolutionary computing method called a genetic quantum algorithm (GQA). GQA is based on the concept and principles of quantum computing such as qubits and superposition of states. Instead of binary, numeric, or symbolic represen- tation, by adopting qubit chromosome as a representation GQA can represent a linear superposition of solutions due to its probabilistic representation. As genetic operators, quantum gates are employed for the search of the best solution. Rapid convergence and good global search ca- pability characterize the performance of GQA. The effec- tiveness and the applicability of GQA are demonstrated by experimental results on the knapsack problem, which is a well-known combinatorial optimization problem. The results show that GQA is superior to other genetic algo- rithms using penalty functions, repair methods, and de- coders.


Comments

  • JohnLast 2820 days ago

    I think this paper is very interesting. A genetic quantum algorythm would have an immediate application regarding trading strategies allowing to optimize better trading strategies with a lot of variables.

    I think there was an independant code for genetic algorythm for mt4.

  • JohnLast 2819 days ago

    This is also interesting. You can see how combining multiple GPU you can outperform IntelCore i7 200 times using genetic quantum algorythm to combinatorial optimization problem. It looks like at the retail level it is possible to do something ;).

     

  • JohnLast 2813 days ago

    Here I will add a link showing that the OpenCL is not really too far from metatrader.

    Check here

    Starting from build 581, the native support of the OpenCL interface has been added for using GPU capacities. (OpenCL is an alternative of CUDA that works only on NVidia graphic cards)

    Here is a small working example of the Mandelbrot fractal calculation in OpenCL. The OpenCL accelerates calculations of fractal approximately up to 100 times in comparison with the CPU calculations.

  • JohnLast 2813 days ago

    However not all GPUs are supported by the OpenCL.

  • JohnLast 2813 days ago

    You can check also this great article.

    http://www.mql5.com/en/articles/405

     

    Here it may be a little bit offtopic so I would make a completely different thing. But take a look. 

    On the link there is a basic introduction about the possibility of the opencl. And it is integrated in the metatrader plateform.

    So I wanted to make tests on my own comparing what I get using the CPU and what I get usign the GPU with the opencl.

    Quite amazing. 

    And I am usign a quite old Ati Mobility Radeon HD4570. On the site of Khronos in the list of supported and conformant products my video card was not listed. But I had a nice surprise to see that AMD had upgraded their drivers telling that they are extracting the maximum of those cards adding opencl support. Nice surprise.

    But let see the results:

    As Jaguar suggested that is a great way to optimize your strategies. And even more you could use more computers and parallel computing to increase the computational power. And if you can hire ressources and if your algorythm is using opencl I am wondering if you could use the GPU of the Agents. If yes the computatinal power at the hands of the retail user would be really big.

     

    image
    От EA tests

     

    image
    От EA tests

     

  • JohnLast 2813 days ago

    The difference between 76.9 seconds and 5.66 seconds is quite noticeable.

  • jaguar1637 2813 days ago

    This 's great

    in one hand, we have the genetic algorithm , the kernel method and linear recurrence system (for calculation) , the Mongo DB to store datas , and now the physical infrastructure design

  • JohnLast 2813 days ago

    The only problem for my chaotic kernel is that the Opencl does not support recurrence calculations by now (or at least that is what I understood that is related with the GPU hardware  limitations). What is necessary is to do an unrolling of the recursive computations for the logistic map.

    Anyway that is not a big problem still the gaussian is very good.

    So the idea is here now, it is about how to implement it correctly. Opencl is quite a big deal.

     

  • jaguar1637 2813 days ago

    OK, for a VPS, what should I ask as requirements ?

    Do you think it's possible to add an ATI video card on a VPS ?

  • JohnLast 2812 days ago

    I think that the Opencl is for data analysis and the VPS is for trade execution.

     

  • JohnLast 2812 days ago

    It lools like the technology is running faster than the ability of the retail traders to incorporate that technology into their practices.

    For example for mt5, not only it uses C++ and quite a few people are really confident with that but the Opencl as I understand it requires to rewrite the code and add opencl code that matches with the individual architecture of the GPU.