学生向王老师反馈代码运行结果与论文不一致

ChatGPT4o 使用 ChatGPT 4o 回答

问题:

translate into polite english but be brief: ​王老师: 您好! 今天下午paper pre的时候,我提到我们联系到《Reinforcement Learning for Optimal Execution when Liquidity is Time-Varying》这篇文章的作者并获得了他们的部分源代码,但是我们运行代码后的结果和论文中并不一致。我们将原作者的代码、我们使用的修改版本和运行产生的结果在附件中一并发给您,内有README.txt作以简单说明。此外,我们观察到复现时test训练好的策略使用的定义market的参数应该和原文中一致,除了S0初始价格,代码中为10,而原文中并无显式说明。 这门课程内容非常详实,在这短短三周的学习中我们收获很大,非常感谢您的辛勤付出!也希望以后有机会还能再听到您的课程!

   敬颂

教安

回答:

Dear Professor Wang,

Good day!

During today's paper presentation, I mentioned that we contacted the authors of the article "Reinforcement Learning for Optimal Execution when Liquidity is Time-Varying" and obtained part of their source code. However, the results we obtained from running the code do not match those in the paper. We have attached the original author's code, our modified version, and the results we obtained, along with a README.txt file for a brief explanation. Additionally, we observed that the parameters defining the market used by the test-trained strategy should match those in the original paper, except for the initial price S0, which is set to 10 in the code but not explicitly stated in the paper.

The content of this course has been very comprehensive, and we have learned a lot in these three short weeks. We greatly appreciate your hard work and dedication! We hope to have the opportunity to attend your classes again in the future.

Best regards, [