Riccardo Poiani

Hi @Jinyu-W thanks for your response and availability. I am trying to replicate results for Offline and Online LP in different configurations. Currently, I am able to achieve 0% and...

@Jinyu-W thanks for you feedback. I will try for sure to integrate the safety reward factor. For what concerns the Online LP, are the methods in [paper 1](https://arxiv.org/pdf/1903.00714.pdf) and [paper...

Also, @Ahmedest61 I have seen something similar to what you are looking for in `temp_gnn` branch, under the folder `examples/cim/ac_gnn`, but I have not tried it. (I do not know...

No idea on this @Ahmedest61. I am currently working on `master`. Please, let me know if the code I pointed you to is what you were looking for and if...

> Hi, as far as I know, the LP methods used in paper 1 and paper 2 are not the same. The one used in paper 2 is more closer...

> Hi @riccardopoiani, I could share more information about the LP methods we used before. > > * The Online LP formulated based on the demand(orders) predicted with a simple...

> > Hi, as far as I know, the LP methods used in paper 1 and paper 2 are not the same. The one used in paper 2 is more...

Hi @wesley-stone, could any of you provide the environment configuration that you used in paper 2? With `master/global_trade_cfg` I have results of 68% fulfillment, while with [these configs](https://github.com/microsoft/maro/pull/169) I am...

Hi, @wesley-stone @Jinyu-W any update on the previous post? It seems to me somehow relevant to provide good configurations within the environment. Also because in [these configs](https://github.com/microsoft/maro/pull/169) no repositioning performs...

Thank you very much @Jinyu-W. I just had a few tests on the environment and this somehow confirmed me what I was thinking. The summary is the following: I believe...