Working repository in context of the bachelor project "Online Marketplace Simulation: A Testbed for Self-Learning Agents" at the research group Enterprise Platform and Integration Concepts
We don't know if we have an off-by-one-error in our policy and market functions. Check that and define our solution.
Edit: (investigated by @SinNeax)
Since the RLAgents determine/randomly choose a price between 0 and config.maxprice - 1, we needed to increment the opponents price in CompetitorLinearRatio1 so that we don't get a divided by zero error.
We don't know if we have an off-by-one-error in our policy and market functions. Check that and define our solution.
Edit: (investigated by @SinNeax) Since the RLAgents determine/randomly choose a price between
0
andconfig.maxprice - 1
, we needed to increment the opponents price inCompetitorLinearRatio1
so that we don't get adivided by zero
error.https://github.com/hpi-epic/BP2021/blob/2535ba00d9ba99405d6874207e923f6f4400701b/recommerce/market/linear/linear_vendors.py#L33-L48
That could be the only off-by-one error because all of the other vendors never recognize the opponents' prices with an increment.