--- tags: Legacy --- # Streamr Research Journal ## Links - [Test Grid](https://docs.google.com/spreadsheets/d/1IJFVdcyH_20Xp8XdLiXoPpHgsR6IO4ZeqWsubtNJlu4/edit#gid=0) ## Entries ### Oct 06th 2021 - [PSuU PoC Experiment](https://github.com/BlockScience/streamr-agreements/blob/6e6654d85f82cda99c6706ce013abcab1d4b719b/Parameter%20Selection.ipynb) - This PoC has a single goal called "System SLA", which is measured through fixed thresholds on the `Avg Downtime` (less than 12%) and `Return on Stake` (more than 300%) KPIs - Parameters were cartesian swept at `min_brokers` (bet. 1 and 3), `max_brokers` (bet. 4 and 8) and `Allocation` (bet. 15 and 35) - N_mc = 0 - Conclusions: - KPIs - Generally, the return on stake is around 0% and 1000%, with outliers to up to 2000% (mode: ~300%) - The avg downtime is between 5% and 14% - The Return on Stake seems to have a linear relationship with the allocation_per_epoch, with negative intercept - Average Downtime % tends to increase linearly with `min_brokers` - System SLA - `allocation_per_epoch` carries about 60% of the weight for deciding if a scenario is going to achieve the SLA - `min_brokers` is the least important, with ~10% of the weight ### Oct 04th 2021 - Scoped experiment for having a minimal demonstration of the PSuU methodology on the Streamr Agreements Model. - Will involve implementation of a basic PSuU workflow ### Oct 02th 2021 - PSuU scaffold written at https://hackmd.io/SV6pDiuKSkuGVIaJVr1INQ