Network Failures - Servicer Load Balancing Threshold Inequality: interruption timestep?

[jshorish]

The load balancing threshold inequality states

The fraction of the Monte Carlo simulations for which the average network load balance recovers to within $\bar d_1$ (as a fraction) of the initial network load balance before interruption should be at least $\bar d_2$.

Thus, the timestep of the service/servicer interruption must be known, to pull from the cadCAD results DataFrame df the average load balancing statistic (KPI-C) immediately prior to interruption. Is this timestep recorded anywhere within df, or will this need to be 'expert knowledge' added manually?

[SeanMcOwen]

Within model/configs/events.py, the variable for time gives the time that it happens for a given string identified event

[SeanMcOwen]

@jshorish Also let me know to what extent you might want/need convenience functions to grab stuff like this automatically. A lot of these are on the later roadmap but if it makes your workflow faster I can prioritize!

[jshorish]

Thx @SeanMcOwen, probably no need for helper functions now given other priorities, apart from possibly extracting event data--but I can likely grab this from the dataframe in any event.

[SeanMcOwen]

I have a simple pull function here: https://github.com/BlockScience/PocketSimulationModel/blob/934bdadaf6a5d82a69e3b219e6d7323c11c1f189/model/config/events.py#L127

Does that work for closing out this issue @jshorish ?

[jshorish]

@SeanMcOwen Looks great, thanks!