Table 2.
Parameters of ZoRLA and preZoRLA2.
Symbol | Description | Range | Used |
---|---|---|---|
pi,t | Clearance effort of SSi during simulation cycle t: the probability that SSi clears a detected compound | [0,1] | 0–1a |
Qi,t | Estimate of SSi's long-term, discounted cost as of events during simulation cycle t | (−∞,∞) | 0–3500 ZoRLAa |
0–1.2 preZoRLA2a | |||
gi | The value of the P-to-P b-signal gradient used by SSi | [0, ∞) | 0–200 ZoRLA |
0–1 preZoRLA2 | |||
pBypass | Probability that a toxin bypasses an SS without being detected | [0,1] | 0.6–0.99 ZoRLA |
0.9 preZoRLA2 | |||
h | A constant that adjusts the randomness of decisions | [0, ∞) | −70 ZoRLA |
0.1 preZoRLA2 | |||
α | Learning rate: determines the weight of newly observed costs in calculation of Q | [0,1] | 0.1 |
β | Discount factor: determines the current importance of future costs | [0,1] | 0.5 ZoRLA |
0 preZoRLA2 | |||
ka | Cost of detecting one r-signal | [0, ∞) | 0–150 ZoRLA |
0–20 preZoRLA2 | |||
kc | Cost of clearing one compound | [0, ∞) | 1 |
Additional parameters for ZoRLA, only | |||
(Xi,Yi) | Coordinates of SSi with respect to the center of the grid space | ℜ2 | (4,4) – (28,28) |
η | Uniform random noise (Eq. (3)) | [0, grid radius] | [0, 15.5] |
Simulation result.