MeetingAgendaDrive.txt

Meeting Agenda:
>>> rewrite clustering section 
>>> rewrite flux section
>>> rewrite conclusion
>>> P(AV-AV) analysis: 
definition: freq / freqtot ; freq = running total number of time an AV is front of another AV , freqtot = running total number of times a car is in front of
observation: reaches equilibrium in all systems
rate of convergence interpretation
>>> analysis:
experiment 2:
fds comparable
fluctuation in type aware smallest
AVs better in oppo and aware
P(AV-AV) similar for base and aware (0.08), oppo has 0.1 and different trend
bad fit for overall fd (need to fix)
Conclusion: not much impact on overall traffic flow
oBSERVATION: 
at time period 1000 (one third occupation rate of road) : P(AV-AV) started to stabilize for all cases 
experiment 1:
low density:
zero clusters for base
comparable size for oppo and aware
fd better for base than oppo than aware (code fault?) —> modelling explanation
aware less transient
crit density:
aware more  clusters and less transient than others
base scenario also shows considerable clustering
p(AV-AV) trend depends on initial conditions and equilibrium establishes
Conclusion: 
aware leads to stronger clustering effects because
lower transiency (time distribution greater)
phenomenon stronger at lower densities as more mobility for AV to follow another AV
———————————————————————————————————-
  AV 4 5	|	|	|	|	|	|HV	|	|	|	
———————————————————————————————————-
        5 5	|	|	|	|	|	|AV	|	|	|	
———————————————————————————————————-
In the situation presented above, in the type aware case the AV would switch lanes leading to greater clustering where a type indifferent AV would not. Such 
phenomenon can only occur if a) such spaces are available (low density) and b) incentive to trail AV exitsts
higher number of overall clustering
higher numbers of distinct clusters
clustering effects most prominent at lower densities
clustering not random (base scenario evidence)
clustering cause:
opportunistic behaviour 
frequency of jam waves (direct correlation as seen through density)
availability of space (if clustered will stay clustered at low space availabilty)
opportunistic and type aware behaviour similar except for lane change dynamics where there are cases where type aware models are incentivized to change lanes where opportunistic models are not
>>> combined plot for experiment 1 and experiment 2 —> ideas from Dr. Li
>>> address changes according to comments
sensitivity analysis: not important for now
explanation and justification of AV behavior
Reviewer 1 comment 1
circular trajectories (if done then the lane sizes will be different or not ?) 

Dr. Li work:
>>> writing: paper, theorem and assumption differentiate, change to Physica A template
>>> sim re-run: P(headways) for all 4 types of headways
>>> math model: analysis of 4 car scenario  game theory stability equilibrium
>>> think: explanation of everything and coherent argument/build up