Meeting Note 4/2 Init, Vacc Conf Detail

# Meeting Note 4/2 Init, Vacc Conf Detail ###### tags: `meeting note` ## Vacc Strat Conf input side: ``` strategy [strategy parameter] rollout 10 first_dose_date 2 second_dose_date 10 ``` - strategy(string): 策略名稱 - rollout(int): 每天施打人數(白天打) - second dose date(int): 開始考慮第二劑 e.g. ``` infect_prob rollout 10 first_dose_date 1 second_dose_date 2 ``` - the `date` is 1-index simulation side: - strategy parameter: if the strategy is location_first, then change to *location_id* ## Vaccine Strategies: Greedy Define a ***heuristic function(score)*** for nodes, which can be measured by 1. Sum of transmission probabilities. - state, age of its neighbors 3. Age-dependent death rate 4. Potential contribution to YLL (Year Life Lost)  ### Second Dose After the first dose, with the resistance, its score is decreased. --- In order to maintain the **score** of each node in every period We might need to make the following operations as efficient as possible 1. update scores (sorted after update) - neighbor change state (e.g. $I \rightarrow R$ or $E\to I$) 3. find top $k$ nodes - vaccine rollout $k$ 5. insert nodes - neighbor $S \to E$ Data structure candidates: 1. Array (update all nodes' score first, then sort the array) 2. Max-Heap * [Quake heap](https://tmc.web.engr.illinois.edu/heap_ianfest.pdf) * Fibonacci heap * Pairing heap 3. Segment tree ??? | | Increase score | Decrease score | Remove-Max | Insert | | ---------- |:--------------:|:--------------:|:----------:|:------:| | Complexity | $O(1)$ | $O(\log({n}))$ | $O(\log({n}))$ | $O(1)$ | - Decrease score = Increase score + Remove-Max + Insert  ### Second dose Set a "second dose date" - Before that date: Separate the nodes that have received first shot from those who haven't - After that date: Merge these two kind of nodes. ## Init Infect Conf ``` k loc_1 n_1 loc_2 n_2 . . . loc_n n_k ``` - $k$組init infect - $loc_i,n_i$ - 地區要init infect幾個人 ## Concluding what Data Process should do - vacc strat - replace census tract name with id(if neccesary) - init infector - replace census tract name with id ## Initialization should **NOT** do (Removing) - process vaccine strategy - vaccine order - init infector ## TO DO - distinguish implementation from declaration ## Report next week - paper structure - discuss data structure - heap? - main objective - with individual modeling, we have more vacc strat - find some strats outperform previous ones on mitigating infections, deaths or YYL. - new feature - transition diagram - $I_{pre},I_{asym},I_{sym}$ - $V_1,V_2$ - more vacc strat