# Report Draft
###### tags: `others`

## Materials, Design, and Methods

We develop a graph-based stochastic epidemic model. Based on the census tract and worker flow information in Taiwan according to [], The graph is generated by clustering the population into a collection of subsets called **contact groups**, which is described as a complete graph mathematically. With nodes being individuals and weighted edges being contact probability between pairs, the transmission are simulated across eight disease states. With some Sars-Cov-2 characteristics, these states are described in Fig ? , and the infection is based on the edges on the graph.
### Initialization Module
The goal of the initialization module is to build a "graph" according to preprocessed data. A "graph" that is convenient for the simulation module to read should contain the following information:
1. number of nodes/individuals
2. number of periods per day
3. number of contact groups
4. number of age groups
5. number of contact matrices
6. number of census tracts
7. type of each contact group
8. contact matrix of each type of contact groups
9. age group, census tract, and contact groups that each individual belongs to

For the simplicity, some of the above parameters are fixed. First, each day is only divided into 2 periods, daytime and night. Second, there are 4 age groups: 0-4, 5-18, 19-64, and 65+. Third, there are 10 types of contact groups:
1. Household: 7 persons(all ages) each
2. Household cluster: 4 households each
3. Play group: 4 persons(0-4) each
4. Daycare center: 14 persons(0-4) each
5. Elementary school: 79 persons(5-18) each
6. Middle school: 128 persons(5-18) each
7. High school: 155 persons(5-18) each
8. Work group: 20 persons(19-64) each
9. Neighborhood: 4 household clusters each
10. Community: 2000 persons(all ages) each

Each individual may belong to several contact groups simutaneously. However, the contact groups at daytime and at night are disjoint. The next paragraph introduces how we assign contact groups for each individual.

## Implementation and Experiments

We divide our implementation into four parts, Preprocessing, Initialization, Simulation, and Visualization module respectively. The data flow is shown in the Fig. ?.

*insert graph*
- preprocessing
- raw data
- Census tract info
- worker flow info
- flu param info
- init-readable data
- initialization
- graph info
- graph construction
- init infector selection
- vacc order selection
- simulation
- statistic info
- visualization
- vis

### Simulation Module