-
Notifications
You must be signed in to change notification settings - Fork 1
/
ReadMe.txt
46 lines (32 loc) · 1.93 KB
/
ReadMe.txt
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
Author: Matthieu Nadini. 03/24/2020
For questions, please contact matthieu.nadini@gmail.com
###########################################################
Sample code for the article "Epidemic spreading and vaccination strategies in a urban-like environment"
- It is about our two-dimensional agent-based model that encapsulates the typical core-peripheral structure of many urban environments
- It considers the SIR disease process only, with few modifications it can be adapted to the SIS as well
###########################################################################
The codes requires either python 2.x or python 3. and c++11 with the boost library
First, you need to modify the run_code.py and type where your boost library is
Then, in order to run the code, type
- python run_code.py
###############################################################################
In Input folder
- Parameters.txt
It contains all input parameters in the model
You may change the value of the input parameters as you like.
- Meaning_parameters.txt
It explains the meaning of all input parameters
- N_runs.txt
Number of independent runs the algorithm will run with the same parameters
##############################################################################
In Output folder
All output files have the following name:
- "Lambda_over_mu_"+NumberToString(int(100.*pjump))+"_"+NumberToString(int(100.*noise))+"_"+NumberToString(run_i)+".txt"
NumberToString(int(100.*pjump)) multiply the input parameter pjump per 100 and convert it into a string
NumberToString(int(100.*noise)) multiply the input parameter noise per 100 and convert it into a string
NumberToString(int(run_i)) convert the number of indipendent run into a string
Example with pjump = 0.5, noise = 0.3, and run_i = 3:
"Lambda_over_mu_50_30_3.txt"
All output file contains the following info:
- first column represents the timestamp
- second column represents the fraction of recovered