Depolarisation of opinions

Link to paper

Authors:

• Christopher Currin

---

1. Clone repo

Options

• git clone [email protected]:ChrisCurrin/opinion_dynamics.git
• git clone https://github.com/ChrisCurrin/opinion_dynamics.git

cd opinion_dynamics

2. Create environment from requirements using conda

conda env create -f environment.yml

conda activate opinion_dynamics

Target Python version: >= 3.6

3. Play around with notebook

jupter notebook opdynamics.ipynb

jupter lab opdynamics.ipynb

4. Run desired scripts

python -m scripts.<script name>

5. Use as a module from the terminal

e.g.

python -m opdynamics --plot=all

Run a simulation with noise with the -D <value> parameter.

python -m opdynamics 1000 10 2 3 -D 0.01 -beta 2 --activity negpowerlaw 2.1 1e-2 -r 0.5 -T 10 --plot summary --save -v

Compare noise parameters by passing multiple values to -D

python -m opdynamics 1000 10 2 3 -D 0.01 0.1 -beta 2 --activity negpowerlaw 2.1 1e-2 -r 0.5 -T 10 --plot summary --save -v

---

Development

Note that the code is formatted using the black Python module.

Update environment.yml using conda env export --from-history > environment.yml

---

TODO:

• Use alternate peak detection alogiorthms
  • peakutils
  • scipy.signal.find_peaks_cwt
• Equations:
  • 
  • 
  • -> 1 other agent versus sample mean of other agents
• Allow simulation.run_product to be run asyncronously.
• Create metric for determining whether opinions are polarised or uniform.
• Add sample_size to SampleChamber.__repr__
• Test more efficient get_social_interaction
• Test more efficient get_connection_probabilities
• Test and plot results for "internal noise".
• Restrict internal noise to be from an agent of opposite opinion.
• Implement noise as coming from agent - "internal noise".
  • with sigmoid transformation (tanh):
    • 
    • 

• plot vs
  • from
  • from
  • with different values
  • with different values
  • with different values
  • with different values

• [low priority] Implement other SDE solvers using https://pypi.org/project/diffeqpy/
• change density activity vs opinion plot to be density per activity
• Apply noise after time
• Calculate and