dwave-examples / structural-imbalance Goto Github PK

Demo for analyzing the structural imbalance on a signed social network.

License: Other

Python 100.00%

defense optimization intermediate social-network

structural-imbalance's Introduction

Structural Imbalance in Signed Social Networks

Social networks map relationships between people or organizations onto graphs, with the people/organizations as nodes and relationships as edges; for example, Facebook friends form a social network with friends represented as nodes with connecting edges. Signed social networks map both friendly and hostile relationships by setting the values of the edges between nodes with either plus ("+") or minus ("-") signs. Such networks are said to be structurally balanced when they can be clearly divided into two, with friendly relations (represented by positive-valued edges) in each faction, and hostile relations (negative-valued edges) between these factions.

The measure of structural imbalance or frustration for a signed social network is the minimum number of edges that violate this rule.

Social theory suggests that increased frustration predicts social instability. In the context of militant organizations, this can result in increased violence.

This demo calculates and shows structural imbalance for social networks of militant organization based on data from the Stanford Militants Mapping Project.

Usage

To run the demo, execute one of the following two commands:

A. Local CPU Execution

python structural_imbalance.py --cpu

B. D-Wave System Execution

python structural_imbalance.py --qpu

Code Specifics

The demo fetches data from the Stanford Militants Mapping Project, calculates the networks, and saves a PNG graphic file of the imbalance network in the root directory of your copy of the demo repository.

Additional command line arguments are available to control the behavior of the demo, such as the region that is considered and whether to display a plot rather than saving a PNG file. Information about the command options is available via:

python structural_imbalance.py --help

License

The translation and solving code is released under the Apache License 2.0. See LICENSE file.

The dataset is used with permission from the Stanford Militants Mapping Project.

References

Mapping Militant Organizations, Stanford University, last modified February 28, 2016, http://web.stanford.edu/group/mappingmilitants/cgi-bin/.

structural-imbalance's People

Contributors

Stargazers

Watchers

structural-imbalance's Issues

Use LeapHybridSampler

Currently, the sampler is hardcoded in the demo.

Either change to use LeapHybridSampler or allow the user to pick the sampler with the set of parameters.

In the demo, the results are plotted and saved. Display them instead.

@arcondello
@randomir

Skip tests that require the QPU when no QPU is available

And possibly run them with classical samplers. This would let us run the tests on PRs.

Remove old demo

#25 added a new implementation of the demo but did not go so far as to clean up the old one.

Map loader fails on Windows

Traceback (most recent call last):
  File "C:\Users\circleci\repo\demo.py", line 86, in <module>
    gssn = GlobalSignedSocialNetwork(qpu)
  File "C:\Users\circleci\repo\dwave_structural_imbalance_demo\interfaces.py", line 60, in __init__
    maps['Global'] = global_signed_social_network()
  File "C:\Users\circleci\repo\dwave_structural_imbalance_demo\mmp_network\loader.py", line 34, in global_signed_social_network
    next(data_iter)
  File "C:\Users\circleci\repo\python\tools\lib\encodings\cp1252.py", line 23, in decode
    return codecs.charmap_decode(input,self.errors,decoding_table)[0]
UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 6347: character maps to <undefined>
ETraceback (most recent call last):
  File "C:\Users\circleci\repo\demo.py", line 86, in <module>
    gssn = GlobalSignedSocialNetwork(qpu)
  File "C:\Users\circleci\repo\dwave_structural_imbalance_demo\interfaces.py", line 60, in __init__
    maps['Global'] = global_signed_social_network()
  File "C:\Users\circleci\repo\dwave_structural_imbalance_demo\mmp_network\loader.py", line 34, in global_signed_social_network
    next(data_iter)
  File "C:\Users\circleci\repo\python\tools\lib\encodings\cp1252.py", line 23, in decode
    return codecs.charmap_decode(input,self.errors,decoding_table)[0]
UnicodeDecodeError: 'charmap' codec can't decode byte 0x9d in position 6347: character maps to <undefined>

Restore commented out tests

a6a4a9f comments out two tests because they are unix-specific. If we wish to continue running these integration tests this way, we should fix and restore them.

CI failing due to "no embedding found"

The unit tests do pass for me locally with DW_2000Q_6 and Advantage. Is there a different QPU architecture that is being used by Circle CI? The fix may be to use a --region flag to define a smaller problem when running structural_imablance.py in the unit test.

Aerospace tag?

I'm having trouble seeing how the aerospace tag applies to this one.

Demo defaults to QPU if CloudClient is installed

Users that have installed the factoring demo will have Cloud Client installed so the following sequence to run on CPU will fail:

pip install .
python demo.py

The demo will try to run on the QPU without having installed .[qpu] --extra-index-url https://pypi.dwavesys.com/simple