New Engine: Stochastic Block Model (sbm)

[sadrasabouri]

What does this implement/fix? Explain your changes.

This PR added a new engine: stochastic block model.
This generation model partitions n vertices into k set: ${C_1, C_2, \cdots, C_k}$, then a matrix $P_{k \times k}$ would assign the probability of existence of an edge between $v_i$ and $v_j$ with probability $P_{ij}$.

Also some minor edits for prior engines have been applied.

Any other comments?

Two points that need to be clarified:

1. Due to the technical debt from the prior inputting system implementation I couldn't add validation for block_sizes and probability_matrix. Let me know what we should do.
2. I implemented self-loop for this engine but we should note that the probability of self-loop is $P_{ii}$ for $v_i$ as well. If you think that's not appropriate we can remove it.

[sepandhaghighi]

@sadrasabouri Thanks for your efforts 💯
I believe that obtaining probability_matrix and block_sizes directly via the CLI is not advisable due to size issue! Instead, we can prompt the user for parameters such as intra_probability, inter_probability, and number_of_blocks. Based on these inputs, we can generate the probability_matrix and block_sizes, restricting custom edits to the config mode.

[codecov]

Codecov Report

Attention: Patch coverage is 78.78788% with 14 lines in your changes missing coverage. Please review.

Project coverage is 96.87%. Comparing base (44afa5f) to head (5872946).
Report is 72 commits behind head on dev.

Files with missing lines	Patch %	Lines
pyrgg/functions.py	36.37%	12 Missing and 2 partials ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##              dev     #189      +/-   ##
==========================================
- Coverage   98.33%   96.87%   -1.46%     
==========================================
  Files           4        8       +4     
  Lines         419      637     +218     
  Branches       68       91      +23     
==========================================
+ Hits          412      617     +205     
- Misses          4       15      +11     
- Partials        3        5       +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[sadrasabouri]

Hey @sepandhaghighi, I applied the changes. This PR needed an immediate post-merge PR which takes into consideration the following points:

• Add new tests for warnings to increase the coverage
• Modify save/load config to make it more engine-specific. We should only save engine-related information in the config file, maybe change the extension to represent the engine, and check for the formatting. An example of a config file now generated for SBM model is presented bellow.

{
  "file_name": "asd",
  "number_of_files": 1,
  "vertices": 10,
  "max_weight": 1,
  "min_weight": 1,
  "min_edge": 0,
  "max_edge": 0,
  "edge_number": 0,
  "sign": true,
  "output_format": "JSON",
  "weight": true,
  "engine": "sbm",
  "direct": false,
  "self_loop": false,
  "multigraph": false,
  "config": true,
  "probability": 0.5,
  "blocks": 3,
  "inter_probability": 0.1,
  "intra_probability": 0.8,
  "block_sizes": [
    3,
    3,
    4
  ],
  "probability_matrix": [
    [
      0.9,
      0,
      0
    ],
    [
      0,
      0.9,
      0
    ],
    [
      0,
      0,
      0.9
    ]
  ],
  "pyrgg_version": "1.6"
}

[sepandhaghighi]

@sadrasabouri Thanks for your efforts 💯
I believe we can eliminate self_loop and direct from the inputs. These parameters can both be managed through the probability_matrix.

[sepandhaghighi]

[Warning] Vertices ...

[sepandhaghighi]

1. Remove In CLI mode,
2. [Warning] Stochastic ...

[sadrasabouri]

@sadrasabouri Thanks for your efforts 💯 I believe we can eliminate self_loop and direct from the inputs. These parameters can both be managed through the probability_matrix.

I agree with removing direct, but that's not the case for the self_loop. probability_matrix identifies the probability of blocks, not nodes.