Network Edge Random add/del Restructure

[vmgaribay]

In an effort to reconceptualize the disruptive network dynamics functions of random add/delete edges in a way that fits better with a tensor-based approach, a new noise-inducing function which temporarily bumps the weights of a small proportion of random edges in the network is being instated. The projected flow will be:

-select random node pairs according to a given proportion or probability parameter (possibly without replacement to avoid self links and bidirectional conflicts) -generate random weights (0-1) matching the number of node pairs -create bidirectional edges for any pairs not already connected -set the edge weights according to the generated tensor

-the trading/exchange of capital then occurs

-the number of edges created in the above process are then deleted at random from the network

-the weights are then updated/recalculated before the process repeats(either end of t or beginning of t+1, whichever makes more sense, probably the latter)

[vanlankveldthijs]

Desired process order:

    dl = local_attachment
    weight_update
    dr = random_edge_noise

    trade_money

    link_deletion(dl+dr)

    agent_update
    data_collection

[vanlankveldthijs]

Restrictions guiding above process order:

• both new local attachments and added random noise should be used when trading money.
• number of links deleted should be the same as number of links created.
• links should be deleted after trading money.
• updating weights (based on agent similarity) should not affect random noise.
• agents should be updated before collecting data

[meiertgrootes]

this has been implemented and forms part of development