Hyper parameters

Hyper-parameter: default value	Type & range	Description
device: "cpu"	{"cpu", "gpu"}	Runs simulations on CPU or GPU.
op: null	{"BalaGoyalOp", "OConnorWeatherallOp", "NoOp"}	Simulates given operator (or model).
seed: 0	Integer, 0 or higher	Generates random number with given seed.
epsilon: 0.0	(-0.5, 0.5)	Action B yields Bernoulli payoffs with probability 0.5 + epsilon.
trials: 10	Integer, 1 or higher	Sets number of Bernoulli trials per simulation step.
lowerupper: 0.99	[0.0, 1.0)	Network converges when all agent beliefs are greater than lower-upper bound.
upperlower: 0.50	[0.0, 1.0)	Network converges when all agent beliefs are less or equal to upper-lower bound.
mistrust: 0.0	Float;	Determines how far apart beliefs have to become before an agent begins to ignore the evidence of neighbours.
antiupdating: False	True or False	Uses anti-updating rule.
init.kind: "uniform"	{"uniform", "gaussian", "constant"}	Distribution agent beliefs are initialised
init.uniform.lower: 0.0	Float;	The lower bound of the uniform distribution.
init.uniform.upper: 1.0	Float;	The upper bound of the uniform distribution
init.gaussian.mean: 0.0	Float;	The mean of the Gaussian distribution
init.gaussian.std: 1.0	Float;	The standard deviation of the Gaussian distribution
init.gaussian.lower: -2.0	Float;	The minimum cutoff value
init.gaussian.upper: 2.0	Float;	The maximum cutoff value
init.gaussian.attempts: 4	Integer;	Number of attempts to generate a truncated Gaussian distribution
init.constant.value: null	Float;	The initial belief value
network.kind: null	{"sample", "wheel", "cycle", "start", "line", "grid", "random", "complete", "karate", "wattsstrogatz", "barabasialbert", "snap", "ogb"}	Generates network of given kind.
network.size: null	Integer;	Sets number of nodes (i.e., agents) in the network.
network.directed: False	True or False	Determines whether edges are directed or not.
network.selfloop: True	True or False	Creates self-loops for each node in the network.
network.random.seed: null	Integer, 0 or higher	Generates random networks with given seed.
network.random.probability: 1.0	Float;	Probability for edge creation
network.wattsstrogatz.knn: 2	Integer;	Joins each node with its knn nearest neighbours in a ring topology.
network.wattsstrogatz.seed: null	Integer, 0 or higher	Generates Watts–Strogatz small-world graphs with given seed.
network.wattsstrogatz.tries: 100	Integer;	Number of attempts to generate a connected graph.
network.wattsstrogatz.probability: 1.0	Float;	Rewires each edge with given probability.
network.barabasialbert.attachments: 1	Integer;	Number of edges to attach from a new node to existing nodes
network.barabasialbert.seed: null	Integer, 0 or higher	Generates Barabasi-Albert networks with given seed.
network.snap.name: null	{LiveJournal, LiveJournal1, EgoFacebook, EgoTwitter, Twitter2010}	Loads SNAP dataset. For more information, see datasets/snap.py
network.ogb.name: "collab"	{collab}	Loads OGB dataset. For more information, see datasets/ogb.py
logging.enabled: True	True or False	Enables logging on standard output (i.e., terminal)
logging.interval: 1	Integer, 1 or higher	Reports simulation progress every interval steps. Value of 0 disables it.
simulation.results: "auto"	String or "auto"	Stores results in specified directory. If directory does not exist, the program creates it. If "auto", stores and organises results in polygraphs-cache directory under user's home directory.
simulation.repeats: 1	Integer, 1 or higher	Repeats a simulation 1 or more times
simulation.steps: 0	Integer, 0 or higher	Permorms at most steps simulation steps. If 0, simulation runs until network has converged.