NetSim (version 0.9)

change_models: Overview of change models implemented in NetSim

Description

Overview of change models implemented in NetSim. The first two and the fifth are define stochastic actor-oriented models for network or behavior change. The third defines a Jackson and Rogers node inclusion model, the fourth a Watts and Strogatz 'small world' model. For the SAOM-related models additional model specifications are necessary.

Usage

create_multinomial_choice_network_change_model(
				focalActorIndex, networkIndex, effectContainer)
	create_multinomial_choice_behavior_change_model(
				focalActorIndex, attributeIndex, effectContainer)
	create_jackson_rogers_change_model(
				networkIndex, 
				pLinkToParentNode = 1.0, pLinkToNeigborNode = 1.0, 
				nParentNodes = 1, nNeighborNodes = 1)
	create_watts_strogatz_change_model(networkIndex)
	create_attribute_multinomial_choice_network_change_model(
				networkIndex, poissonAttributeIndex, 
				updater = create_tie_swap_updater(networkIndex))

Arguments

focalActorIndex

Index of focal actor of the multinomial (network or behavior) change model

networkIndex

Index of the network that is subject to change (dependent network)

effectContainer

Container object storing effects as defined in the SIENA manual

attributeIndex

Index of the attribute container that is subject to change (dependent attribute)

pLinkToParentNode

Probability to link to a 'parent node'. See Jackson \& Rogers (2007), p.894

pLinkToNeigborNode

Probability to link to a selected neighbor of a parent. See Jackson \& Rogers (2007), p.894

nParentNodes

Number of 'parent node'. See Jackson \& Rogers (2007), p.894

nNeighborNodes

Number of neighbors considered of parent node. See Jackson \& Rogers (2007), p.894

updater

Defines a change that is considered by the focal actor after choosing a particular tie. By default, this is a tie swap updater.

poissonAttributeIndex

Index of a attribute container containing a number of concurring Poisson parameters

References

Ripley, R. M.; Snijders, T. A. B. & Preciado Lopez, P.: Manual for SIENA 4.0. (2012), Oxford: University of Oxford, Department of Statistics; Nuffield College. URL: http://www.stats.ox.ac.uk/~snijders/siena/RSiena_Manual.pdf

Jackson, M. O. & Rogers, B. W. Meeting strangers and friends of friends: How random are social networks? American Economic Review, 2007, 97, 890-915

Watts, D. J. & Strogatz, S. H. Collective dynamics of 'small-world' networks Nature, 1998, 393, 440-442

See Also

add_change_model create_tie_swap_updater create_effect_container add_to_effect_container create_effect add_effect

Examples

## The following example is taken from section 4.1 and 4.2 in package vignette
## C. Stadtfeld: "Netsim: A Social Networks Simulation tool in R" (2013)
## Examples of the other models can be taken from that source
 
# construct process state
nActors <- 21
mat <- matrix(0, nActors, nActors)
att <- c(rep(0, nActors/3),rep(1, nActors/3),rep(2, nActors/3))
network <- create_network(mat)
attributeContainer <- create_scale_attribute_container(att,
min=0, max=3, by = 1)
processState <- create_process_state()
processState <- add_network(processState, network, name = "network")
processState <- add_attribute_container(processState, attributeContainer,
name = "attribute")
networkIndex <- get_network_index(processState)
attributeIndex <- get_attribute_container_index(processState)

# construct effect container
# homophily model with basic network effects
effectContainerHomophily <- create_effect_container()
effectContainerHomophily <- add_to_effect_container(
	effectContainerHomophily,
	create_effect("density", networkIndex),
	-2.5)
effectContainerHomophily <- add_to_effect_container(
	effectContainerHomophily,
	create_effect("recip", networkIndex),
	2.5)
effectContainerHomophily <- add_to_effect_container(
	effectContainerHomophily,
	create_effect("transTrip", networkIndex),
	0.5)
effectContainerHomophily <- add_to_effect_container(
	effectContainerHomophily,
	create_effect("cycle3", networkIndex),
	-0.5)
effectContainerHomophily <- add_to_effect_container(
	effectContainerHomophily,
	create_effect("simX",
		attributeIndex,
		networkIndex,
		0.5),
	1.0)
	
# effect container of the behavior change model
effectContainerInfluence <- create_effect_container()
effectContainerInfluence <- add_to_effect_container(
	effectContainerInfluence,
	create_effect("linear",
		attributeIndex),
	0.0)
effectContainerInfluence <- add_to_effect_container(
	effectContainerInfluence,
	create_effect("quad",
		attributeIndex),
	0.0)
effectContainerInfluence <- add_to_effect_container(
	effectContainerInfluence,
	create_effect("totSim",
		attributeIndex,
		networkIndex, 10/18),
	2.0)

# Definition of SAOM network model manager
modelManager <- create_model_manager()
# assign homophily model to all actors
for (i in c(0 : (nActors - 1) ) ){
	# Poisson models
	poissonParameterInfluence <- 5
	poissonModelInfluence <- create_poisson_model(
	poissonParameterInfluence)
	
	poissonParameter <- 40
	poissonModel <- create_poisson_model(poissonParameter)
	
	# saom change models
	saomHomophilyModel <- create_multinomial_choice_network_change_model(
		i,
		networkIndex,
		effectContainerHomophily)
	behaviorSaom <- create_multinomial_choice_behavior_change_model(
		i,
		attributeIndex,
		effectContainerInfluence
	)
	# updaters
	setAttributeUpdater <- create_actor_attribute_set_updater(
	attributeIndex, i)
		
	tieSwapUpdater <- create_tie_swap_updater(networkIndex)
	
	# define model chains
	modelManager <<- add_time_model(modelManager,
		poissonModel)
	modelManager <<- add_change_model(modelManager,
		poissonModel,
		saomHomophilyModel)
	modelManager <<- add_updater(modelManager,
		saomHomophilyModel,
		tieSwapUpdater)
	modelManager <<- add_time_model(modelManager,
        poissonModelInfluence)
	modelManager <<- add_change_model(modelManager,
        poissonModelInfluence,
        behaviorSaom)
	modelManager <<- add_updater(modelManager,
        behaviorSaom,
        setAttributeUpdater)
		
} # for loop

simulator <- create_simulator(processState, modelManager, 10)
# commented out due to Windows compilation problems. Further tests needed!
## Not run: simulate(simulator)