Package 'levelnet'

two-mode network from a data.frame

Description

Create a two-mode network from a data.frame

Usage

bipartite_from_data_frame(df, type1, type2)

Arguments

df	data.frame
type1	column name of mode 1
type2	column name of mode 2

Value

two mode network as igraph object

Author(s)

David Schoch

---

Bill cosponsorship data for the 115th Senate

Description

Bill cosponsorship data for the 115th Senate

Usage

cosponsor_senate_15

Format

a data frame of bill cosponsorships

References

govtrack.us

---

Disparity Filter

Description

Extract significant edges of a weighted network or one-mode projection with the disparsity filter.

Usage

disparsity_filter(g, proj = "true", alpha = 0.05, cut_mode = "or")

Arguments

g	igraph object. either two-mode or weighted network
proj	string. Which mode to project on ("true"/"false")
alpha	significant level
cut_mode	'and' (retain edge if both directions are significant) or 'or' (retain edge if one direction is significant)

Value

backbone of weighted network as igraph object

Author(s)

David Schoch

References

Serrano et al. (2009). Extracting the multiscale backbone of complex weighted networks

---

Permutation induced by Fiedler vector

Description

Returns the permutation induced by sorting the Fiedler vector of the Laplacian matrix of a graph

Usage

fiedler_order(g, mode = "cols")

Arguments

g	an igraph object or a (0,1)-matrix
mode	one of "mcl" (clique vertex matrix), "cols" (Lazarus count of columns) "rows" (Lazarus count of rows) or "sym" (Lazarus count of both columns and rows).

Value

numeric vector

Author(s)

David Schoch

---

Random Indifference Graph

Description

Create a random indifference graph. An indifference graph is an interval graph where intervals have length 1.

Usage

graph_indifference(n, r = 2)

Arguments

n	number of nodes
r	radius

Details

'n' points (x) are sampled uniformly at random between 0 and 'r'. The interval is then given by (x,x+1)

Value

indifference graph as igraph object and interval representation (a,b)

Author(s)

David Schoch

See Also

[graph_interval,graph_tolerance]

Examples

graph_indifference(n = 10)

---

Random Interval Graph

Description

Create a random interval graph. In an interval graph, each node is characterized by an interval on the real line. Two nodes are connected, if their intervals overlap.

Usage

graph_interval(n, r = 2, sd = 0.5)

Arguments

n	number of nodes
r	radius (see details)
sd	standard deviation (see details)

Details

Interval graphs are created as follows. First, n random points x are created uniformly at random between 0 and 'r'. For each point, a value Y is created from a normal distribution with mean X and standard deviation is 'sd'. In this way, it is possible to control the density of the network. The larger 'r' and the larger 'sd' the more likely do intervals overlap.

Value

interval graph as igraph object and interval representation as node attribute (a,b)

Author(s)

David Schoch

See Also

[graph_indifference,graph_tolerance]

Examples

graph_interval(n = 10)

---

generate random roll-call votes based on ideology space

Description

generate random roll-call votes based on ideology space

Usage

graph_random_vote(
  M = 101,
  D = 1,
  p = 4,
  pd = 2,
  beta = 1,
  r = 9,
  noprob = 0.05,
  Nrand = 1000,
  N = 525
)

Arguments

M	number of members
D	distance between means
p	dimension of space
pd	dimensioms where distributions are separated
beta	scaling parameter for probabilistic voting
r	radius of hypersphere for random generation
noprob	probabilit of non voting
Nrand	number of randomly generated votes
N	number of votes to sample from randomly generated votes

Value

list with random votes and ideologies

Author(s)

David Schoch

References

Aldrich, John H., and Montgomery, Jacob M., and Sparks, David B. (2014). Polarization and Ideology: Partisan Sources of Low Dimensionality in Scaled Roll Call Analyses. Political Analysis 22:435-456

---

Boxicity 2 graph

Description

Create a random graph with boxicity 2.

Usage

graph_rectangle(n, r = 2, sd = 0.5)

Arguments

n	number of nodes
r	radius
sd	standard deviation

Value

Boxicity 2 graph as igraph object

Author(s)

David Schoch

---

Random Tolerance Graph

Description

Create a random tolerance graph. A tolerance graph is an interval graph, where nodes are only connected if the overlap is larger than a nodes tolerance level. These graphs are directed.

Usage

graph_tolerance(n, r = 2, sd = 0.5, tol = 0.5)

Arguments

n	number of nodes
r	radius (see details)
sd	standard deviation (see details)
tol	tolerance

Details

Tolerance graphs are created as follows. First, n random points x are created uniformly at random between 0 and 'r'. For each point, a value Y is created from a normal distribution with mean X and standard deviation is 'sd'. In this way, it is possible to control the density of the network. The larger 'r' and the larger 'sd' the more likely do intervals overlap. When overlaps are calculated, it is checked whether the overlap is larger than the tolerance of the node. If so, the edge is included.

Value

tolerance graph as igraph object and interval representation and tolerance as node attributes

Author(s)

David Schoch

See Also

[graph_interval,graph_indifference]

Examples

graph_tolerance(n = 10)

---

helper function

Description

small helper functions

Usage

clique_vertex_mat(g)

is_bipartite1(g)

Arguments

g	igraph object.

Value

igraph object

Author(s)

David Schoch

---

Check whether graph is interval graph

Description

Check whether graph is interval graph.

Usage

is_interval(g)

Arguments

g	igraph object

Details

This function is not very efficient since it relies on the clique vertex matrix. More efficient linear time algorithms will be implemented in the future.

Value

Logical scalar, whether graph is an interval graph

Author(s)

David Schoch

---

Laplacian Vectors

Description

Returns Laplacian eigenvectors associated with the k smallest positive eigenvalues

Usage

laplacian_vectors(g, k = 2)

Arguments

g	igraph object
k	number of vectors to return

Value

data.frame of vectors

Author(s)

David Schoch

---

Lazarus Count

Description

Calculates the Lazarus count of a matrx/graph.

Usage

lazarus_count(g, perm = NULL, mode = "cols")

Arguments

g	either an igraph object or a (0,1)-matrix
perm	permutation or NULL
mode	one of "mcl" (clique vertex matrix), "cols" (Lazarus count of columns) "rows" (Lazarus count of rows) or "sym" (Lazarus count of both columns and rows).

Details

The Lazarus count of a matrix is the number of "holes" in each column. A hole is a number of zero entries surrounded by ones. For an interval graph, this count is zero for the [clique_vertex_mat]. If 'perm' is NULL, a row permutation based on the Fiedler vector of the Laplacian is calculated.

Value

Lazarus count of g

Author(s)

David Schoch

Examples

set.seed(10)
#the lazarus count of an interval graph is zero
g <- graph_interval(n = 10,r = 1)
lazarus_count(g, mode = "mcl")

---

Multisweep Lex-BFS

Description

Multisweep lexicograpical BFS

Usage

multiLexBFS(g, k = 4)

Arguments

g	igraph object
k	number of sweeps

Details

LexBFS is used to recognize interval graphs. Not fully implemented yet.

Value

permutation

Author(s)

David Schoch

---

Box representation from permutations

Description

Create a box representation from permutations

Usage

perm2box(g, perm, dim)

Arguments

g	igraph object.
perm	integer vector of length n times dim
dim	integer. dimensionality of boxes

Value

coordinates

References

Chandran, L. S., Francis, M. C. & Sivadasan, N. Geometric representation of graphs in low dimension using axis parallel boxes. Algorithmica 56, 129.

---

Flexible Stochastic Degree Sequence Model

Description

Flexible Stochastic Degree Sequence Model.

Usage

fsdsm(
  g,
  row_constr,
  proj = "true",
  model = "logit",
  max_iter = 1000,
  alpha = 0.05,
  params = list(b0 = 0.1, b1 = 5e-05, b2 = 5e-05, b3 = 5e-05, a = 0.01),
  verbose = FALSE
)

sdsm_prob(
  g,
  proj = "true",
  model = "logit",
  max_iter = 1000,
  params = list(b0 = 0.1, b1 = 5e-05, b2 = 5e-05, b3 = 5e-05, a = 0.01),
  verbose = FALSE
)

Arguments

g	igraph object. The two-mode network
row_constr	constraint matrix
proj	string. Which mode to project on ("true"/"false")
model	string. which link to be used ('logit','probit','cloglog' or 'scobit')
max_iter	number of randomly sampled networks
alpha	significance level
params	named parameter list for scobit model
verbose	print status during execution

Details

a flexible implementation of the stochastic degree sequence model, allowing for the addition of constraints (use sdsm from the backbone package for the regular model)

Value

backbone of one-mode projection

Author(s)

David Schoch

References

Neal, Zachary (2014). The backbone of bipartite projections: Inferring relationships from co-authorship, co-sponsorship, co-attendance and other co-behaviors

---

sdsm model diagnostics

Description

check which binary outcome model fits the data best

Usage

sdsm_diagnostic(
  g,
  proj = "true",
  iter = 10,
  verbose = FALSE,
  params = list(b0 = 0.1, b1 = 5e-05, b2 = 5e-05, b3 = 5e-05, a = 0.01)
)

Arguments

g	igraph object. The two-mode network
proj	string. Which mode to project on
iter	number of fits per model
verbose	logical. print additional information (default: FALSE)
params	named parameter list for scobit model

Value

rmse and runtime of various models

Author(s)

David Schoch

---

Supergraph with given boxicity

Description

Create a supergraph with given boxicity using simulated annealing (SA)

Usage

superbox_graph(
  l,
  dim = 1,
  perm = NULL,
  iter = 15000,
  temp = 10,
  tmax = 5,
  verbose = FALSE
)

Arguments

dim	integer. target boxicity
perm	starting permutation for SA. If NULL, a random permutation is created
iter	integer. number of iterations for SA
temp	integer. starting temperature for SA
tmax	integer. number of function evaluations at each temperature for SA
verbose	logical. print report during SA (defaults to FALSE)
g	igraph object

Value

a list with entries

perm	permutation vector. All permutations are concatenated to one long vector
ged	graph edit distance from original graph
A	adjacency matrix of supergraph with given boxicity

References

Chandran, L. S., Francis, M. C. & Sivadasan, N. Geometric representation of graphs in low dimension using axis parallel boxes. Algorithmica 56, 129.

---