Skip to content

GitLab

Commit 54f5c469 authored by Tiago Peixoto's avatar Tiago Peixoto
Browse files

Modify absolute_trust() algorithm 

Path sampling is done using different bias, and the edge sampling is now
done correctly. There is now only one parameter 'n_iter' which replaces
'min/max_iter' and 'epslon'.
parent e05fa740
Hide whitespace changes
Inline Side-by-side
Showing with 154 additions and 151 deletions
src/graph/centrality/graph_absolute_trust.cc
View file @ 54f5c469
......@@ -32,8 +32,7 @@ using namespace boost;
using namespace graph_tool;
void absolute_trust(GraphInterface& g, int64_t source, boost::any c,
boost::any t, double epslon, size_t min_iter,
size_t max_iter, bool reversed, size_t seed)
boost::any t, size_t iter, bool reversed, size_t seed)
{
rng_t rng(static_cast<rng_t::result_type>(seed));
......@@ -44,8 +43,7 @@ void absolute_trust(GraphInterface& g, int64_t source, boost::any c,
run_action<>()(g,
bind<void>(get_absolute_trust(), _1, g.GetVertexIndex(),
source, _2, _3, epslon,
make_pair(min_iter, max_iter), reversed,
source, _2, _3, iter, reversed,
ref(rng)),
edge_floating_properties(),
vertex_floating_vector_properties())(c, t);
......
src/graph/centrality/graph_absolute_trust.hh
View file @ 54f5c469
......@@ -38,17 +38,13 @@ struct get_absolute_trust
template <class Graph, class VertexIndex, class TrustMap,
class InferredTrustMap>
void operator()(Graph& g, VertexIndex vertex_index, int64_t source,
TrustMap c, InferredTrustMap t, double epslon,
pair<size_t,size_t> iter_range, bool reversed,
rng_t& rng) const
TrustMap c, InferredTrustMap t, size_t n_iter,
bool reversed, rng_t& rng) const
{
typedef typename property_traits<TrustMap>::value_type c_type;
typedef typename property_traits<InferredTrustMap>::value_type
::value_type t_type;
size_t min_iter = iter_range.first;
size_t max_iter = iter_range.second;
unchecked_vector_property_map<vector<t_type>, VertexIndex>
t_count(vertex_index, num_vertices(g));
unchecked_vector_property_map<vector<size_t>, VertexIndex>
......@@ -73,112 +69,130 @@ struct get_absolute_trust
for (i = (source == -1) ? 0 : source;
i < ((source == -1) ? N : source + 1); ++i)
{
typename graph_traits<Graph>::vertex_descriptor v =
vertex(i, g);
if (v == graph_traits<Graph>::null_vertex())
continue;
// walk hash set
tr1::unordered_set<size_t> path_set;
tr1::uniform_int<size_t>
random_salt(0, numeric_limits<size_t>::max());
size_t salt = random_salt(rng);
t_type delta = epslon + 1;
size_t iter = 0;
while (delta > epslon || iter < min_iter)
{
delta = 0;
typename graph_traits<Graph>::vertex_descriptor v =
vertex(i, g);
if (v == graph_traits<Graph>::null_vertex())
continue;
typename graph_traits<Graph>::vertex_descriptor pos = v;
t_type pos_t = 1.0;
t_type pweight = 1.0;
v_mark[v][vertex_index[v]] = iter+1;
size_t path_hash = salt;
hash_combine(path_hash, i);
// start a self-avoiding walk from vertex v
vector<typename graph_traits<Graph>::edge_descriptor> out_es;
vector<t_type> out_prob;
do
for (size_t bias = 0; bias < 3; ++bias)
for (size_t iter = 0; iter < n_iter; ++iter)
{
out_es.clear();
out_prob.clear();
// obtain list of candidate edges to follow
typename graph_traits<Graph>::out_edge_iterator e, e_end;
for (tie(e, e_end) = out_edges(pos, g); e != e_end; ++e)
typename graph_traits<Graph>::vertex_descriptor pos = v;
t_type pos_t = 1.0;
t_type pweight = 1.0;
v_mark[v][vertex_index[v]] = bias*n_iter + iter + 1;
size_t path_hash = salt;
hash_combine(path_hash, i);
// start a self-avoiding walk from vertex v
vector<typename graph_traits<Graph>::edge_descriptor>
out_es;
vector<t_type> out_prob;
do
{
typename graph_traits<Graph>::vertex_descriptor t =
target(*e,g);
if (v_mark[v][vertex_index[t]] <= iter)
{
if (c[*e] < 0)
continue;
out_es.push_back(*e);
if (out_prob.empty())
out_prob.push_back(c[*e]+0.1);
else
out_prob.push_back(out_prob.back()+c[*e]+0.1);
}
}
if (!out_es.empty())
{
// select edge according to its probability
typename graph_traits<Graph>::edge_descriptor e;
typedef tr1::uniform_real<t_type> dist_t;
tr1::variate_generator<rng_t&, dist_t>
random(rng, dist_t(0, out_prob.back()));
out_es.clear();
out_prob.clear();
t_type u;
// obtain list of candidate edges to follow
typename graph_traits<Graph>::out_edge_iterator e,
e_end;
for (tie(e, e_end) = out_edges(pos, g); e != e_end; ++e)
{
#pragma omp critical
u = random();
typename graph_traits<Graph>::vertex_descriptor t =
target(*e,g);
if (v_mark[v][vertex_index[t]] <
bias*n_iter + iter + 1)
{
if (c[*e] < 0 || c[*e] > 1)
continue;
out_es.push_back(*e);
if (bias != 1)
{
double prob = c[*e];
if (bias == 2)
prob = 1.0 - prob;
if (out_prob.empty())
out_prob.push_back(prob);
else
out_prob.push_back(out_prob.back() +
prob);
}
}
}
e = out_es[lower_bound(out_prob.begin(),
out_prob.end(), u) -
out_prob.begin()];
// new position in random walk
pos = target(e,g);
size_t posi = vertex_index[pos];
//update current path trust
pos_t *= c[e];
if (reversed && boost::source(e, g) != v)
pweight *= c[e];
// get path hash
hash_combine(path_hash, posi);
if (path_set.find(path_hash) == path_set.end())
if (!out_es.empty())
{
// if new path, modify vertex trust score
path_set.insert(path_hash);
t_type old = 0;
if (t_count[v][posi] > 0)
old = t[v][posi]/t_count[v][posi];
t[v][posi] += pos_t*pweight;
t_count[v][posi] += pweight;
delta += abs(old - t[v][posi]/t_count[v][posi]);
// select edge according to its probability
typename graph_traits<Graph>::edge_descriptor e;
if (bias != 1)
{
typedef tr1::uniform_real<t_type> dist_t;
tr1::variate_generator<rng_t&, dist_t>
random(rng, dist_t(0, out_prob.back()));
t_type u;
{
#pragma omp critical
u = random();
}
e = out_es[upper_bound(out_prob.begin(),
out_prob.end(), u) -
out_prob.begin()];
}
else
{
tr1::uniform_int<size_t>
random(0, out_es.size()-1);
#pragma omp critical
e = out_es[random(rng)];
}
// new position in random walk
pos = target(e,g);
size_t posi = vertex_index[pos];
// update current path trust
pos_t *= c[e];
if (reversed && boost::source(e, g) != v)
pweight *= c[e];
// get path hash
hash_combine(path_hash, posi);
if (path_set.find(path_hash) == path_set.end())
{
// if new path, modify vertex trust score
path_set.insert(path_hash);
t_type old = 0;
if (t_count[v][posi] > 0)
old = t[v][posi]/t_count[v][posi];
t[v][posi] += pos_t*pweight;
t_count[v][posi] += pweight;
}
if (!reversed)
pweight *= c[e];
// stop if weight drops to zero
if (pweight == 0)
break;
// mark vertex
v_mark[v][posi] = bias*n_iter + iter + 1;
}
if (!reversed)
pweight *= c[e];
// stop if weight drops to zero
if (pweight == 0)
break;
v_mark[v][posi] = iter+1; // mark vertex
}
while (!out_es.empty());
}
while (!out_es.empty());
++iter;
if (max_iter > 0 && iter == max_iter)
break;
}
}
#pragma omp parallel for default(shared) private(i) schedule(dynamic)
......
src/graph_tool/centrality/__init__.py
View file @ 54f5c469
......@@ -390,9 +390,8 @@ def eigentrust(g, trust_map, vprop=None, norm=False, epslon=1e-6, max_iter=0,
else:
return vprop
def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
min_iter=100, max_iter=None, reversed=False, seed=None,
ret_iter=False):
def absolute_trust(g, trust_map, source=None, vprop=None, n_iter=100,
reversed=False, seed=None):
r"""
Samples the absolute trust centrality of each vertex in the graph, or only
for a given source, if one is provided.
......@@ -409,15 +408,9 @@ def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
values, instead of all the vertices in the graph.
vprop : PropertyMap, optional (default: None)
Vertex property map where the values of eigentrust must be stored.
epslon : float, optional (default: 0.001)
Convergence condition. The iteration will stop if the total delta of all
vertices are below this value.
min_iter : int, optional (default: 100)
If supplied, this will limit the minimal number of iterations (per
source vertex).
max_iter : int, optional (default: None)
If supplied, this will limit the total number of iterations (per
source vertex).
n_iter : int, optional (default: 100)
Number of iterations (independent self-avoiding walks) per source
vertex.
reversed : bool, optional (default: False)
Calculates the "reversed" trust instead: The direction of the edges are
inverted, but the path weighting is preserved in the original direction
......@@ -425,14 +418,12 @@ def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
seed : int, optional (default: None)
The initializing seed for the random number generator. If not supplied
a different random value will be chosen each time.
ret_iter : bool, optional (default: False)
If true, the total number of iterations is also returned.
Returns
-------
absolute_trust : PropertyMap
A vertex property map containing the absolute trust vector from the
corresponding vertex to the rest of the network. Each e lement i of the
corresponding vertex to the rest of the network. Each element i of the
vector is the trust value of the vertex with index i, from the given
vertex.
......@@ -464,8 +455,14 @@ def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
such that the direct trust of the last edge on the path is not considered.
The algorithm progressively samples all possible paths, until the trust
values converge, and has a topology-dependent complexity.
The algorithm performs only an approximation of the above measure, by doing
several self-avoiding random walks per source vertex, and computing the
trust for all different paths found. Each complete walk is done by one of
three types of biased choices, which govern the probability of following a
specific edge: 1. With probability proportional to an edge's trust value;
2. With probability proportional to the complement of an edge's trust value
:math:`(1-t_{e})`; 3. All edges with equal probability. The parameter
"n_iter" controls how many walks of each type are performed.
If enabled during compilation, this algorithm runs in parallel.
......@@ -478,31 +475,31 @@ def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
>>> trust.get_array()[:] = random(g.num_edges())
>>> t = gt.absolute_trust(g, trust)
>>> print array(t[g.vertex(10)])
[ 4.51906352e-01 1.01429943e-01 0.00000000e+00 0.00000000e+00
0.00000000e+00 2.44861537e-01 0.00000000e+00 3.02651461e-01
0.00000000e+00 1.74730923e-02 0.00000000e+00 4.29689123e-03
9.92174035e-04 4.87376703e-02 7.96102941e-06 2.33366853e-01
2.54084432e-01 1.21249107e-02 1.68187723e-01 1.63299672e-02
3.63286096e-02 7.26969527e-01 2.66379387e-01 2.60306326e-03
3.00337293e-04 3.45095945e-02 4.74197414e-02 3.47040393e-01
4.15994365e-01 2.58366543e-01 1.98886295e-02 1.84395387e-01
3.01526472e-01 4.17581597e-01 5.08199168e-01 2.39325151e-01
1.16653396e-01 0.00000000e+00 2.32122209e-01 1.74458369e-01
0.00000000e+00 2.28833708e-02 8.40459672e-02 1.13238046e-01
1.15211877e-01 1.64368525e-01 5.42106779e-03 8.21849155e-01
0.00000000e+00 1.10064601e-01 3.94971607e-01 2.40544425e-01
1.81781785e-01 4.27661874e-01 1.96505012e-01 2.22664312e-01
1.05942366e-01 1.45053992e-01 3.17854637e-01 0.00000000e+00
7.19074080e-02 6.01425735e-02 1.80652580e-01 1.64203118e-06
1.51091732e-01 2.15091245e-01 6.59231333e-01 5.29298126e-01
3.39640072e-01 2.85420503e-01 0.00000000e+00 0.00000000e+00
3.76245906e-01 2.26621429e-02 2.16958451e-01 5.43678807e-01
0.00000000e+00 7.01804977e-02 1.50118832e-01 3.44142938e-03
3.63454763e-01 2.55747857e-01 1.58862377e-01 4.74257967e-02
9.79787266e-02 1.33262415e-01 4.44120403e-01 2.13954368e-02
1.65309874e-01 2.22217618e-01 2.80359630e-02 3.78916751e-01
9.22800991e-03 2.89978731e-03 9.21281004e-02 0.00000000e+00
5.20985973e-01 5.47754676e-02 6.27828075e-01 2.28123051e-01]
[ 4.34269834e-01 1.84086073e-02 2.71138999e-03 0.00000000e+00
0.00000000e+00 2.34151772e-01 0.00000000e+00 2.74629641e-01
3.82398108e-02 3.51593064e-03 0.00000000e+00 2.25062582e-03
7.34404022e-02 3.02164679e-02 1.10468968e-01 3.36602968e-01
1.43712489e-01 1.09583017e-02 9.97860841e-03 6.20179844e-03
3.92453395e-03 5.93950226e-01 2.42696614e-01 1.10854185e-02
5.30694827e-02 8.71279579e-02 1.71495793e-01 4.53369014e-01
4.08880943e-01 2.43222424e-01 2.31048774e-02 1.97243934e-01
2.28690841e-02 3.89067069e-01 5.65879939e-01 2.64268559e-01
1.02377787e-02 0.00000000e+00 2.24908740e-01 1.37897878e-01
0.00000000e+00 3.84890236e-02 1.36013255e-01 1.13238046e-01
4.42613337e-02 1.54047730e-01 7.26844315e-04 9.74848849e-01
2.01388742e-03 2.98355979e-01 5.06984528e-01 1.99649387e-01
2.77657386e-02 4.29460962e-01 1.02848076e-01 4.89428001e-02
8.00972222e-02 1.44641703e-01 5.47481542e-02 6.50547255e-01
8.27105379e-02 1.39682293e-01 1.53401434e-01 1.86305773e-03
1.10090329e-01 1.66675096e-01 6.19209248e-01 4.98140945e-01
2.94691587e-01 2.72221755e-01 0.00000000e+00 0.00000000e+00
3.26706417e-01 4.91620740e-02 1.97732120e-01 5.43678807e-01
3.22718844e-01 3.99852058e-03 1.28240520e-01 5.80458086e-02
6.15485883e-04 2.65672284e-01 1.44784379e-01 5.08912552e-02
1.02175200e-01 2.88217333e-01 4.58941581e-01 2.21592754e-04
2.43926919e-01 2.29463331e-01 2.23860692e-02 3.96450827e-01
1.50189446e-01 3.84811442e-02 1.43428384e-01 1.96921016e-02
4.77979398e-01 7.78812012e-02 5.56363349e-01 2.12033381e-01]
"""
if seed != 0:
......@@ -520,17 +517,14 @@ def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
else:
source = -1
if max_iter == None:
max_iter = 0
if reversed:
g.stash_filter(reversed=True)
g.set_reversed(True)
ic = libgraph_tool_centrality.\
libgraph_tool_centrality.\
get_absolute_trust(g._Graph__graph, source,
_prop("e", g, trust_map), _prop("v", g, vprop),
epslon, min_iter, max_iter, reversed, seed)
n_iter, reversed, seed)
if reversed:
g.pop_filter(reversed=True)
......@@ -538,8 +532,5 @@ def absolute_trust(g, trust_map, source=None, vprop=None, epslon=0.001,
vprop_temp.get_array()[:] = numpy.array(vprop[g.vertex(source)])
vprop = vprop_temp
if ret_iter:
return vprop, ic
else:
return vprop
return vprop
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment