/**
* Returns the statistic corresponding to the given ego with respect to the
* given values of the behavior variable.
*/
double AltersCovariateAvAltEffect::egoStatistic(int ego, double * currentValues)
{
double statistic = 0;
const Network * pNetwork = this->pNetwork();
int neighborCount = 0;
for (IncidentTieIterator iter = pNetwork->outTies(ego);
iter.valid();
iter.next())
{
int j = iter.actor();
if (!this->missing(this->period(), j) &&
!this->missing(this->period() + 1, j) &&
!this->missingCovariate(j,this->period()))
{
statistic += currentValues[j] * this->covariateValue(j);
neighborCount++;
}
}
if ((neighborCount > 0) && (this->ldivide))
{
statistic *= currentValues[ego] / neighborCount;
}
return statistic;
}
/**
* Calculates the contribution of a tie flip to the given actor.
*/
double SameCovariateTransitiveTripletsEffect::calculateContribution(
int alter) const
{
// If we are introducing a tie from the ego i to the alter j, then each
// two-path from i to j with v_i = v_j contributes one unit;
// in addition, each in-star i -> h <- j with v_i = v_h
// also contributes one unit.
// This number is not stored in a table and is calculated from scratch.
int contribution1 = 0;
const Network * pNetwork = this->pNetwork();
if (this->inequalityCondition(this->value(alter) - this->value(this->ego())))
{
contribution1 = this->pTwoPathTable()->get(alter);
}
// The following probably can be done more efficiently
// using CommonNeighborIterator.
// Iterate over ego's outgoing ties
for (IncidentTieIterator iter = pNetwork->outTies(this->ego());
iter.valid();
iter.next())
{
// Get the receiver of the outgoing tie.
int h = iter.actor();
if (this->inequalityCondition(this->value(h) - this->value(this->ego())) &&
pNetwork->tieValue(alter, h) >= 1)
{
contribution1++ ;
}
}
return contribution1;
}
/**
* Returns the value of this function for the given alter. It is assumed
* that the function has been initialized before and pre-processed with
* respect to a certain ego.
*/
double DifferentCovariateOutStarFunction::value(int alter)
{
int statistic = 0;
if (!(this->lexcludeMissing && this->missing(alter)))
{
const Network * pNetwork = this->pNetwork();
// Iterate over incoming ties in network W
for (IncidentTieIterator iter =
pNetwork->inTies(this->ego());
iter.valid();
iter.next())
{
// Get the sender of the incoming tie.
int h = iter.actor();
// in-2-stars:
if (!(this->lexcludeMissing && this->missing(h)))
{
if ((fabs(this->CovariateNetworkAlterFunction::value(h)
- this->CovariateNetworkAlterFunction::value(this->ego()))
> EPSILON) &&
((lnotBothDifferent) || (fabs(this->CovariateNetworkAlterFunction::value(h)
- this->CovariateNetworkAlterFunction::value(alter))
> EPSILON)) &&
(pNetwork->tieValue(alter, h) >= 1))
{
statistic++ ;
}
}
}
}
return statistic;
}
/**
* Calculates the change in the statistic corresponding to this effect if
* the given actor would change his behavior by the given amount.
*/
double SimilarityEffect::calculateChangeContribution(int actor,
int difference)
{
double contribution = 0;
const Network * pNetwork = this->pNetwork();
if (pNetwork->outDegree(actor) > 0)
{
// The formula for the average similarity effect:
// s_i(x) = avg(sim(v_i, v_j) - centeringConstant) over all neighbors
// j of i.
// sim(v_i, v_j) = 1.0 - |v_i - v_j| / observedRange
// We need to calculate the change delta in s_i(x), if we changed
// v_i to v_i + d (d being the given amount of change in v_i).
// To this end, we can disregard the centering constant and
// compute the average change in similarity, namely,
// avg(sim(v_i + d, v_j) - sim(v_i, v_j)) =
// avg(1 - |v_i+d-v_j|/range - 1 + |v_i-v_j|/range) =
// avg(|v_i-v_j| - |v_i+d-v_j|) / range,
// the average being taken over all neighbors of i.
// The reasoning for avg. similarity x popularity alter effect is
// similar.
// This is what is calculated below.
int oldValue = this->value(actor);
int newValue = oldValue + difference;
int totalChange = 0;
for (IncidentTieIterator iter = pNetwork->outTies(actor);
iter.valid();
iter.next())
{
int j = iter.actor();
int alterValue = this->value(j);
int change =
std::abs(oldValue - alterValue) - std::abs(newValue - alterValue);
if (this->lalterPopularity)
{
change *= pNetwork->inDegree(j);
}
totalChange += change;
}
contribution = ((double) totalChange) / this->range();
if (this->laverage)
{
contribution /= pNetwork->outDegree(actor);
}
if (this->legoPopularity)
{
contribution *= pNetwork->inDegree(actor);
}
}
return contribution;
}
/**
* Returns the statistic corresponding to the given ego with respect to the
* given values of the behavior variable.
*/
double AltersCovariateAvSimEffect::egoStatistic(int ego,
double * currentValues) {
const Network * pNetwork = this->pNetwork();
double statistic = 0;
int neighborCount = 0;
for (IncidentTieIterator iter = pNetwork->outTies(ego); iter.valid();
iter.next()) {
int j = iter.actor();
if (!this->missing(this->period(), j)
&& !this->missing(this->period() + 1, j)
&& !this->missingCovariate(j, this->period())) {
double tieStatistic = this->similarity(currentValues[ego],
currentValues[j]);
statistic += tieStatistic * this->covariateValue(j);
neighborCount++;
}
}
if (neighborCount > 0) {
statistic /= neighborCount;
}
return statistic;
}
/**
* Returns the value of this function for the given alter. It is assumed
* that the function has been initialized before and pre-processed with
* respect to a certain ego.
*/
double SameCovariateMixedTwoPathFunction::value(int alter)
{
int statistic = 0;
if (!(this->lexcludeMissing && this->missing(alter)))
{
const Network * pFirstNetwork = this->pFirstNetwork();
const Network * pSecondNetwork = this->pSecondNetwork();
for (IncidentTieIterator iter = pSecondNetwork->outTies(this->ego());
iter.valid();
iter.next())
{
// Get the receiver of the outgoing tie.
int h = iter.actor();
// 2-paths:
if (!(this->lexcludeMissing && this->missing(h)))
{
if ((fabs(this->CovariateMixedNetworkAlterFunction::value(h) -
this->CovariateMixedNetworkAlterFunction::value(this->ego()))
< EPSILON) &&
(pFirstNetwork->tieValue(h, alter) >= 1))
{
statistic++ ;
}
}
}
}
return statistic;
}
/**
* For each j and the given i, this method calculates the sum
* sum_h w_{ih} x_{hj}.
*/
void WXXClosureEffect::calculateSums(int i,
const Network * pNetwork,
double * sums) const
{
int n = pNetwork->n();
// Initialize
for (int j = 0; j < n; j++)
{
sums[j] = 0;
}
// Iterate over all h with non-zero non-missing w_{ih}
for (DyadicCovariateValueIterator iterH = this->rowValues(i);
iterH.valid();
iterH.next())
{
int h = iterH.actor();
// Iterate over all j with a tie from h
for (IncidentTieIterator iterJ = pNetwork->outTies(h);
iterJ.valid();
iterJ.next())
{
int j = iterJ.actor();
// Add the term w_{ih} x_{hj} (= w_{ih})
sums[j] += iterH.value();
}
}
}
/**
* Returns the statistic corresponding to the given ego as part of
* the endowment function with respect to the initial values of a
* behavior variable and the current values.
*/
double AltersCovariateAvSimEffect::egoEndowmentStatistic(int ego,
const int * difference, double * currentValues) {
double statistic = 0;
const Network * pNetwork = this->pNetwork();
if (difference[ego] > 0 && !this->missingDummy(ego)
&& (pNetwork->outDegree(ego) > 0)) // otherwise, nothing to calculate...
{
int oldValue = this->value(ego); // ego's behavior value before moving on behavior scale
int newValue = oldValue + difference[ego]; // ego's behavior value after moving on behavior scale
double totalChange = 0; // will keep track of changes
for (IncidentTieIterator iter = pNetwork->outTies(ego); // loops over outgoing ties of ego
iter.valid(); iter.next()) {
int j = iter.actor(); // identifies alter
int alterValue = this->value(j); // identifies behavior value of alter
int change = std::abs(oldValue - alterValue)
- std::abs(newValue - alterValue);
// calculates impact of ego's movement on absolute difference to alter
totalChange += change * this->covariateValue(j); // weighting change statistic acc. to covariate value of alter
}
statistic -= totalChange / this->range(); // standardize by behavior range for SIMILARITY function
statistic /= pNetwork->outDegree(ego); // divide by outdegree to get AVERAGE similarity change statistic
}
return statistic;
}
/**
* Calculates the change in the statistic corresponding to this effect if
* the given actor would change his behavior by the given amount.
*/
double AltersCovariateAvSimEffect::calculateChangeContribution(int actor,
int difference) {
double contribution = 0;
const Network * pNetwork = this->pNetwork();
if (pNetwork->outDegree(actor) > 0) // otherwise, nothing to calculate...
{
int oldValue = this->value(actor); // ego's behavior value before moving on behavior scale
int newValue = oldValue + difference; // ego's behavior value after moving on behavior scale
double totalChange = 0; // will keep track of changes
for (IncidentTieIterator iter = pNetwork->outTies(actor); // loops over outgoing ties of ego
iter.valid(); iter.next()) {
int j = iter.actor(); // identifies alter
int alterValue = this->value(j); // identifies behavior value of alter
int change = std::abs(oldValue - alterValue)
- std::abs(newValue - alterValue);
// calculates impact of ego's movement on absolute difference to alter
totalChange += change * this->covariateValue(j); // weighting change statistic acc. to covariate value of alter
}
contribution = totalChange / this->range(); // standardize by behavior range for SIMILARITY function
contribution /= pNetwork->outDegree(actor); // divide by outdegree to get AVERAGE similarity change statistic
}
return contribution;
}
/**
* Returns the statistic corresponding to the given ego with respect to the
* given values of the behavior variable.
*/
double AverageSimmelianAlterEffect::egoStatistic(int i, double * currentValues)
{
double statistic = 0;
clearSimmelian();
updateSimmelian((const OneModeNetwork*) this->pNetwork()); // simulated state
const Network * pNetwork = this->pNetwork();
int simmelianOutdegree = 0;
for (IncidentTieIterator iter = pNetwork->outTies(i);
iter.valid();
iter.next())
{
if (pSimmelian->tieValue(i, iter.actor()) > 0)
{
statistic += currentValues[iter.actor()];
simmelianOutdegree ++;
}
}
if (simmelianOutdegree > 0)
{
statistic *= currentValues[i];
if (this->ldivide)
{
statistic /= simmelianOutdegree;
}
}
return statistic;
}
/**
* Calculates the change in the statistic corresponding to this effect if
* the given actor would change his behavior by the given amount.
*/
double AltersCovariateAvAltEffect::calculateChangeContribution(int actor,
int difference)
{
double contribution = 0;
const Network * pNetwork = this->pNetwork();
if (pNetwork->outDegree(actor) > 0)
{
double totalAlterValue = 0;
for (IncidentTieIterator iter = pNetwork->outTies(actor);
iter.valid();
iter.next())
{
int j = iter.actor(); // identifies alter
double alterValue = this->centeredValue(j) * this->covariateValue(j);
totalAlterValue += alterValue;
}
if (this->ldivide)
{
contribution = difference * totalAlterValue /
pNetwork->outDegree(actor);
}
else
{
contribution = difference * totalAlterValue;
}
}
return contribution;
}
/**
* Returns the statistic corresponding to the given ego as part of
* the endowment function with respect to the initial values of a
* behavior variable and the current values.
*/
double AltersCovariateAvAltEffect::egoEndowmentStatistic(int ego,
const int * difference,
double * currentValues)
{
double statistic = 0;
const Network * pNetwork = this->pNetwork();
if (difference[ego] > 0 && !this->missingDummy(ego) && (pNetwork->outDegree(ego) > 0)) // otherwise, nothing to calculate...
{
double totalAlterValue = 0;
for (IncidentTieIterator iter = pNetwork->outTies(ego);
iter.valid();
iter.next())
{
int j = iter.actor(); // identifies alter
double alterValue = this->centeredValue(j) * this->covariateValue(j);
totalAlterValue += alterValue;
}
if (this->ldivide)
{
statistic -= difference[ego] * totalAlterValue /
pNetwork->outDegree(ego);
}
else
{
statistic -= difference[ego] * totalAlterValue;
}
}
return statistic;
}
/**
* Modifies the two-path count given the case the observed network is a
* two mode network.
* @param[in] rNetwork The observed network
* @param[in] ego The ego of the modified tie
* @param[in] alter The alter of the modified tie
* @param[in[ val The magnitude of modification
*/
void DistanceTwoLayer::modify2PathCountTwoMode(const Network& rNetwork, int ego,
int alter, int val) {
// in a two mode network the exist no triangles, therefore it is
// sufficient to iterate over all incoming ties of alter
for (IncidentTieIterator iter = rNetwork.inTies(alter); iter.valid();
iter.next()) {
if (iter.actor() != ego) {
modifyTieValue(ego, iter.actor(), val);
}
}
}
/**
* Calculates the statistic corresponding to the given ego. The variable
* pSummationTieNetwork is the current network in the case of an evaluation
* effect and the network of lost ties in the case of an endowment effect.
*/
double NetworkEffect::egoStatistic(int ego,
const Network * pSummationTieNetwork)
{
double statistic = 0;
for (IncidentTieIterator iter = pSummationTieNetwork->outTies(ego);
iter.valid();
iter.next())
{
statistic += this->tieStatistic(iter.actor());
}
return statistic;
}
/**
* Create primary setting based on a network
*/
std::vector<int> * primarySetting(const Network * pNetwork, int ego)
{
std::vector<int> *setting = new std::vector<int>;
std::set<int> neighbors;
for (IncidentTieIterator iter = pNetwork->outTies(ego);
iter.valid();
iter.next())
{
neighbors.insert(iter.actor());
}
for (IncidentTieIterator iter = pNetwork->inTies(ego);
iter.valid();
iter.next())
{
neighbors.insert(iter.actor());
}
neighbors.insert(ego);
// when finished (not all done here) copy to the vector
for (std::set<int>::const_iterator iter1 = neighbors.begin();
iter1 != neighbors.end(); iter1++)
{
setting->push_back(*iter1);
}
return setting;
}
/**
* This method marks as invalid all actors that are iterated over by the given
* iterator. The fact that an actor i is invalid is represented by setting
* lflag[i] = lround. It is assumed that lflag[i] <= lround for all actors,
* meaning that lflag[i] < lround holds for valid actors. The variable
* validActorCount keeps track of the still valid actors.
*/
void InStructuralEquivalenceEffect::markInvalidActors(IncidentTieIterator iter,
int & validActorCount)
{
while (iter.valid())
{
if (this->lflag[iter.actor()] < this->lround)
{
this->lflag[iter.actor()] = this->lround;
validActorCount--;
}
iter.next();
}
}
/**
* Calculates the contribution of a tie flip to the given actor.
*/
double CovariateIndirectTiesEffect::calculateContribution(int alter)
const
{
double change = 0;
// If there are enough two-paths from the ego i to the alter j, then
// we loose the distance 2 pair (i,j) by introducing the tie between
// them.
if (this->pTwoPathTable()->get(alter) != 0)
{
change -= this->value(alter);
}
// This variable is to simplify the later tests if a two-path through
// the given alter makes a difference.
int criticalTwoPathCount = 0;
if (this->outTieExists(alter))
{
criticalTwoPathCount = 1;
}
// Consider each outgoing tie of the alter j.
for (IncidentTieIterator iter = this->pNetwork()->outTies(alter);
iter.valid();
iter.next())
{
int h = iter.actor();
// If h is not the ego i, there's no tie from i to h, and the
// introduction or withdrawal of the tie (i,j) makes a difference
// for the pair <i,h> to be a valid distance two pair,
// then increment the contribution.
if (h != this->ego() &&
!this->outTieExists(h) &&
this->pTwoPathTable()->get(h) == criticalTwoPathCount)
{
change += this->value(h);
}
}
return change;
}
/**
* Initializes the layer given the reference network is a two mode
* network.
*/
void DistanceTwoLayer::initializeTwoMode(const Network& rNetwork) {
// this is a two mode network so we do not need to check for loops
// nor do we have to store the reciever two paths.
for (int i = 0; i < rNetwork.m(); ++i) {
// construct all pairs
for (IncidentTieIterator outerIter = rNetwork.inTies(i);
outerIter.valid(); outerIter.next()) {
int outerActor = outerIter.actor();
// copy the iterator
IncidentTieIterator innerIter(outerIter);
// move to the next position
innerIter.next();
for (; innerIter.valid(); innerIter.next()) {
modifyTieValue(outerActor, innerIter.actor(), 1);
}
}
}
}
/**
* Returns the average in-degree of the neighbors of the given actor in
* the current network (0, if the actor has no outgoing ties).
*/
double PopularityAlterEffect::averageInDegree(int i) const
{
const Network * pNetwork = this->pNetwork();
double inDegree = 0;
if (pNetwork->outDegree(i) > 0)
{
for (IncidentTieIterator iter = pNetwork->outTies(i);
iter.valid();
iter.next())
{
inDegree += pNetwork->inDegree(iter.actor());
}
inDegree /= pNetwork->outDegree(i);
}
return inDegree;
}
/**
* Returns the statistic corresponding to the given ego with respect to the
* given values of the behavior variable.
*/
double SimilarityEffect::egoStatistic(int ego,
double * currentValues)
{
const Network * pNetwork = this->pNetwork();
double statistic = 0;
int neighborCount = 0;
for (IncidentTieIterator iter = pNetwork->outTies(ego);
iter.valid();
iter.next())
{
int j = iter.actor();
if (!this->missing(this->period(), j) &&
!this->missing(this->period() + 1, j))
{
double tieStatistic =
this->similarity(currentValues[ego], currentValues[j]);
if (this->lalterPopularity)
{
tieStatistic *= pNetwork->inDegree(j);
}
statistic += tieStatistic;
neighborCount++;
}
}
if (this->laverage && neighborCount > 0)
{
statistic /= neighborCount;
}
if (this->legoPopularity)
{
statistic *= pNetwork->inDegree(ego);
}
return statistic;
}
/**
* Returns the statistic corresponding to the given ego with respect to the
* currentValues given for the behavior variable.
*/
double AverageAlterDist2Effect::egoStatistic(int i, double * currentValues)
{
double statistic = 0;
const Network * pNetwork = this->pNetwork();
int neighborCount = 0;
for (IncidentTieIterator iter = pNetwork->outTies(i);
iter.valid();
iter.next())
{
int j = iter.actor();
double alterValue = 0;
int tieToi = 0;
for (IncidentTieIterator iteri = pNetwork->outTies(j);
iteri.valid();
iteri.next())
{
if (i != iteri.actor())
{
alterValue += currentValues[iteri.actor()];
}
else
{
tieToi = 1;
}
}
// tieToi = this->pNetwork()->tieValue(iter.actor(), i);
if ((pNetwork->outDegree(j) > tieToi) & (this->ldivide2))
{
alterValue /= (pNetwork->outDegree(j) - tieToi);
}
statistic += alterValue;
neighborCount++;
}
if (neighborCount > 0)
{
statistic *= currentValues[i];
if (this->ldivide1)
{
statistic /= neighborCount;
}
}
return statistic;
}
/**
* Calculates the contribution of a tie flip to the given actor.
*/
double SameCovariateActivityEffect::calculateContribution(int alter) const
{
double myvalue = this->value(this->ego());
double contribution = 0;
const Network * pNetwork = this->pNetwork();
if ((lsame) && (fabs(this->value(alter) - myvalue) < EPSILON))
{
for (IncidentTieIterator iter = pNetwork->outTies(this->ego());
iter.valid();
iter.next())
{
// Get the receiver of the outgoing tie.
int h = iter.actor();
if (fabs(this->value(h) - myvalue) < EPSILON)
{
contribution++;
}
}
if (this->outTieExists(alter))
{
contribution--;
}
contribution *= 2;
contribution++;
}
if ((!lsame) && (fabs(this->value(alter) - myvalue) >= EPSILON))
{
for (IncidentTieIterator iter = pNetwork->outTies(this->ego());
iter.valid();
iter.next())
{
// Get the receiver of the outgoing tie.
int h = iter.actor();
if (fabs(this->value(h) - myvalue) >= EPSILON)
{
contribution++;
}
}
if (this->outTieExists(alter))
{
contribution--;
}
contribution *= 2;
contribution++;
}
return contribution;
}
/**
* The contribution of the tie from the implicit ego to the given alter
* to the statistic. It is assumed that preprocessEgo(ego) has been
* called before.
*/
double SameCovariateActivityEffect::tieStatistic(int alter)
{
double contribution = 0;
const Network * pNetwork = this->pNetwork();
if (!((this->missing(alter)) || (this->missing(this->ego()))))
{
double myvalue = this->value(this->ego());
if (lsame)
{
if (fabs(this->value(alter) - myvalue) < EPSILON)
{
for (IncidentTieIterator iter = pNetwork->outTies(this->ego());
iter.valid();
iter.next())
{
// Get the receiver of the outgoing tie.
int h = iter.actor();
if ((!this->missing(h)) &&
(fabs(this->value(h) - myvalue) < EPSILON))
{
contribution++;
}
}
}
}
else
{
if (fabs(this->value(alter) - myvalue) >= EPSILON)
{
for (IncidentTieIterator iter = pNetwork->outTies(this->ego());
iter.valid();
iter.next())
{
// Get the receiver of the outgoing tie.
int h = iter.actor();
if ((!this->missing(h)) &&
(fabs(this->value(h) - myvalue) >= EPSILON))
{
contribution++;
}
}
}
}
}
return contribution;
}
/**
* Calculates the change in the statistic corresponding to this effect if
* the given actor would change his behavior by the given amount.
*/
double AverageSimmelianAlterEffect::calculateChangeContribution(int actor,
int difference)
{
double contribution = 0;
int simmelianOutdegree = 0;
clearSimmelian();
updateSimmelian((const OneModeNetwork*) this->pNetwork()); // simulated state
const Network * pNetwork = this->pNetwork();
if (pNetwork->outDegree(actor) > 0)
{
// The formula for the effect:
// s_i(x) = v_i * avg(v_j) over all simmelian neighbors j of i.
// We need to calculate the change delta in s_i(x), if we changed
// v_i to v_i + d (d being the given amount of change in v_i).
// This is d * avg(v_j), the average being taken over all neighbors
// of i. This is what is calculated below.
// if (not divide), instead of avg the total is used.
for (IncidentTieIterator iter = pNetwork->outTies(actor);
iter.valid();
iter.next())
{
if (pSimmelian->tieValue(actor, iter.actor()) > 0)
{
contribution += this->centeredValue(actor);
simmelianOutdegree ++;
}
}
contribution *= difference;
if ((this->ldivide) && (simmelianOutdegree >= 1))
{
contribution /= simmelianOutdegree;
}
}
return contribution;
}
/**
* Does the necessary preprocessing work for calculating the tie flip
* contributions for a specific ego. This method must be invoked before
* calling NetworkEffect::calculateTieFlipContribution(...).
*/
void TwoNetworkDependentBehaviorEffect::preprocessEgo(int ego)
{
// set up the covariate based on current values of the network and behavior
const Network * pFirstNetwork = this->pFirstNetwork();
for (int i = 0; i < pFirstNetwork->n(); i++)
{
this->lfirstTotalAlterValues[i] = 0;
if (pFirstNetwork->outDegree(i) > 0)
{
for (IncidentTieIterator iter = pFirstNetwork->outTies(i);
iter.valid();
iter.next())
{
int j = iter.actor();
this->lfirstTotalAlterValues[i] += this->centeredValue(j);
// Rprintf("%d %f %d %d %d %d\n",
// j,
// this->centeredValue(j),
// this->period(),
}
}
else
{
this->lfirstTotalAlterValues[i] = 0;
}
// Rprintf("%d %f\n", i,this->ltotalAlterValues[i]);
}
for (int i = 0; i < pFirstNetwork->m(); i++)
{
this->lfirstTotalInAlterValues[i] = 0;
if (pFirstNetwork->inDegree(i) > 0)
{
for (IncidentTieIterator iter = pFirstNetwork->inTies(i);
iter.valid();
iter.next())
{
int j = iter.actor();
this->lfirstTotalInAlterValues[i] += this->centeredValue(j);
}
}
else
{
this->lfirstTotalInAlterValues[i] = 0;
}
}
}
/**
* Returns the statistic corresponding to the given ego with respect to the
* currentValues given for the behavior variable.
*/
double AverageAlterInDist2Effect::egoStatistic(int i, double * currentValues)
{
double statistic = 0;
const Network * pNetwork = this->pNetwork();
int neighborCount = 0;
for (IncidentTieIterator iter = pNetwork->outTies(i);
iter.valid();
iter.next())
{
int j = iter.actor();
double alterValue = 0;
for (IncidentTieIterator iteri = pNetwork->inTies(j);
iteri.valid();
iteri.next())
{
if (i != iteri.actor())
{
alterValue += currentValues[iteri.actor()];
}
}
// tieFromi = this->pNetwork()->tieValue(i, iter.actor());
if ((pNetwork->inDegree(j) > 1) & (this->ldivide2))
{
alterValue /= (pNetwork->inDegree(j) - 1);
// there always is a tie i -> iteri.actor()
}
statistic += alterValue;
neighborCount++;
}
if (neighborCount > 0)
{
statistic *= currentValues[i];
if (this->ldivide1)
{
statistic /= neighborCount;
}
}
return statistic;
}
/**
* Indicates if all ties are reciprocated with the same value.
*/
bool OneModeNetwork::symmetric() const {
// The current implementation is linear in the total number of ties.
// The time complexity can be reduced to a constant by maintaining
// the number of non-symmetric ties.
// Assume the network is symmetric until we can disprove it.
bool rc = true;
// Test the incoming and outgoing ties of each actor in turn.
for (int i = 0; i < this->n() && rc; i++) {
if (this->outDegree(i) == this->inDegree(i)) {
IncidentTieIterator outIter = this->outTies(i);
IncidentTieIterator inIter = this->inTies(i);
// No need to test both iterators for validity, as the numbers
// of incoming and outgoing ties are the same.
while (outIter.valid() && rc) {
if (outIter.actor() != inIter.actor()
|| outIter.value() != inIter.value()) {
// Found a mismatch.
rc = false;
}
outIter.next();
inIter.next();
}
} else {
// The numbers of incoming and outgoing ties differ, which
// destroys the symmetry immediately.
rc = false;
}
}
return rc;
}
/**
* Returns the statistic corresponding to the given ego as part of
* the endowment function with respect to the initial values of a
* behavior variable and the current values.
*/
double AverageSimmelianAlterEffect::egoEndowmentStatistic(int ego,
const int * difference,
double * currentValues)
{
double statistic = 0;
int simmelianOutdegree = 0;
clearSimmelian();
updateSimmelian((const OneModeNetwork*) this->pNetwork()); // simulated state
const Network * pNetwork = this->pNetwork();
if (difference[ego] > 0)
{
if (pNetwork->outDegree(ego) > 0)
{
double thisStatistic = 0;
double previousStatistic = 0;
for (IncidentTieIterator iter = pNetwork->outTies(ego);
iter.valid();
iter.next())
{
double alterValue = currentValues[iter.actor()];
double alterPreviousValue = currentValues[iter.actor()]
+ difference[iter.actor()];
if (pSimmelian->tieValue(ego, iter.actor()) > 0)
{
simmelianOutdegree ++;
thisStatistic += alterValue;
previousStatistic += alterPreviousValue;
}
}
if (simmelianOutdegree >= 1)
{
thisStatistic *= currentValues[ego];
previousStatistic *= (currentValues[ego] + difference[ego]);
statistic = thisStatistic - previousStatistic;
if (this->ldivide)
{
statistic /= simmelianOutdegree;
}
}
}
}
return statistic;
}
/**
* Calculates the change in the statistic corresponding to this effect if
* the given actor would change his behavior by the given amount.
* It is assumed that preprocessEgo(ego) has been called before.
*/
double AverageAlterDist2Effect::calculateChangeContribution(int actor,
int difference)
{
double contribution = 0;
const Network * pNetwork = this->pNetwork();
if (pNetwork->outDegree(actor) > 0)
{
// The formula for the effect:
// s_i(x) = v_i * avg(v_j) over all neighbors j of i,
// where v_j is the average behavior of j's neighbors,
// excluding i.
// We need to calculate the change delta in s_i(x), if we changed
// v_i to v_i + d (d being the given amount of change in v_i).
// This is d * avg(v_j) and is calculated below.
// if (not divide1) or (not divide2),
// instead of "avg" or "average" the total is used.
double sumAlterValue = 0;
for (IncidentTieIterator iter = pNetwork->outTies(actor);
iter.valid();
iter.next())
{
double alterValue = this->totalAlterValue(iter.actor());
int tieValue = this->pNetwork()->tieValue(iter.actor(), actor);
if (tieValue == 1)
{
alterValue -= this->centeredValue(actor);
}
if (((pNetwork->outDegree(iter.actor()) - tieValue)> 0) & (this->ldivide2))
{
alterValue /= (pNetwork->outDegree(iter.actor()) - tieValue);
}
sumAlterValue += alterValue;
}
contribution = difference * sumAlterValue;
if (this->ldivide1)
{
contribution /= pNetwork->outDegree(actor);
}
}
return contribution;
}
/**
* This method tests if there are at most <i>k</i> two-paths from <i>i</i>
* to <i>j</i>.
* If the test is positive, the number of such paths is stored
* in the variable <i>twoPathCount</i>.
*/
bool OneModeNetwork::atMostKTwoPaths(int i, int j, int k,
int & twoPathCount) const {
this->checkSenderRange(i);
this->checkReceiverRange(j);
// Iterate the outgoing ties of i and incoming ties of j simultaneously
// and count the number of matching neighbors. Stop as soon as their number
// exceeds k.
IncidentTieIterator outIter = this->outTies(i);
IncidentTieIterator inIter = this->inTies(j);
twoPathCount = 0;
while (outIter.valid() && inIter.valid() && twoPathCount <= k) {
if (outIter.actor() < inIter.actor()) {
outIter.next();
} else if (outIter.actor() > inIter.actor()) {
inIter.next();
} else {
twoPathCount++;
outIter.next();
inIter.next();
}
}
return twoPathCount <= k;
}
