Cluster Struct Reference

Public Attributes
int	order
	Clusterizing order : See details in Config::order.
int	sK
	parameters of pCCE
int	max_trial
int	max_iter
bool	kmeans_pp
bool	iter_detail
char	method [MAX_CHARARRAY_LENGTH]
	Clustering algorithm : See details in Config::method.
int *	nk
	The number of clusters for each order "k".
bool	addsubclus
	Include all sub-clusters of the highest order : on | off.
int ***	clusinfo
	The clusters for each order "k".

Member Data Documentation

addsubclus

bool Cluster::addsubclus

Include all sub-clusters of the highest order : on | off.

When calcuting the coherence in electron spin bath, CCE with order >= 3 gives lots of diverged point in coherence fucntion. If turn on this option, it will include all sub-clusters of the highest order and give the more robust result. (default : off)

clusinfo

int*** Cluster::clusinfo

The clusters for each order "k".

clusters[i][j][k] i : i-th order j : (if j==0) The number of clusters at order i (if j!=0) j-th clusters at order i k : (if k==0) The number that you will have to multiply(+)/divide(-) in the coherence calcualtion (if k!=0) The index of the spin in the cluster

e.g. clusters[i=0][j=0][k=0] : The number how many you should calculate in 0-th order clusters[i=1][j=0][k=0] : The number of 1st order clusters clusters[i=1][j!=0][k=0] : The number how many you should calculate the coherence for 1st order and j-th cluster clusters[i=1][j!=0][k!=0] : The spin index in j-th cluster at 1st order

iter_detail

bool Cluster::iter_detail

If you want to print detail for parition information, True.

kmeans_pp

bool Cluster::kmeans_pp

If you want to use kmeans++ method, True. (Default: True)

max_iter

int Cluster::max_iter

It's maximum value for kmeans algorithm opitimization.

max_trial

int Cluster::max_trial

It's maximum value for kmeans algorithm opitimization.

nk

int* Cluster::nk

The number of clusters for each order "k".

nk is useful when you calculate CCE with order >= 3 At order >=3, Creating clusters is based on rdip, a ton of cluster will be formed (at least 1e6).
This value would consider nk # clusters for each order after sorting this clusters by coupling-strength. ( NOTE : this parameter is only available at "clusalgo = hash" )

Input file format : nk = k:n (0<=k<=order, 0<=n<=Maximum number of clusters) e.g. nk = { 1:1000, 2:100, 3:10, 4:5 } nk = { 2:100000, 3:100000 } nk = { 4:0 } (if n=0, it means we will include all clusters at order 4)

Array format: nk[order + 1] = {0,} (default, including all clusters for each order) nk[i=0] = The number of clusters at order 0 (e.i. nk[0] = 1 always) nk[i>i] = The number of clusters at order i

sK

int Cluster::sK

parameters of pCCE

pcce

Todo

Hyeonsu should be write this part in detail.

Partition CCE variable pCCE(order N, sK)

---

The documentation for this struct was generated from the following file:

• /home/huijin/git/CCEX/include/cluster.h