How to estimate the fractal dimension of each data point？

I used the Em algorithm to perform point-dimensional estimation and estimated the following parameters.

The input is the distance between the data and the nearest point　（20000×1)．

The output is

d: A 1 x M vector of dimension parameters

delta: A 1 x L (L = sum( [ m_1, m_2, .., m_M ] )) vecter of density parameters

prior_prob: A 1 x L vecter of base probabilities of clusters

conditional_prob: K x L matrix of conditional probability of cluster assignment given each data.

LLfinal: The log likelihood of the best model.

MapToDim: A binary M x L matrix indicating which density clusters (column) is assigned to each dimensional cluster (each row).　

SettingsSearch: A struct variable containing meta information about the model search process.

How can I calculate the fractal dimension of all the data after finding the point dimension of each cluster?