d2lib is a C++ library of discrete distribution (d2) based data processing framework. It also contains a collection of computing tools supporting the analysis of d2 data.

[under construction]

• discrete distribution over Euclidean space
• discrete distribution with finite possible supports in Euclidean space (e.g., bag-of-word-vectors and sparsified histograms)
• n-gram data with cross-term distance
• dense histogram

• distributed/serial IO
• compute Wasserstein distance between a pair of D2.

• nearest neighbors [TBA]
• D2-clustering [TBA]
• Dirichlet process [TBA]

• document analysis: from bag-of-words to .d2s format [TBA]