To do

project

• elaborate the basic procedures for general temporal quantities with general time intervals
• apply these indexes to real-life data (Franzosi, SN5)
• betweenness centrality based on semiring from About semirings; closeness centrality; can the Brandes approach be used? 2001,2008
• triad census - see Moody
• temporal largest eigen-value / eigen-vector !? power method;Lanczos; Arbenz
• measures from Siena (manual from page 37- and Statnet/ERGM ; TERGM, Tore; 1, 2, 3
• temporal weak/strong components (reachability semiring) paper; W = (A+A^T)^*; S = A^* ⋂ (A^*)^T
  • transforming temporal equivalence into a partition
    • multiply matrix with 1:n or random diagonal - equivalent nodes get the same value; if necessary repeat; intersections. Analysis of evolution.
    • can Union-find algorithm be used?
• temporal cores; paper
• temporal patterns PNAS
• accumulative values / differential values
• add symmetric attribute and representation of symmetric matrices with triangular matrices [ [a11], [a21,a22], [a31,a32,a33], …] and adapt the corresponding procedures.
• which procedures can be efficiently implemented for large networks?
• can the approach be extended to temporal networks with non-zero latency?
• clustering/blockmodeling of temporal networks

21.4.2014: Ustvari za vozlišča pare (št. v razbitju 1, št. v razbitju 2) in jih uredi v padajočem vrstnem redu glede na pogostost. Najpogostejši pari prenesejo številko (če je še prosta) iz razbitja 1 na razbitje 2.

• Vozlišča iz istega razreda imajo enake vrstične vsote v A.v . Za v je najbrž smiselno izbrati slučajno permutacijo 1:n , tako da se izognemo odvisnostim zaradi številčenja po sosedih.
• Za preverjanje pravilnosti razbitja zadostuje, da za vsako skupino izberemo eno vozlišče in preverimo ali je povezano z ostalimi iz razreda.