Skip to main content

Table 1 Comparison of various distributed QR factorization algorithms

From: Distributed Gram-Schmidt orthogonalization with simultaneous elements refinement

 

Total number of

Total amount of

Local memory

 

sent messages

data (real numbers)

requirements

   

per node

DS-CGS

N·I (d)

\(N\cdot I^{(d)} \cdot \frac {m^{2}+5m}{2}\)

O(mn/N + m 2)

AC-CGS

\(N \cdot I^{(s)}\cdot \frac {(m+1)m}{2}\)

\(N \cdot I^{(s)} \cdot \frac {(m+1)m}{2}\)

O(mn/N + m 2)

G-CGS

N·R·(2m−1)

\(N \cdot R \cdot \frac {m^{2}+5m-2}{2}\)

O(nm/N)

G-MGS

N·R·(2m−1)

\(N \cdot R \cdot \frac {m^{2}+5m-2}{2}\)

O(nm/N)

  1. I (d) denotes the number of iterations of “dynamic” consensus, I (s) the number of iterations of “static” consensus, R the number of rounds per push-sum, N the number of nodes, m the number of columns of the input matrix