Computing Atlas

How Computing Was Built
Sign In
Text size
100%
Theme
Algorithm

LU Decomposition Algorithm

Numerical Algorithm

LU decomposition factors a square matrix into the product of a lower triangular matrix L and an upper triangular matrix U, typically computed as a byproduct of running Gaussian elimination and recording the multipliers used at each elimination step into L while the resulting eliminated matrix becomes U. Once a matrix is factored this way, solving a linear system against it, or against several different right-hand sides, becomes much cheaper than repeating full Gaussian elimination each time, since forward and back substitution against triangular matrices are fast. The technique is named for the lower and upper triangular matrices it produces and is a standard building block of numerical linear algebra software.

Facts
Partially Attested
Time Complexity
2/3 n^3 floating-point operations 1
Source states 2/3 n^3 floating-point operations when computed by Gaussian elimination, ignoring lower-order terms
Credited To
Tadeusz Banachiewicz 1
Sources
1. Wikipedia: LU decomposition
  • Lead section
    introduced by the Polish astronomer Tadeusz Banachiewicz in 1938
  • Using Gaussian elimination
    floating-point operations, ignoring lower-order terms
View the Source
Comments (0)
No comments yet. Be the first to share a thought.
Reader Challenges (0)
No disputes yet. Spotted an error or a better source? Open the first one.