Skip to content
GitLab
Projects Groups Topics Snippets
  • /
  • Help
    • Help
    • Support
    • Community forum
    • Submit feedback
  • Sign in
  • S s2dverification
  • Project information
    • Project information
    • Activity
    • Labels
    • Members
  • Repository
    • Repository
    • Files
    • Commits
    • Branches
    • Tags
    • Contributor statistics
    • Graph
    • Compare revisions
  • Issues 65
    • Issues 65
    • List
    • Boards
    • Service Desk
    • Milestones
  • Merge requests 7
    • Merge requests 7
  • Deployments
    • Deployments
    • Releases
  • Monitor
    • Monitor
    • Incidents
  • Analytics
    • Analytics
    • Value stream
    • Repository
  • Wiki
    • Wiki
  • Snippets
    • Snippets
  • Activity
  • Graph
  • Create a new issue
  • Commits
  • Issue Boards
Collapse sidebar
  • Earth SciencesEarth Sciences
  • s2dverification
  • Issues
  • #212
Closed
Open
Issue created Apr 20, 2018 by Llorenç Lledó@llledoDeveloper

Bad normalization in EOF

Hello, I think (!) there is a bug in the EOF function.

# Factor to normalize the EOF.
eof.patt.nn <- EOF[e, , ] * mask
eof.patt.ms <- sum(eof.patt.nn^2, na.rm = TRUE)

# Normalize the EOF     
eof.patt <- eof.patt.nn/eof.patt.ms

But the norm of the eof pattern is indeed:

eof.patt.ms <- sqrt(sum(eof.patt.nn^2, na.rm = TRUE))

Luckily this never does nothing, as mask is typically all ones, and EOF[e,,] already has norm 1, because La.svd returns u and v matrices normalized already.

Edited Apr 20, 2018 by Llorenç Lledó
Assignee
Assign to
Time tracking