Thomas Lumley 9/27/2016

Large quadratic forms

Read Original

The article discusses the statistical and computational challenges of analyzing large quadratic forms, particularly in genomics (e.g., SKAT tests). It explains how these forms arise from misspecified likelihood ratios or weighted test statistics, leading to distributions based on eigenvalues. The core problem is the high computational cost (O(n³)) for large matrices with thousands of variants, prompting a proposal to approximate using only the largest eigenvalues and the Satterthwaite method.

Large quadratic forms

Comments

No comments yet

Be the first to share your thoughts!

Browser Extension

Get instant access to AllDevBlogs from your browser