search

stephenslab / susieR

R package for "sum of single effects" regression.
https://stephenslab.github.io/susieR
Other
176 stars 45 forks source link

Summary version check R for positive semi-definite takes too much time #65

Closed gaow closed 5 years ago

gaow commented 5 years ago

Currently we do eigen decomposition for the input R matrix and check for eigen values to ensure it is PSD. It is very time consuming. Should we check this at all? I think it is a good idea to check it. From a few data-sets I analyzed for others, their input R can be none PSD and susieR is able to capture that. So the question is how to check it more efficiently.

stephens999 commented 5 years ago

what is PSD?

do you mean Positive semi-definite? (this is a non-standard abbreviation) On Wed, Oct 24, 2018 at 5:34 PM gaow notifications@github.com wrote:

Currently we do eigen decomposition for the input R matrix and check for eigen values to ensure it is PSD. It is very time consuming. Should we check this at all? I think it is a good idea to check it. From a few data-sets I analyzed for others, their input R can be none PSD and susieR is able to capture that. So the question is how to check it more efficiently.

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub, or mute the thread.

gaow commented 5 years ago

@stephens999 Okay I've changed it. Maybe SPSD is a better abbreviation (at least I'm sure I've seen it in class)

pcarbo commented 5 years ago

The standard abbreviation is SPD = symmetric positive definite. I don't know of any other standard abbreviations.

pcarbo commented 5 years ago

@gaow In mashr we check that a matrix is SPD using chol. Would that work for your aims? I don't know if it is possible to distinguish (in finite-precision arithmetic) between positive definitive and positive semi-definite.