Truncated normal variance #276
Closed
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| Truncated{typeof(d),S}(d,l,u, cdf(d, u) - cdf(d, l)) | ||
| cdf_u = cdf(d, u) | ||
| cdf_l = cdf(d, l) | ||
| nc = if cdf_u > 0.98 && cdf_l > 0.98 |
There was a problem hiding this comment.
Should we do anything to calibrate these magic numbers? They seem sensible, but I wonder if we might want to test them.
There was a problem hiding this comment.
Not clear: they really tradeoff computation time against accuracy (because if the test passes then you've got to do more work). Some people will care more about speed, others will care more about accuracy.
|
Changes Unknown when pulling 65263ec on LaurenceA:truncated-moments into * on JuliaStats:master*. |
|
Yes. Seem right. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes Issue #264. There are three changes, all of which help us to compute the moments of a truncated Gaussian.
First, there was a bug in the implementation of the variance for Truncated Normal moments.
Second, the moment calculations often fail because the normalizing constant has a large error. This has been fixed by making sure that we use ccdf if cdf(d, u) and cdf(d, l) are close to 1.
Third, tests have been added, one of which fails if the second alteration is not made.