Fix bug in DP README.md #1236
Fix bug in DP README.md #1236
Conversation
fix math formatting
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #1236 +/- ##
=========================================
Coverage 92.82% 92.82%
=========================================
Files 200 202 +2
Lines 31166 32520 +1354
=========================================
+ Hits 28930 30188 +1258
- Misses 2236 2332 +96 ☔ View full report in Codecov by Sentry. |
ipa-core/src/protocol/dp/README.md
Outdated
| $$ num_bernoulli \geq 8 \log(2/\delta) /\varepsilon^2$$ | ||
| paper above is that for small epsilon (TODO, how small required?), we require the following number of samples: | ||
|
|
||
| $$ num\\_bernoulli \geq \frac{8 \log \left( \frac{2}{\delta}\right)}{\varepsilon^2}$$ |
There was a problem hiding this comment.
This formula is a high level summary of the paper's result. The complete calculation for determining num_bernoulli is more involved in what the code actually does. Was using this to get some early estimates but now that we have the full analysis implemented we should probably just point to that from the readme and drop this approximation.
There was a problem hiding this comment.
that works too. just noticed that it wasn't formatting correctly so I figured I'd fix it 🙂
There was a problem hiding this comment.
it would be useful to provide the instructions for computing the expectation and arriving at the unbiased estimate (even if it's not yet implemented.)
remove formula based
|
@bmcase I removed the formula. is this good to close? |
|
yep, looks good. thanks for the edits |
fix math formatting