You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
@@ -134,7 +134,7 @@ In case of non-distributed data streams, the [martingale estimator](src/main/jav
134
134
can be used, which gives slightly better estimation results as the asymptotic storage factor is $6\ln 2 = 4.159$.
135
135
This gives a relative standard error of $\sqrt{\frac{6\ln 2}{6m}} = \frac{0.833}{\sqrt{m}}$.
136
136
The theoretically predicted estimation errors have been empirically confirmed by [simulation results](doc/hyperloglog-estimation-error.md).
137
-
* UltraLogLog: This algorithm is described in detail in this [paper](https://arxiv.org/abs/2308.16862).
137
+
* UltraLogLog: This algorithm is described in detail in this [paper](https://doi.org/10.14778/3654621.3654632).
138
138
Like for HyperLogLog, a precision parameter $p$ defines the number of registers $m = 2^p$.
139
139
However, since UltraLogLog uses 8-bit registers to enable fast random accesses and updates of the registers,
140
140
$m$ is also the state size in bytes.
@@ -211,7 +211,7 @@ The following consistent hashing algorithms are available:
211
211
*[Improved Consistent Weighted Sampling](https://doi.org/10.1109/ICDM.2010.80): This algorithm is based on improved
212
212
consistent weighted sampling with a constant computation time independent of the number of buckets. This algorithm is faster than
213
213
JumpHash for a large number of buckets.
214
-
*[JumpBackHash](https://arxiv.org/abs/2403.18682): In contrast to JumpHash, which traverses "active indices" (see [here](https://doi.org/10.1109/ICDM.2010.80) for a definition)
214
+
*[JumpBackHash](https://doi.org/10.1002/spe.3385): In contrast to JumpHash, which traverses "active indices" (see [here](https://doi.org/10.1109/ICDM.2010.80) for a definition)
215
215
in ascending order, JumpBackHash does this in the opposite direction. In this way, floating-point operations can be completely avoided.
216
216
Further optimizations minimize the number of random values that need to be generated to reach
217
217
the largest "active index" within the given bucket range in amortized constant time. The largest "active index",
0 commit comments