TY  - COMP
TI  - Streaming Statistics
AB  - In the context of a larger effort for in situ data analytics, there is a need to calculate basic statistics metrics (e.g., count, mean, median) online as new data points become available. Originally, the code for such online, or in other words streaming, statistics was part of the TALASS (Topological Analysis of Large- Scale Simulations) library. We isolated the relevant code and created a standalone library from it called Streaming Statistics. We also added an ability to serialize and deserialize the statistics objects so that the library can be used in distributed, task-based processing. To use the Streaming Statistics library, the user chooses a statistic, constructs an object for it, and then "adds" values to it, which means the statistic is augmented.
AU  - Shudler, Sergei
AU  - Bremer, Peer-Timo
DO  - https://doi.org/10.11578/dc.20220324.4
UR  - https://www.osti.gov/doecode/biblio/72100
CY  - United States
PY  - 2022
DA  - 2022-03-01
LA  - English
C1  - Research Org.: Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
C2  - Sponsor Org.: USDOE National Nuclear Security Administration (NNSA)
C4  - Contract Number: AC52-07NA27344
ER  -