Use case class 5 -- User Story

Use case class 5:

A user story on improving algorithms for anomaly-based triggers (in search for dark matter)

Present Jocelyn and Subramanyan:

Jocelyn and Subramanyan are investigating different classes of algorithms to search for signatures of dark matter. They have originally started studying reprocessed data at a single astroparticle experiment like CRESST. In their analysis they have optimized expected sensitivities for several models describing properties of dark matter. Moreover, their algorithm improves the sensitivity of the experiment significantly. The algorithm would further  improve the performance of CRESST if applied already in the real-time analysis, by selecting more dark matter candidate events and less background. Jocelyn and Subramanyan discuss the usage of their algorithm with the CRESST collaboration, and it is individually adapted to the CRESST online software. Now, a competing collaboration becomes aware of Jocelyns and Subramanyans work. Since their software shares hardly any interfaces with the CRESST software, the whole work has to start from the beginning: evaluation in simulation, implementation in the software, writing analyses to cross-check the performance relative to orthogonal triggers, and so on.

Future Jocelyn and Subramanyan:

The implementation of actual solutions for real-time anomaly detection will be one focus of task area 5 within the  PUNCH consortium. Jocelyn and Subramanyan implement the definition of their algorithm using generic interfaces to identify anomalous signals in huge data streams. The anomaly validation including simulated data can then be addressed generically in the same framework. Then, the new anomaly-based trigger is available directly in several particle physics and astrophysics experiments which share the PUNCH interface. This way a cross-experiment and actually cross-community collaboration to search online also for signatures of dark matter will be established enhancing the potential for its discovery substantially.