Event data


This section is incomplete!

The various tracking components in ACTS can be assembled into a full reconstruction chain. Between these components, data needs to be exchanged in a well-defined way. This is achieved through the EDM (Event Data Model), which is a set of data types and interfaces representing the content of an event. Until very recently, ACTS has focused mainly on an internal EDM, which is really focused on efficient interchange between components inside the toolkit, sometimes at the cost of usability for clients. With the main reconstruction chain becomes more and more mature, however, the focus has shifted to a more client-oriented EDM encapsulating the outputs of tracking.


Fig. 22 Diagram showing the stages and data flow of a track reconstruction chain. The boxes show EDM objects that are passed between the stages.

Fig. 22 shows an overview of the EDM data types and how they form a data-flow sequence. Measurements coming from the experiment software are the main inputs of the chain. This data-type is abstracted in ACTS in a way that allows the details of these measurements to be fully experiment-specific. See Uncalibrated measurements / source links for details.

ACTS ships with a clusterization algorithm, which can turn segmented raw measurements into clusters, the second EDM object in the chain, representing particle intersections with the sensors. For the creation of track seeds, clusters need to be converted into three-dimensional space-points, combining information from multiple clusters where needed, e.g. for one-dimensional silicon strip sensors. The space-points and seeds are part of the EDM as well, since they are handed over to the track finding component. This component is responsible for the creation of completed tracks, which can then optionally be refitted with a precision track fitter.

Both the track finding and the precision track fit produce track objects, which are the primary output of the tracking chain. More on the track EDM can be found in High-level Track Event Data Model.