60 likes | 152 Views
Experts asked for information / comments. Following Tim’s request at the Coll. meeting 19 th June, SH contacted experts from both the US and Europe to find out about current capabilities of frameworks and envisaged development.
E N D
Experts asked for information / comments • Following Tim’s request at the Coll. meeting 19th June, SH contacted experts from both the • US and Europe to find out about current capabilities of frameworks and envisaged development. • from the US asked John Jaros, Norman Graf, Tim Barklow and Aurelio Juste • from Europe contacted Ties Behnke, Frank Gaede and on one specific issue Alexei Raspereza • Questions asked to gather information on: • Interoperability, in particular if SiD tracking is / will be available in European framework • Current level of usage of LCFI code in the community and projection of future usage • Best approach to make high-profile contributions to LoIs in the WP1 area • Replies from the European experts were reported in the Physics meeting of 26th June. • From the US, Aurelio Juste and Norman Graf sent replies.
Interoperability of software frameworks • replies confirm what we heard from Ties Behnke and Frank Gaede earlier: • Is interoperability planned to be ensured in both directions? Are you aware of • plans to make ALCPG/SiD tracking available / usable from within the European • software framework, and if so, on what timescale? • Norman Graf: If by interoperability you mean having MarlinReco calling org.lcsim Drivers, • then I am not aware of any interest in doing that. • However, there is an effort to provide a cross-framework geometry system, which would go • a long ways towards solving some of the problems. [Comment: stalled, as we heard last meeting] • My guess is that people would prefer to write a C++ implementation of an algorithm than to • use the Java code. • What is your best realistic estimate of the timescale on which the software frameworks will be • fully interoperable? • Norman Graf: I doubt they will ever be fully interoperable.
Views on contributions to LoIs • What is likely to have a higher profile within the LOIs: • Results from benchmark physics processes performed by our group? • Contributions to code development (extension of Vertex Package)? • Aurelio Juste: I believe the former. • (cf reply from Ties Behnke: The current thinking is that the LoI's will have a strong part on • realistic physics analyses. …Visibility will come from both aspects: analyses and code • development. It is hard to say whether one will be stronger than the other - probably not) • Consider using Eur. Framework to perform first LDC study and then switch to SiD detector and • ALCPG tracking. For the SiD group, would benchmark studies performed using the ALCPG • framework and SiD detector from the start be preferred to the suggested approach? • Aurelio Juste: I personally don't believe it matters what framework is used. The main reason to • prefer the studies to be performed using the ALCPG framework and SiD detector from the start • would be the fact that they would become available earlier, rather than having to wait for the LDC • studies to be completed.
LCFI code in US software framework • both Aurelio and Norman emphasised the importance of our code being made • available in the ALCPG software framework: • What is your estimate for the number of analyses that will be using the LCFI Vertex Package? • Aurelio Juste: Within SiD, I can tell you that right now there aren't too many vertex-related • benchmarking analyses that I know of going on or planned (maybe ~3-5), partly because of • lack of sophisticated and realistic enough tools. Obviously this will have to change in order for • the concepts to be able to deliver LOIs in the timescales we are talking about. • I personally believe the goal should be to make the LCFI Vertex Package be THE VERTEXING • TOOL within SiD, provided any technical issues can be overcome. This would ensure that • any vertex-related benchmarking study would be using the LCFI Vertex Package. • What approach would be more likely to result in an increase in this number? • Norman Graf: Clearly, the more flexible the LCFI package is, the greater the chances of it • being used in analyses.
LCFI code in US software framework contd • What is the timescale on which you plan to write & test the drivers for the LCFI Vertex Package? • Norman Graf: This is proceeding quite slowly at the moment due to the general lack of • manpower and to the summer vacation schedules. We would clearly like to have this done • by ALCPG07.
Conclusions • At present, there is great interest in our code in Europe and in the US. • In the short term, whether it will be used in the US will depend on whether it will become • available on the timescale needed for the LoIs – a delay could delay the completion of the • SiD LoI; given this, ALCPG07 seems rather late lack of manpower is evident. • In the medium term, the competitiveness of our code will depend on how well it will be • be maintained and extended (NOTE: currently realistic detector comparisons are not • possible due to hardwired geometry for “hadronic interactions workaround”). • A careful balance is needed between benchmark studies and code extensions and between • work in the US and in the European framework, in order to maximise our impact on both LoIs. • In deciding how to divide effort, both the short and the medium term should be kept in mind – • it would not make sense trying to maximise impact on LoIs by working primarily on benchmark • studies at the expense of neglecting code extensions and risking to lose the leading position • in the EDR phase.