[Halld-offline] reconstruction problem in master branch of sim-recon

[Shepherd, Matthew]

This may be a little thread creep but....

 Could Simon explain the functional differences in the new approach?  I was working to implement the neural net for flagging hadronic showers and split offs.  It uses the reference trajectory in the DTrackWireBased object.  Two things I want to know:

1) With the new system how do I retrieve the location of a track at the upstream face of the FCAL?

2) is the new system a change of interface or change of underlying algorithm?  I’d like to know if I need to potentially retrain the algorithm using the new underlying DTrackWireBased objects.

Thanks,

Matt

> On Dec 22, 2017, at 11:03 AM, Mark Ito <marki at jlab.org> wrote:
> 
> Folks,
> 
> As mentioned at the analysis meeting on Wednesday, there is an issue with the master branch of sim-recon that causes reduced efficiency for photon reconstruction in the FCAL:
> 
> https://halldweb.jlab.org/wiki-private/index.php/December_20,_2017,_Analysis_Working_Group#Minutes
> 
> Since then Alex has identified a major set of changes to the matching routines as the cause of the problem. These were introduced with pull request #1008:
> 
>   https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_sim-2Drecon_pull_1008&d=DwIF-g&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=oyKV5joTkJsuRYv6hh48IMTw3i-IrYD-ZUAHHU0DdAY&m=5eJdPttkwfiSH0bBVbgKZIVez1OpVuGJdPnaucdMiFg&s=xY1Wuy35AKEerwIrEYcaMxk-dGIl8naGJHH2Zm9mLkE&e= 
> 
> With the holiday break upon us, there is no plan to institute some sort of quick fix. The latest tagged release:
> 
>   https://urldefense.proofpoint.com/v2/url?u=https-3A__github.com_JeffersonLab_sim-2Drecon_releases_tag_2.20.1&d=DwIF-g&c=lz9TcOasaINaaC3U7FbMev2lsutwpI4--09aP8Lu18s&r=oyKV5joTkJsuRYv6hh48IMTw3i-IrYD-ZUAHHU0DdAY&m=5eJdPttkwfiSH0bBVbgKZIVez1OpVuGJdPnaucdMiFg&s=YiE6YcSHaCfil6ankqZjoAfgmiiWkEWr_9ZxgAX9ngc&e= 
> 
> pre-dates the change and can be used as a fall-back.
> 
> If folks really need a "fixed" version, let me know and we can discuss options.
> 
>   -- Mark
> 
> -- 
> Mark Ito, marki at jlab.org, (757)269-5295
> 
> _______________________________________________
> Halld-offline mailing list
> Halld-offline at jlab.org
> https://mailman.jlab.org/mailman/listinfo/halld-offline