Possible discrepancy between xml file and geometry converted into root file

[andread3vita]

Check duplicate issues.

• Checked for duplicates

Goal

Hi all!
I am working with genfit to write a gaudi functional that can fit the tracks from the IDEA reconstruction pipeline. In order to test its functionalities, I have also a local version of genfit.

Genfit has to be initialized with the geometry of the detector and I have tried two different approaches:

• in my local genfit I am initializing it with the root file of geometry. After creating the root file with dd4hep2root, I initialize genfit as follows:

new TGeoManager("Geometry", "IDEA geometry");
TGeoManager::Import("/eos/user/a/adevita/saveSpace/testingRoom/testROOTvsXML/TGeoIDEA.root");
genfit::MaterialEffects::getInstance()->init(new genfit::TGeoMaterialInterface());

• in the version where I am trying to include genfit within a gaudi functional, I retrieve the geometry from the GeoSvc that I have initialized in the steering file with the xml file. I can do that because I verified that GeoSvc is initializing a TGeoManager.

materialEffects = genfit::MaterialEffects::getInstance();
materialEffects->init(new genfit::TGeoMaterialInterface());

The xml file that I am using to initialize the GeoSvc and the one that I used to create the root file are the same.

In principle I should get the same results, but I get different outputs. Since the inputs to the genfit are the same in both my local version and the gaudi functional, I think that there is something different at the level of the initialization of the geometry.

I assume that the TGeoManager instances created in the previous two examples are the same, but maybe it is not true. Moreover I am using the same initialization for genfit.

Thank you in advance for all the suggestions.
Andrea

Operating System and Version

Alma Linux 9.5

compiler

GCC 14.2.0

ROOT Version

ROOT Version: 6.32.04

DD4hep Version

current

Reproducer

.

Additional context

No response

[andresailer]

Did you have a look at the different geometries? Or just at what genfit gives you?

[andread3vita]

I gave a look only at the output of genfit, but I am using exactly the same lines of code for the two examples. The only difference is the initialization of the geometry.

[MarkusFrankATcernch]

Try to checksum the geometry for both cases using DetectorChecksum.
If the checksums differ using the same compiler etc. you need to investigate further why.

[andread3vita]

Okay, I will try to do that.

[andread3vita]

Hi @MarkusFrankATcernch! Could you explain to me how to use the DetectorChecksum?
I tried the following:

std::string xml_file = "/eos/user/a/adevita/saveSpace/k4geo/FCCee/IDEA/compact/IDEA_o1_v03/IDEA_o1_v03.xml";
Detector& description = Detector::getInstance();
description.fromCompact(xml_file); 

DetElement top = description.world();
dd4hep::detail::DetectorChecksum checksum(description);

checksum.analyzeDetector(top);
auto hash_entry = checksum.handleHeader();

std::cout << "Checksum: " << hash_entry.hash << std::endl;

and I get:

DetectorChecksum       ++ ==> Computing checksum for tree: /world
DetectorChecksum       ++ Handled 0 materials.
DetectorChecksum       ++ Handled 161479 solids.
DetectorChecksum       ++ Handled 729644 volumes.
DetectorChecksum       ++ Handled 69 vis.attributes.
DetectorChecksum       ++ Handled 3 fields.
DetectorChecksum WARN  +++ No Detector header information availible from the geometry description.
Checksum: 0

Is it the expected output? Thank you for your help!

Andrea

[MarkusFrankATcernch]

@andread3vita This is not the expected output.
Already this is highly suspicious:

DetectorChecksum       ++ Handled 0 materials.

Please have a look e.g. in the example:
CLICSiD_check_checksum_EcalBarrel
You should get something like this here:

+++ 0    Combined hash code                      ec2a9dbb0aeccacc  (10375 sub-codes)

And yes. Your XML needs a header structure, which apparently is missing:

DetectorChecksum WARN  +++ No Detector header information availible from the geometry description.

[atolosadelgado]

Hi @MarkusFrankATcernch! Could you explain to me how to use the DetectorChecksum? I tried the following:

std::string xml_file = "/eos/user/a/adevita/saveSpace/k4geo/FCCee/IDEA/compact/IDEA_o1_v03/IDEA_o1_v03.xml";
Detector& description = Detector::getInstance();
description.fromCompact(xml_file); 

DetElement top = description.world();
dd4hep::detail::DetectorChecksum checksum(description);

checksum.analyzeDetector(top);
auto hash_entry = checksum.handleHeader();

std::cout << "Checksum: " << hash_entry.hash << std::endl;

and I get:

DetectorChecksum       ++ ==> Computing checksum for tree: /world
DetectorChecksum       ++ Handled 0 materials.
DetectorChecksum       ++ Handled 161479 solids.
DetectorChecksum       ++ Handled 729644 volumes.
DetectorChecksum       ++ Handled 69 vis.attributes.
DetectorChecksum       ++ Handled 3 fields.
DetectorChecksum WARN  +++ No Detector header information availible from the geometry description.
Checksum: 0

Is it the expected output? Thank you for your help!

Andrea

Hi Andrea,

I see k4geo in the path, are you using k4geo compact files as they are in the repository?

Best,
Alvaro

[andread3vita]

Hi @atolosadelgado! Yes, I am using them as they are in the repository.

Andrea

[andread3vita]

@MarkusFrankATcernch I will try to check the example! Thanks!

Andrea

[MarkusFrankATcernch]

If I take the existing plugin as an example, this should do the job:

  DetectorChecksum wr(description);
  DetElement de = description.world();
  DetectorChecksum::hashes_t hash_vec;
  DetectorChecksum::hash_t checksum = 0;

  wr.analyzeDetector(de);
  hash_vec.push_back(wr.handleHeader().hash);
  wr.checksumDetElement(0, de, hash_vec, true);
  checksum = detail::hash64(&hash_vec[0], hash_vec.size()*sizeof(DetectorChecksum::hash_t));
  printout(ALWAYS,"DetectorChecksum","+++ Checksum for %s 0x%016lx",
           de.path().c_str(), checksum);