Antonio Giannini authored 1 month ago and Adam Edward Barton committed 1 month ago

changes for qg tagger (no pre-processing model)

Zach Marshall authored 1 month ago and Adam Edward Barton committed 1 month ago

Modifying infrastructure so that HEPMC file is a tgz

Need to update the transform to zip things (and say it's doing so,
because that takes a long time in a slightly weird place with no obvious
output) and then to adjust the event counting to do the correct thing.

This should be what's expected based on ATLMCPROD-11650

For event generation, there is often no input file, and we should still
allow AthenaMP to run.

Jean-Baptiste De Vivie De Regie authored 1 month ago and Duc Ta committed 1 month ago

update condition tags for MC21 Run4 and MC23

Nils Erik Krumnack authored 1 month ago and Tadej Novak committed 1 month ago

decorate electrons and photons in PHYSLITE with Emva

This is needed to give the correct input energy to the columnar
EgammaCalibrationAndSmearingTool.  It should also allow to run solely
the smearing part of the tool in regular event-wise jobs.

In my test this was 44 bytes/event for PHYSLITE.  So that's notable, but
it is also part of the long-term plans for PHYSLITE.  Long-term we will
be able to drop the variables that are only needed for the calibration,
as we will be able to skip calibation on PHYSLITE.

Nils Erik Krumnack authored 1 month ago and Tadej Novak committed 1 month ago

add samplingPattern decoration to e-gamma clusters in PHYSLITE

For historic reason the sampling pattern is stored in the cluster object
instead of as a decoration.  To make this available for columnar
analysis it also needs to exist as a decoration.  For now that means we
have two copies of it, but it was already agreed to drop it from the
cluster object, so this is a temporary situation.

Michael Revering authored 1 month ago and Jean-Baptiste De Vivie De Regie committed 1 month ago

Merge branch 'modernizeMSVtxAlg' into 'main'

Modernize the MS vertex finding algorithm

See merge request !74525

Scott Snyder authored 1 month ago and Johannes Elmsheuser committed 1 month ago

CxxUtils: Don't try to use the x87 fsincos instruction any more.

Since a long time, we've had CxxUtils::sincos, which calculates sin and cos
using the x87 fsincos instruction.  At the time, this was significantly
better than the alternatives.  However, with current compilers on x86_64,
it is in fact faster to use the sincos() library function than the x87 instruction.
So remove the inlining bit.
sincos() is still a GNU extension, so don't remove this entirely from CxxUtils ---
this allow us to keep the #ifdefs in one place.

Grigori Rybkin authored 1 month ago and Johannes Elmsheuser committed 1 month ago

PyJobTransforms: Make MT file validation default

Tools/PyJobTransforms/python/transform.py: Make MT file validation default

Alaettin Serhan Mete authored 2 months ago and Frank Winklmeier committed 2 months ago

PyJobTransforms: Rework creating REntry in python

Alaettin Serhan Mete authored 2 months ago and Frank Winklmeier committed 2 months ago

RNTupleART: Add a new test for reading/writing ## containers in RNTuple

Daiya Akiyama authored 2 months ago and Duc Ta committed 2 months ago

Fixing the bug on the SCT bad error judgement

Antonio De Maria authored 2 months ago and Frank Winklmeier committed 2 months ago

ATR-30599: adding missing elec+tau for 2023/mc23d for Run3 matching

Matous Vozak authored 2 months ago and Duc Ta committed 2 months ago

Apply 1 suggestion(s) to 1 file(s)

Co-authored-by: Johannes Junggeburth <johannes.josef.junggeburth@cern.ch>

Savannah Clawson authored 2 months ago and Frank Winklmeier committed 2 months ago

Adding AFP SiT and ToF hits containers to PHYS

Antonio De Maria authored 2 months ago and Vakhtang Tsulaia committed 2 months ago

ATR-30599: add lep+tau T&P triggers to the list of run3 triggers to get matching

John Derek Chapman authored 2 months ago and Duc Ta committed 2 months ago

PyJobTransforms.trfAMI: improve handling of pyAMI.exception.Error exceptions

Follows from !76577. In the case that `getAMIClient` or `get_ami_tag` throws an
exception and the two specific cases are not found a general TransformAMIException
should be thrown rather than letting the job fail with an `UnboundLocalError` when
it tries to access a non-existent variable. Also remove obsolete debug message.

Scott Snyder authored 2 months ago and Duc Ta committed 2 months ago

GdbUtils: New package.

gdb helper code, previously maintained in a separate tree.

Thomas Strebler authored 2 months ago and Vakhtang Tsulaia committed 2 months ago

Expand CPAlgorithms tests to AF3

Matous Vozak authored 2 months ago and Duc Ta committed 2 months ago

Apply 1 suggestion(s) to 1 file(s)

Co-authored-by: Johannes Junggeburth <johannes.josef.junggeburth@cern.ch>

Martina Javurkova authored 2 months ago and Jean-Baptiste De Vivie De Regie committed 2 months ago

Add steering option to enable FastChain transform for running derivations and producing DAOD output format.

Thomas Strebler authored 2 months ago and Walter Lampl committed 2 months ago

Update JetAnalysisConfig.py with new forward jet calibration

Linghua Guo authored 2 months ago and Walter Lampl committed 2 months ago

Update R22 high-mu ey calibration recommendation with fixedE1E2 scale

Thomas Strebler authored 2 months ago and Walter Lampl committed 2 months ago

Updated Run3 MC Workflow AOD input to MC23

Tadej Novak authored 2 months ago and Walter Lampl committed 2 months ago

Apply 1 suggestion(s) to 1 file(s)

Co-authored-by: Baptiste Ravina <baptiste.ravina@cern.ch>

Rename /eos/atlas/atlascerngroupdisk/proj-sit -> /eos/atlas/atlascerngroupdisk/proj-ascig for CI and tools infrastructure

Julian Friedrich Wack authored 2 months ago and Walter Lampl committed 2 months ago

Modernize the algorithm

Adding Gaudi properties
Changing to range based for loops where possible
Using smart pointers

Tadej Novak authored 2 months ago and Duc Ta committed 2 months ago

Add explicit DBReplicaSvc configuration

Scott Snyder authored 2 months ago and Walter Lampl committed 2 months ago

Tier0ChainTests: Disable MT derivation tests.

Derivations are not expected to run in MT mode.
At the least, the truth dressing tool and overlap removal need to be
redesigned to avoid modifying the same decoration from different algorithms.
These tests were failing, so disable them.

Michael Holzbock authored 3 months ago and Frank Winklmeier committed 3 months ago

Fix seed for jet building in fJvt calculation

Do not fail `mkdir` when directory already exists. This is needed in
case the RC decides to overwrite existing references.