BUG. OMP_NUM_THREAD variable interfers with nproc return value
nproc should return all installed processors. See #3108

Grigory Kozlov authored 1 year ago and Sergey Gorbunov committed 1 year ago

- Added template-based CaParameters
- Deleted inherited classes CaStationV/F/D, CaFieldV/F/D, have to be used with <DataT> instead

Administrator authored 1 year ago and Volker Friese committed 1 year ago

When comapring real numbers for equality one shouldn't compare the difference
of both values with zero. Due to rounding effects the result can be slightly
differnt from zero. If the difference of the values is smaller than a a small
value ( e.g. 0.00001 ) both values are equal.

Use the newly created or updated containers which contain the development
packages for lzma and zstd. Without the new containers the linking of the
flesnet binaries fail due to missing dependencies.

The Debian10 container is superseded by the container vae23. Remove the old
container from the test matrix and use onlx the new one.

Administrator authored 1 year ago and Pierre-Alain Loizeau committed 1 year ago

The Debian10 container is superseded by the container vae23. Remove the old
container from the test matrix and use onlx the new one.

Ca: comment out an unused pointer to the TimeSlice object to avoid assertions in event-by-event mcbm processing

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Use the same output format as all the other detector systems.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Up to now the input and event numbers where added to the match when the match
was created. In time based mode the digi and match objects are not necessarily
created at the same event to witch the MvdPoint belongs. This results in
wrongly matching the digis to the MvdPoints.
Add information about input and event number to the pixel charge to store the
the correct information about input and event numbers. Use this information
when creating the match object.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

The latency defines the time how much can happen between digi creation and
readout. Only after this time the daq buffer can be written to the output
timeslice without timly disordered digis.
Increase the default time and allow to set the time from the macro.
Add info output to the CbmDaq when the latency is changed.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Pass the correct information about the reconstruction mode (event or time
based) to the MVD cluster and hit finder. Switch of MVD reconstruction when
running in event based mode using the real event builder since the MVD is
currently not supported by the real event builder.

The tests from macro/C2F, macro/much and macro/dimuon still uses the old macro
run_reco_event.C which uses the macro reconstruct.C. The file defines the
reconstruction topology. The reconstruction used in C2F needs to use the
Ideal eventbuilder while the reco macro in dimuon doesn't produce tracks when
using the Ideal event builder.
Add the posibility to switch the usage of the ideal event builder on/off.
The test from both directories should use the macro run_reco.C in future.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Add CbmMvdQa to the reconstruction test to create residuals for the MVD.
Small fixes in CbmMvdQa to get the task running again. Attach tracks only of
requested. Fix issue when writting the output.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Add a fixed error of 8mum for the x- and y-position of the MvdHit.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Change the code such that the cluster finder also works event based.
Loop over all frames such that all data is written to the output after the
Exec function.
Refactor code.
Move code for cleanup in extra function.
Since TClonesArray->AbsorbAt() removes entries from the array looping from
front to back doesn't work correctly such loop from back to front.
Mark digis below the threshold as used such that they are removed from the
input array during cleanup.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

This latency is used by the CbmDaq to decide when the data can be moved from
the daq buffer to the final timeslice. If the latency is currently less than
110 microseconds some of the MVD digis are not properly time sorted. The
latency needs to be adjusted after the digitization parameters are finally
fixed.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Receice at the end of the run all data which is still in the buffers of the
mvd sensors, create digis and send them to the DAQ.

Administrator authored 1 year ago and Sergey Gorbunov committed 1 year ago

Fix the digi time. There were two problems pilled up. The first problem was to
get the event time to late during execution. By accident the event time was
then also used twice, once when creating the temporary CbmPixelCharge buffer
and a second time when creating the final digi.

Flush all data which is still stored in the temporary buffer at the end of
the run.