2017-09-08 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.2.3
* ConstituentSubtractor.cc
- In the previous version the zero pt massive corrected particles had pt set to 1e-200. This caused a warning when the active area was used during clustering. Now, the zero pt is set to 1e-50, which removed the warning.
* Makefile
- Temporarily removed the test example_whole_event_using_charged_info. The reason is that it gives different output particles when running on MacOS 10.12.6, Apple LLVM version 8.1.0 (clang-802.0.42) as Gavin Salam found. The difference is tiny: one zero pt and zero mass particle appears additionally when running on MAC, which has no impact on the performance of the correction.
2017-08-21 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.2.2
* ConstituentSubtractor.cc
* ConstituentSubtractor.hh
- Added function set_max_standardDeltaR for backward compatibility. It was replaced in version 1.1.3 by function set_max_distance. In this version, it does the same as the function set_max_distance.
- Further speed improvement for whole event subtraction (especially for small _max_distance parameters, such as 0.25)
2017-08-07 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.2.1
* ConstituentSubtractor.cc
- Fixed bug for the distance enum ConstituentSubtractor::angle. Previously, the optimized rapidity ranges were used also for this distance type which was wrong. Now, the whole rapidity range is used (without optimization).
- Checking if ghosts are constructed for the jet-by-jet subtraction. In case they are constructed, error is thrown since such ConstituentSubtractor should be used only for whole event subtraction, and not jet-by-jet.
2017-07-20 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.2.0
* ConstituentSubtractor.cc
* ConstituentSubtractor.hh
- Important change for the output corrected particles in case of correction of massive particles. The corrected zero pt particles with non-zero corrected delta_m (massive particles in rest) are now included in the output. Previously, all the zero pt corrected particles were discarded. Now, only the zero pt and zero delta_m corrected particles are discarded. The corrected particles with zero pt and non-zero delta_m are kept (their pt is set to very small value (1e-200 GeV)). This should have better performance for jet energy and jet mass. If the user wants, he/she can still discard those zero pt particles in his/her own code, e.g. by requiring particles with pt > 1e-100 GeV.
- Implemented changes to significantly increase the speed of the CS algorithm for the whole event subtraction. The performance is exactly the same. Only the speed is much higher. For small values of _max_distance parameter, the speed is now ~10-times higher. For larger values of the _max_distance parameter, the speed is ~2-times higher.
* RescalingClasses.cc
* RescalingClasses.hh:
- Added new rescaling classes:
- BackgroundRescalingYPhiFromRoot - rescaling of background using Root TH2 histogram binned in rapidity vs azimuth
- BackgroundRescalingYPhiUsingVectorForY - rescaling of background for heavy ions - using eliptic flow parameters and rapidity dependence recorded in a vector<double>
- BackgroundRescalingYPhiUsingVectors - rescaling of background using objcet vector<vector<double> > containing the rapidity vs azimuth dependence
* example_background_rescaling.cc
- Added more examples for usage of rescaling classes
* example_background_rescaling.ref
* example.ref
* example_whole_event.ref
* example_whole_event_using_charged_info.ref
- Updated example outputs due to the inclusion of massive zero pt corrected particles in the output.
2017-07-11 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.1.6
* ConstituentSubtractor.cc
* ConstituentSubtractor.hh
- Implemented suggestion from Marta Verweij:
- Added new functions: "get_ghosts" and "get_ghosts_area". They return the vector of ghosts and the corresponding ghost areas used for whole event subtraction.
- Changed the function "construct_ghosts_uniformly" to be public
- Now the user does not need to provide the BackgroundEstimator to CS, but he/she can do background estimation independently and then assign the estimated background to ghosts. The procedure is the following:
- construct the ghosts with function "construct_ghosts_uniformly"
- get them with functions "get_ghosts" and "get_ghosts_area"
- do background estimation independently and assign the corresponding background estimates to ghosts
- use the function "do_subtraction" to subtract the modified ghosts from real particles.
* RescalingClasses.cc
* RescalingClasses.hh:
- Added new rescaling class template BackgroundRescalingYFromRootPhi - useful mainly for heavy ion events. The rapidity dependance is taken from a Root TH1 object provided by the user. The next parameters are used to parametrize the eliptic flow.
* example_background_rescaling.cc
- Added example for usage of the new rescaling class BackgroundRescalingYFromRootPhi (commented at the beginning of the macro).
2017-01-25 Gavin Salam <gavin.salam@cern.ch>
* VERSION:
updated VERSION to 1.1.5
* Makefile:
- resolved issue with make fragile-shared, caused by repeated
SRCS lines in the Makefile. Also removed other duplicates.
2017-01-20 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.1.4
* ConstituentSubtractor.cc
- fixed rounding issues for zero masses, when doing the correction of term mtMinusPt
- more precise evaluation of corrected 4-momenta, which affected the mass of corrected particles with low masses
- added new function set_do_mass_subtraction
* ConstituentSubtractor.hh
- fixed initialization of member variable "_do_mass_subtraction"
- updated reference file. Added also all corrected particles.
2016-11-14 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
- Set version number to 1.1.3
* ConstituentSubtractor.cc
* ConstituentSubtractor.hh:
- Fixed bug in "do_subtraction" function for the unsubtracted background proxies.
- Added new function "sequential_subtraction" - for testing purposes so far, example will be added later. The ConstituentSubtraction is performed twice: first with small maximal allowed deltaR distance, then the background proxies are again uniformly redistributed and the ConstituentSubtraction runs second time with higher allowed maximal deltaR
- Added new function "set_distance_type". With this function, the user can change the definition of distance between particles and background proxies. There are two possibilities using enums Distance: deltaR (longitudinal Lorentz invariant distance in rapidity vs. phi) or angle (in Euclidean space). The default distance is deltaR, i.e. the same definition of distance as in previous versions.
* Makefile:
- Removed dependence on Root. Using template in RescalingClasses now.
* RescalingClasses.cc
* RescalingClasses.hh:
- Removed dependence on Root for the rescaling using TH1 histogram. Using template now. Also the name of the class changed. Old name: BackgroundRescalingYTH1. New name: BackgroundRescalingYFromRoot
- Changed the name of the rescaling class for heavy ion events. Old name: BackgroundRescalingYPhiHeavyIon. New name: BackgroundRescalingYPhi. Also the description of the rescaling function is updated.
* example_rescaling_using_TH1.cc
* example_rescaling_using_TH1.ref:
- Removed
* example_background_rescaling.cc
* example_background_rescaling.ref:
- Added new example for usage of Heavy Ion background rescaling function
- The file example_background_rescaling.cc contains also commented example for usage of rapidity rescaling according to root's TH1 distribution
* example_whole_event.cc:
- Updated this example with usage of the "set_distance_type" function.
2016-03-11 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
set version number to 1.1.2
* Makefile:
- added CXXFLAGS and LDFLAGS in case the "ROOTSYS" variable is defined from ROOT. In case it is not defined, the standard usage still works, just the RescalingClasses are not compiled.
* RescalingClasses.cc:
* RescalingClasses.hh:
- New files: contain classes for rho estimation rescaling. Two classes were added:
- BackgroundRescalingYTH1 - input is a TH1 object (stabdard histogram in ROOT) containg the actual rapidity dependence of backgound pt
- BackgroundRescalingYPhiHeavyIon - input are 8 parameters. The first four parameters parametrize the elliptic flow. The next four parameters parametrize the rapidity dependence using two Gaussian distributions
* example_rescaling_using_TH1.cc:
* example_rescaling_using_TH1.ref:
- New files: example in which the ConstituentSubtractor is used for the whole event subtraction using rapidity rescaling for the rho estimation. The input for rapidity rescaling is a TH1 object
2016-02-09 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
set version number to 1.1.1
* ConstituentSubtractor.hh:
* ConstituentSubtractor.cc:
- Faster algorithm for subtraction in the "do_subtraction" member function.
- New member function: subtract_event_using_charged_info
- whole event subtraction
- 4 additional inputs: two vectors of PseudoJets for charged background and charged signal, and two doubles, charged background scale (CBS) and charged signal scale (CSS)
- the scales CBS and CSS scale the input charged particles. Useful if one assumes correlation between charged and neutral particles or in case the inputs from calorimeter are miscalibrated wrt tracks. In case CBS=CSS=0, the input charged particles are not used. In case CBS=CSS=1, the input charged particles are not scaled. Recommending to try several combinations for CBS and CSS from range [0.8, 1.5].
- the background estimation is built in this function using GridMedianBackgroundEstimator with possibility to rescale
- see example file "example_whole_event_using_charged_info.cc" for usage
* example_whole_event_using_charged_info.cc:
* example_whole_event_using_charged_info.ref:
- New files: example in which the ConstituentSubtractor is used for the whole event subtraction using tracking information for charged particles
* functions.h:
- Updated function "read_event". Now, obtaining also the vectors of charged particles originating from signal and background.
2015-11-05 Peter Berta <berta@ipnp.troja.mff.cuni.cz>
* VERSION:
set version number to 1.1.0
* Makefile:
- added "-fPIC" CXX flag
* ConstituentSubtractor.hh:
* ConstituentSubtractor.cc:
- updated output for result(PseudoJet): copying the "user_index" from the input jet and its constituents to the output corrected jet and its constituents
- discarded the original max_deltaR parameter, and replaced it by parameter max_standardDeltaR defined as standardDeltaR=sqrt(deltay^2 + deltaphi^2). This new parameter restricts the usage of ghost-particle pairs which have too large distance standardDeltaR, i.e. not the distance deltaR which is the standardDeltaR multiplied by other factors as in the previous version.
- added a new multiplicative factor (sin(theta_i))^polarAngleExp into the definition of distance for a ghost-particle pair. Angle theta_i is the polar angle of partile i. The polarAngleExp is a free parameter with default value polarAngleExp=0, i.e. the default definition of deltaR has not changed wrt previous version. It can be changed using set_polarAngleExp member function
- changed the structure of the package and its member functions
- the standard usage with result(PseudoJet) is the same and it gives the same results as in previous version
- new member functions are available to simplify the whole event subtraction: subtract_event, set_background_estimator, set_scalar_background_density, see example file example_whole_event.cc
- similar changes as in GenericSubtractor due to FJ>=3.1:
- added set_common_bge_for_rho_and_rhom() and removed use_common_bge_for_rho_and_rhom()
- added set_use_bge_rhom_rhom() to tweak how rho_m is obtained when ConstituentSubtractor has been constructed with two estimators
- altered common_bge_for_rho_and_rhom() and added use_bge_rhom_rhom() to retrieve the behaviour
- added a warning when rhom is non-zero and unused
- Updated the terminology:
- ghosts - soft particles distributed with high density
- background proxies - particles used in the "do_subtraction" member function. They can be obtained from ghosts or in other way. Previously, these particles were also called "ghosts"
* example_whole_event.cc:
* example_whole_event.ref:
-new files: example in which the ConstituentSubtractor is used for the whole event
* functions.h:
-new file: contains useful functions for the example files
* example.hh:
-removed. Content moved to file functions.h
2014-04-06 Peter Berta <berta@ipnp.troja.mff.cuni.cz>