LCFI Flavour Tag Input Variables

LCFIAIDAPlot

Code:

If you use JAIDA (preferred):	If you use RAIDA:
LCFIAIDAPlot.cc	LCFIAIDAPlot.cc
LCFIAIDAPlot.h	LCFIAIDAPlot.h
ftplot.xml	(use same xml file)

The difference between the JAIDA and RAIDA versions is the use of IDataPointSets to make some of the plots.
However, IDataPointSets are not fully implemented in RAIDA so instead the plots are ICloud2Ds, which doesn't contain the errors on the parameters.

Examples of the plots and text produced are below. Under construction at the moment!
Comments are very welcome!

Let me know if you think anything else should be added or removed.
Do the titles and axis labels of the histograms make sense?
(At the moment, the colours are chosen by JAS3, but any comments on those are welcome too.)

I am in the process of writing a macro to plot these sensibly.
I want to keep this as lightweight as possible. I think a sample analysis package could be written to show people how to access some of the other information.

Example Output:

Sample of e+e- --> Z --> jet events at ECM=200 GeV. Generated by Ben Jeffrey.
Cuts applied:

| cos(theta_jet) | < 0.9
no jet pT cut

Files produced:

Optional printout of the numbers of jets: NeuralNetOutputFile.text
AIDA file: MyAida.aida (10 MB)

An example of the aida file directory structure is shown:

All plots produced by LCFIAIDAPlot are under the "MyLCFIAIDAPlot" directory. This contains at least two subdirectories:

A directory for each version of the tags you are looking at: in this case "FlavourTag". This case four subdirectories:

"AnyNumberOfVerticies",
"OneVertex",
"TwoVerticies",
"ThreeVertices".

Each of these contains plots showing the neural net outputs, efficiencies, leakage rates, purities etc

A directory called "FlavourTagInputs". It cotains four subdirectories:

A tuple containing all the quantities used to form the neural net flavour tags.
"bJets": histograms of all the quantities used to form the neural net flavour tags for true (MC) b-jets.
"cJets": histograms of all the quantities used to form the neural net flavour tags for true (MC) c-jets.
"udsJets": histograms of all the quantities used to form the neural net flavour tags for true (MC) light-jets: u, d or s.

Examples of the directory stucture are shown in JAS3 screenshots below.

Plots of the tagging output (B-tag, C-tag, BC-tag) for different jet samples.

	Output of the three flavour tags: B-tag, C-tag and BC-tag for B-jets, C-jets. 100 bins per plot
	Output of the three flavour tags: B-tag, C-tag and BC-tag for light-jets, and backgrounds. i.e. B-tag value of non B-jets (UDSC-jets) C-tag value of non C-jets (UDSB-jets) 100 bins per plot

Purity, Efficiency and Leakage Rates

	Purity-efficiency for the B-tag, C-tag and BC-tag for different numbers of verticies.
	Leakage rates for the B and C tags. i.e. how many of the wrong type of jets are tagged as a function of the tag NN cut.
	Leakage Rates vs Effiency of the tagger

Histograms of the Flavour Tag input variables, seperated for B-jets, C-jets and UDS-jets.

Only if a second vertex is found are the following histograms filled: DecayLength, DecayLength(SeedToIP), SecondaryVertexProbabilty,

PTMassCorrection is supposed to be PTMass. The names are picked up automatically from the

	FlavourTag input variables for B-jets (1)
	FlavourTag input variables for B-jets (2)
	FlavourTag input variables for C-jets (1)
	FlavourTag input variables for C-jets (2)
	FlavourTag input variables for UDS-jets (1)
	FlavourTag input variables for UDS-jets (2)

A tuple of the Vertex Tagging Inputs, one entry per jet.

Should we have a tuple in this program? It could be very useful, but I think it makes the output very large!
I will leave it as an option to be selected by the user.

	Vertex charge - optmised for b-jets
	Vertex charge - optmised for c-jets
	Significance of 1st track d0
	Significance of 2nd track d0 -100 is when no second track is found
	Decay length of

	Distance from vertex to IP


	Momentum of primary track
	Momentum of second track If there is no second track p=0 is filled
	Number of tracks found eminating from secondary vertex
	Number of vertices found in jet - including the IP
	Corrected jet momentum
	Jet momentum
	Probability that a secondard vertex is found in the jet
	True (MC) flavour of the jet (this is not one of the flavour tag inputs!!)
	Significance of the 1st track d0
	-100 set if there is no second track

victoria.martin@ed.ac.uk 04/09/07