Computentp, Neural Nets and MCLIMITS

This page has been substantially rewritten (and remains a work in progress) to focus just on the information required for a successful run of the Computentp and Neural Net package, to deliver exclusions. For information on results obtained using inputs created in v12 of Athena, please refer to the archive. This page also describes how to run on GlaNtp – the version of the code set up for use in Glasgow, with no CDF dependencies. To use the previous version of the code (there are some important differences) refer to r93 and earlier.

To set up the current version of GlaNtp on Glasgow AFS, create a symbolic link to the setup script:

ln -s /afs/

then set up the environment:

source ./ -v 00-00-72 -b /afs/ -s GlaNtp\
Packagev17 -a

this will set up the environment for working in a v17 release of Athena, and it will make available the GlaNtp commands in the current working environment.

Note that an overview of the GlaNtp framework, made in doxygen, may be found here.

Project Aims

This project aims to document the use of an Artificial Neural Network (ANN) system and fitting software for the analysis of data from inclusive Higgs searches at ATLAS involving a lepton trigger and Higgs decay to b+bbar. This will use input from the Computentp software, designed to automate the weighting of the input files as required.


ANN :- This is a kind of algorithm with a structure consisting of "neurons" organised in a sequence of layers. The most common type, which is used here, is the Multi-Layer Perceptron (MLP), which comprises three kinds of layer. The input neurons are activated by a set trigger, and once activated they pass data on to a further set of "hidden" neurons (which can in principle be organised into any number of layers, but most frequently one or two - in our case one), and finally the processed data is forwarded to the output neurons.

The key feature of a neural network is its ability to be "trained" to recognise patterns in data, allowing high efficiency algorithms to be developed with relative ease. This training is typically done with sample data which has been generated artificially, resulting in an algorithm that is very effective at recognising certain patterns in data sets. The only shortcoming is the danger of "over-training" an ANN, meaning that it becomes overly discriminating and searches across a narrower range of patterns than is desired (one countermeasure is to add extra noise to training data).

Computentp – Simply running the code as above will result in less than optimal Neural Net training. The training procedure requires equal numbers of events from signal and from background (in this case it results in half of the signal events being used in training, half for testing). However, the above code will take events from the background signal samples in proportion to the file sizes - these result in proportions not quite in accordance with physical ratios. As the Neural Net weights results according to information about the cross-section of the process and so on stored in the tree, the final result is that while the outputs are weighted in a physical fashion, the Net is not trained to the same ratios, and so is not optimally trained. To solve this problem, Computentp is used to mix together all background and signal samples and assign TrainWeights to them, so that the events are weighted correctly for the Net's training.

Preparing samples for the Neural Net

Previous work went into producing samples for the Neural Net from AODs - results have previously been obtained for MC samples derived from v12 and v15 of athena, with work directed toward upgrading this to v16. The inputs are created from AODs using the TtHHbbDPDBasedAnalysis package (currently 00-04-18 and its branches are for v15, 00-04-19 is for v16), which can be found here. However, the work on athena v16 ran into problems with application of trigger matching, and so was curtailed for the moment.

Now inputs are produced in a two-step process from D3PDs. First, the desired D3PDs have cleaning cuts applied to them, and our desired event-by-event information is stored in a flat ntuple. This is performed by the code found here. Then we add in the global event variables (ones that are constant throughout the sample, such as luminosity and cross-section).

Current samples in use

Input data and cross-sections

These cross-sections are for the overall process, at √s = 7 TeV.

The ttH sample cross-sections are provided for the overall process – the MC is divided into two samples with W+ and W- independent of one another. These two samples are merged before being put through the ANN.

The tt samples were initially generated to produce the equivalent of 75 fb-1 of data, based on the LO cross-sections. Taking into account the k-factor of 1.84, this means that now all samples simulate 40.8 fb-1 of data. These samples have also had a generator-level filter applied - most events (especially for tt+0j) are of no interest to us, so we don't want to fill up disk-space with them, so we apply filters based on the numbers of jets etc. The Filter Efficiency is the fraction of events that pass from the general sample into the final simulated sample. To clarify how all the numbers hang together, consider the case of tt+0j. We have simulated 66,911 events - as said above, this corresponds to 40.8 fb-1 of data. We have a Filter Efficiency of 0.06774, so the full number of events that a complete semi-leptonic event would be comes to 987,762 events in 40 fb-1. Divide this by 40 to get the number of events in 1 fb-1 (i.e. the cross-section), and you get 24,694 events per fb-1. Our starting point for our cross-section is 13.18, with a k-factor of 1.84, which gives a cross-section of 24.25 – so all the numbers compare with each other pretty favourably. This of course makes getting from the number of sensible state events to the number expected per fb-1 rather easy – simply divide by 40.8. You'll notice that the cross-section includes all the branching ratios already, so we don't need to worry about that.

**IMPORTANT** The Filter Efficiency for these samples was calculated based on a no-pileup sample. The filter is generator level, and one of the things it will cut an event for is not enough jets. However, pileup adds jets, but these are added well after the filter. The net result is that a number of events that failed the filter would have passed, had the pileup been added earlier in the process. This means the filter efficiency (and thus the cross-sections) are incorrect, by a yet to determined amount....

For the other samples, however, we do need to worry about branching ratios - the quoted initial cross-section includes all final states, so we need to apply branching ratios to the cross-section to reduce it down, so that it reflects the sample we've generated. We then subsequently need to reduce the cross-section further so that it reflects the number of sensible states.

Sample Dataset numbers Cross-section (pb) Branching Ratios Filter Efficiency What the multiplicative factors are Effective cross-section (pb) Sources
ttH 109840, 109841 0.09756 0.676*0.216*2*0.675 0.8355 Overall 0.01607 Initial cross-section:
      0.676   W → hadrons   Branching ratios: 2008 PDG Booklet
      0.216   W → leptons (electron/muon)    
      2   Account for the 2 W decay routes    
      0.675   H → bb    
        0.8355 Lepton filter efficiency   Filter eff:
tt + 0j 105894, 116102 13.18 1.84   For sample 105894 24.25120 Initial cross-section and filter efficiency:
        0.06774 Filter efficiency for sample 116102   k-factor:
      1.84   k-factor    
tt + 1j 105895, 116103 13.17 1.84   For sample 105895 24.23280 Initial cross-section and filter efficiency:
        0.2142 Filter efficiency for sample 116103   k-factor:
      1.84   k-factor    
tt + 2j 105896, 116104 7.87 1.84   For sample 105896 14.48080 Initial cross-section and filter efficiency:
        0.4502 Filter efficiency for sample 116104   k-factor:
      1.84   k-factor    
tt + 3j 105897, 116105 5.49 1.84   For sample 105897 10.10160 Initial cross-section and filter efficiency:
        0.5860 Filter efficiency for sample 116105   k-factor:
      1.84   k-factor    
gg → ttbb (QCD) 116101 0.8986 0.676*0.216*2*1.84   Overall 0.48285 Initial cross-section:
      0.676   W → hadrons (electron/muon)   Branching ratios: 2008 PDG Booklet
      0.216   W → leptons (electron/muon)   k-factor: - need to verify there's nothing more suitable than applying tt+X value!
      2   Account for the 2 W decay routes    
      1.84   k-factor    
qq → ttbb (QCD) 116106 0.1416 0.676*0.216*2*1.84   Overall 0.07609 Initial cross-section:
      0.676   W → hadrons (electron/muon)   Branching ratios: 2008 PDG Booklet
      0.216   W → leptons (electron/muon)   k-factor: - need to verify there's nothing more suitable than applying tt+X value!
      2   Account for the 2 W decay routes    
      1.84   k-factor    
gg → ttbb (EWK) 116100 0.0875 0.676*0.216*2*1.84   Overall 0.04702 Initial cross-section:
      0.676   W → hadrons (electron/muon)   Branching ratios: 2008 PDG Booklet
      0.216   W → leptons (electron/muon)   k-factor: - need to verify there's nothing more suitable than applying tt+X value!
      2   Account for the 2 W decay routes    
      1.84   k-factor    
qq → ttbb (EWK) 116107 0.0101 0.676*0.216*2*1.84   Overall 0.00543 Initial cross-section:
      0.676   W → hadrons (electron/muon)   Branching ratios: 2008 PDG Booklet
      0.216   W → leptons (electron/muon)   k-factor: - need to verify there's nothing more suitable than applying tt+X value!
      2   Account for the 2 W decay routes    
      1.84   k-factor    

These cross-sections and branching ratios are correct as of 8 Feb 2011. qq→ttbb (EWK) is currently not being used, thanks to a bug in the production of the MC

Number of events surviving preselection, weights and TrainWeights

(See later in the TWiki for an explanation of weights and TrainWeights.) This table will be completed with all the relevant weights and TrainWeights at a later date - these values are to be compared to the output from Computentp to ensure everything is working as intended, and are calculated for the sensible cross-sections/events. (A quick check of the TrainWeight is to multiply the number so events of each background by their TrainWeight and sum them - by design, this should equal the number of entries in the ttH sample.)

Sample Dataset Number Pileup?   Number of events     Cross-section (fb)  
      Total Passing Preselection Sensible States Total Passing Preselection Sensible States
ttH (W+ sample) 109840 Yes 29968 2685 1936      
    No   2497 1761      
ttH (W- sample) 109841 Yes 29980 2764 2020      
    No   2600 1879      
ttH (total)   Yes 59948 5449 3956 16.07 1.460 1.060
    No   5097 3640   1.366 0.976
tt + 0j 105894 No 25487 6 5 24251 5.709 4.758
  116102 Yes 66911 149 123      
    No   78 66      
tt + 1j 105895 No 26980 21 18 24233 18.862 16.167
  116103 Yes 211254 960 787      
    No   638 517      
tt + 2j 105896 No 17487 69 53 14481 57.138 43.889
  116104 Yes 265166 2478 1957      
    No   2026 1548      
tt + 3j 105896 No 10990 96 77 10102 88.240 70.776
  116105 Yes 241235 3946 3022      
    No   3469 2619      
gg → ttbb (QCD) 116101 Yes 89887 3550 2560 483 19.070 13.752
qq → ttbb (QCD) 116106 Yes 19985 496 366 76.09 1.888 1.393
gg → ttbb (EWK) 116100 Yes 19987 981 706 47.02 2.308 1.661

Running the Neural Net

Things to do


In the script used to make the webpage showing the results, the reference to H6AONN5MEMLP is hardwired. It should become a argument. It is the name of the method you give TMVA in the training, and so if it changes in one you should be able to change in the other

Overview of the process

  1. Computentp takes in the original root files from ${ntuple_area}, and is steered based on steerComputentp.txt (created by genemflat_batch) . Computentp calculates the TrainWeight etc based on nGenForType etc, stored within the root files - no external hard-coding (I think).

  2. There are two main outputs from Computentp:

    1. computentp_output/ contains a file for each root file, containing all the variables for the tree - including weight and trainweight.

    2. trees/ contains all the same information, in one large file.

  3. The training runs on the Computentp output (the single large file), and produces the weights file (weights/) . It uses TrainWeight (branch inside Computentp output).

  4. The templating is what actually produces the NN score plots. It uses one file per signal/background - uses the original root files, and the ANN weight files as produced by the training. Will calculate the scale-factor to apply to each event (also often referred to as the weight) based upon NGenForType etc, which are within the root files.

  5. Using the MCLIMIT program, these ANN outputs are used to generate 1,000 pseudoexperiments (this number is set in the code, and can be adjusted if desired – the variable NPE in genemflat_batch).

    1. For each pseudoexperiment we simulate a Background-only sample of ANN output, which is subjected to both systematic and statistical (Poisson) variations. The Poisson variations are applied twice – once to each individual background, and then again to the sum of the backgrounds. (A possible bug?)

    2. For comparison to these null hypothesis datasets, for both signal and background an array of 10,000 templates is created, each of these templates being subjected to systematic and Poisson variation.

    3. The pseudodata and the array are passed to the function cslimit, which uses them to calculate an exclusion for each pseudoexperiment.

    4. Based on these 1,000 separate exclusions, we produce a final exclusion.

Setting up

Initial Setup

In the file, you should correct the line:

#PBS -j oe -m e -M

with your e-mail address – this enables the batch system to send you an e-mail informing you of the completion (successful or otherwise) of your job.

Trainweights and Weights

Initial versions of the code used to have the weights for the ANN hard-coded into the runcards (General Parameter FixWeight in the FlatReader file). However, later versions of the code do not need this hardcoding – weights are calculated from values found in the input files themselves, and so FixWeight has been set to 1 (this can be used to multiply certain samples' cross-sections by a given number, if desired). However, the formulae to obtain the weights are still quoted below, so that you can check the Computentp's work and make sure that it makes sense.

N.B. These weights are wrong for the ttjj (5212) sample. The input that was produced in v12 of athena was initially produced using MC@NLO. This produces both positive (+1) and negative (-1) weighted events (an easy way to consider this is to consider the negatively weighted events as destructive events, that interfere with the positively weighted events, with the net result of decreasing the cross-section of the process). We considered all events equally for our calculation of the weights, simply considering the total number of events in the files. This problem will disappear when we switch to v15 inputs, where the ttjj samples have been produced using Alpgen.

The first weight to be considered is TrainWeight – the scale factor we multiply each of the background events by so that they are in physically realistic ratios in relation to one another, while enforcing the requirement of the ANN training that we have equal numbers of signal and background events – used for the training of the ANN.

The calculation for the Trainweight is:

(Number of generated signal events / Number of generated events for that background) * (Cross-section for that background / Cross-section for all backgrounds combined)

The next weight is simply called Weight – this is the scale factor used to produce a physically realistic input for the ANN – now with the signal weighted as well. The formula used to find this is simply

(Number of events expected for your desired luminosity) / (Number of events present in input dataset)

Both of these numbers are calculated by Computentp, and can be checked in the file trees/NNInputs_120.root, where they are in branches labelled by 'TrainWeight' and 'weight'. However, it should be noted that while TrainWeight as produced by Computentp is used by the ANN in the training sequence, the final results are produced independently Computentp – the ANN calculates the weight on its own.

FlatReader and FlatPlotter

Then, creates a file called FlatPlotterATLAStth${prefix}.txt . This file is used in the 'templating' phase of and is based on the templates provided in teststeerFlatPlotterATLAStthSemileptonic-v15.txt.

And so via the FlatPlotter file the FlatReader files are included in the call:
runFlatPlotter \$steerPlotter ...
which produces a template for each of the signal and individual background samples.

User Setup

To set up the neural net,

  • check out the latest version of the code running framework from subversion (check what this is on trac) using the command

    svn co %BR%

  • check out a version of the GlaNtp code into your home directory (or set up to point at someone else's installation of the the code). The procedure for how to do this is described in the next section.

  • ensure you know the ntuple_area variable to be passed in at run-time to This will be the directory where the input ntuples are stored.

  • The BASEBATCHDIR is now set automatically to the working directory when the script is executed.

Getting a copy of GlaNtp

  • Set yourself up for access into SVN (using a proxy to access SVN, as described here)

    source /data/ppe01/sl5x/x86_64/grid/glite-ui/latest/external/etc/profile.d/

  • Create the directory where you want to set up your copy, and get a copy of the setup script (afraid the best place to get this script is from the scripts area of the GlaNtp code you're checking out. I am aware of the tautology of getting a script from the package so you can get the package, but that's the way it is. Just download this one file and go from there - you can delete it later when you've got the whole thing. (The code below assumes you're checking out from the trunk. Generally better to check out a specific tag, but the latest tag and the trunk should be the same, so you should just be able to copy and paste the below code.)

    mkdir /home/ahgemmell/GlaNtp
    cd /home/ahgemmell/GlaNtp
    svn co
  • You then need to set up your environment ready for the validation. This is done with the script, which is available within the NNFitter package. (Yes, I know - another case of getting the code before getting the code...) You run the script (which is also used for debugging the code) with

  • Make a directory to hold the code itself:

    mkdir GlaNtpPackage

  • not only checks out and compiles the code, it also then goes and validates it. sets up the environment variables so the validation data can be found.

  • You now run the script in the parent directory of GlaNtpPackage, specifying whether you want a specific tag (e.g. 00-00-10), or just from the head of the trunk (h) so you're more free to play around with it. It's always a good idea to check out a specific tag, so that whatever you do to the head, you can still run over a valid release.

    ./ SVN 00-00-10
  • This will check out everything, and run a few simple validations - the final output should look like this (i.e. don't be worried that not everything seems to have passed validation!):

HwwFlatFitATLAS Validation succeeded
Done with core tests
Result of UtilBase                          validation:  NOT DONE: NEED
Result of Steer                             validation:  OK
Result of StringStringSet                   validation:  OK
Result of StringIntMap                      validation:  OK
Result of ItemCategoryMap                   validation:  OK
Result of FlatSystematic                    validation:  OK
Result of LJMetValues                       validation:  OK
Result of PhysicsProc                       validation:  OK
Result of FlatNonTriggerableFakeScale       validation:  OK
Result of FlatProcessInfo                   validation:  OK
Result of PaletteList                       validation:  OK
Result of CutInterface                      validation:  NOT DONE: NEED
Result of NNWeight                          validation:  NOT DONE: NEED
Result of FlatFileMetadata                  validation:  OK
Result of FlatFileMetadataContainer         validation:  OK
Result of Masks                             validation:  NOT DONE: NEED
Result of FFMetadata                        validation:  OK
Result of RUtil                             validation:  NOT DONE: NEED
Result of HistHolder                        validation:  NOT DONE: NEED
Result of GlaFlatFitCDF                     validation:  OK
Result of GlaFlatFitBigSysTableCDF          validation:  OK
Result of GlaFlatFitBigSysTableNoScalingCDF validation:  OK
Result of GlaFlatFitATLAS                   validation:  OK
Result of FlatTuple                         validation:  OK
Result of FlatReWeight                      validation:  OK
Result of FlatReWeight_global               validation:  OK
Result of FlatReWeightMVA                   validation:  OK
Result of FlatReWeightMVA_global            validation:  OK
Result of TreeSpecGenerator                 validation:  OK
Result of FlatAscii                         validation:  OK
Result of FlatAscii_global                  validation:  OK
Result of FlatTRntp                         validation:  OK

Variables used by the GlaNtp package

The variables used by the package can be divided into two sets. The first are those variables that are constant throughout the sample - the 'global' variables (e.g. cross-section of the sample). These can be specified in their own tree, where they will be recorded (and read by GlaNtp) once only. If desired, these variables can be defined within the main tree of the input file - however, then they will be recorded once per event, and read in once per event. This is obviously a bit wasteful, but for historical reasons it can be done. To determine which of these behaviours you use, set LoadGlobalOnEachEvent in FlatPlotter and FlatReader to 1 for the events to be read in on an event-by-event basis, or 0 to be read in once from the global tree (or from the first event only). For more information on this switch, refer to this. The other variables are those that change on an event-by-event basis. These variables include both the variables we are going to train the Neural Net on (more information relevant to those variables is given in the relevant section of this TWiki), and other useful variables, such as filter flags (that tell GlaNtp whether an event is sensible or not). All of these variables are listed in the file VariableTreeToNTPATLASttHSemiLeptonic-v15.txt

The file maps logical values to their branch/leaf. The tree can be the global tree or the event tree.

GeneralParameter string 1 FlatTupleVar/<variable_name>=<tree>/<variable_name_in_tree>

Also specified are the name of the leaf for the cutmask and invert word -- these are global values for a file.

GeneralParameter string 1 CutMaskString=cutMask
GeneralParameter string 1 InvertWordString=invertWord

The structure of Computentp's output is specified by

ListParameter   EvInfoTree:1  1 NN_BJetWeight_Jet1:NN_BJetWeight_Jet1/NN_BJetWeight_Jet1

If you want a parameter to be found in the output, best to list it here....

Calculating my_integral (The Magic Formula)

1. Check the critical formula: The most important forumla is the first thing to check:

weight*= GetSF()*GetXsect()*GetBrFrac()*GetFilterEff()*GetLumiForType()/GetNGenForType(); my_brFrac my_filterEff my_xSect my_lumiForType my_nGenForType

You need a scale factor: my_sf?

2. Check the stuff that FlatReader uses. This is documented in GlaNtp/NtpAna/test/VariableTreeToNtp.txt. Variable Tree to Ntp is the one that maps logical values to their physical branch/leaf. Anything prefaced with FlatTupleVar needs to be specified or is useful to specify.

Values are divided in to those that can change on each event (kept in the "ev" tree) and those that are the same for a file (kept in the "global" tree). As you know you can set the tree names. You really should create a global tree for the global file values now. We have procrastinated on this a long time.

There you see # # Values that are required from global # GeneralParameter string 1 FlatTupleVar/BrFrac=globalInfo/BrFrac GeneralParameter string 1 FlatTupleVar/FilterEff=globalInfo/FilterEff GeneralParameter string 1 FlatTupleVar/Fraction=Fraction/Fraction GeneralParameter string 1 FlatTupleVar/Integral=Integral/Integral GeneralParameter string 1 FlatTupleVar/XSect=globalInfo/Xsect

# This specifies the name of the leaf for the cutmask and invert word. #Again, these are global values for a file. GeneralParameter string 1 CutMaskString=cutMask GeneralParameter string 1 InvertWordString=invertWord

This confirms that Fraction and Integral are needed. my_fraction my_integral

Here are the ones that are required for ev: # # Values that are required from ev # GeneralParameter string 1 FlatTupleVar/Channel=evInfo/Channel GeneralParameter string 1 FlatTupleVar/DilMass=evInfo/Mll GeneralParameter string 1 FlatTupleVar/Entry=evInfo/ientry GeneralParameter string 1 FlatTupleVar/Event=evInfo/eventNumber GeneralParameter string 1 FlatTupleVar/Lep1En=evInfo/lep1_E GeneralParameter string 1 FlatTupleVar/Lep2En=evInfo/lep2_E GeneralParameter string 1 FlatTupleVar/LumiForType=evInfo/lumiForType GeneralParameter string 1 FlatTupleVar/MEVal=LRInfo/LRHWW GeneralParameter string 1 FlatTupleVar/NGenForType=evInfo/nGenForType GeneralParameter string 1 FlatTupleVar/Njets=evInfo/Njets GeneralParameter string 1 FlatTupleVar/Rand=evInfo/Rand GeneralParameter string 1 FlatTupleVar/Run=evInfo/runNumber GeneralParameter string 1 FlatTupleVar/Weight=LRInfo/weight GeneralParameter string 1 FlatTupleVar/cutWord=evInfo/cutWord GeneralParameter string 1 FlatTupleVar/lep1_Type=evInfo/lep1_Type GeneralParameter string 1 FlatTupleVar/lep2_Type=evInfo/lep2_Type GeneralParameter string 1 FlatTupleVar/sf=evInfo/sf

Some can be used with the default values that FlatTuple gives: GeneralParameter string 1 FlatTupleVar/Channel=evInfo/Channel GeneralParameter string 1 FlatTupleVar/DilMass=evInfo/Mll GeneralParameter string 1 FlatTupleVar/MEVal=LRInfo/LRHWW GeneralParameter string 1 FlatTupleVar/Rand=evInfo/Rand GeneralParameter string 1 FlatTupleVar/lep1_Type=evInfo/lep1_Type GeneralParameter string 1 FlatTupleVar/lep2_Type=evInfo/lep2_Type

Some are ok to leave if you dont want to use it: there are switches that turn on the use of these GeneralParameter string 1 FlatTupleVar/Lep1En=evInfo/lep1_E GeneralParameter string 1 FlatTupleVar/Lep2En=evInfo/lep2_E GeneralParameter string 1 FlatTupleVar/Weight=LRInfo/weight

Some are useful for plotting: GeneralParameter string 1 FlatTupleVar/Njets=evInfo/Njets

I think you have my_Eventtype as channel my_failEvent as cutword

Variables that must be listed in the event (not the global) tree

nGenForType, LumiForType, Eventtype

Variables used for training the Neural Net

The list of variables on which the neural net is to train is set in the shell script, under TMVAvarset.txt (this file is created when the script runs). At present, these variables are:

The b-weights for the six 'leading' jets - currently the jets are ranked according to their b-weights, but it is possible to rank them according to pT and energy. The decision about how to rank them is done in the AOD -> NTuple stage:

The masses and pT of the various jet combinations (only considering the four 'top' jets - i.e. if ranked by b-weights, the jets that we expect to really be b-jets in our signal:

The sums of the eT of the two reconstructed tops, for each of the top three states:

And the differences between the eta and phi of the two reconstructed tops, again from the top three states:

You also need to provide addresses to the Neural Net so that it can find the variables in the input trees. This is done inside VariableTreeToNTPATLASttHSemiLeptonic-v15.txt :

ListParameter   EvInfoTree:1  1 NN_BJetWeight_Jet1:NN_BJetWeight_Jet1/NN_BJetWeight_Jet1

Currently all information is in the EvInfoTree, which provides event level information. However, future work will involve trying to establish a GlobalInfoTree, which contains information about the entire sample, such as cross-section - this will only need to be loaded once, and saves having to write the same information into the tree repeatedly, and subsequently reading it repeatedly.

Variable Weights in the Neural Net

To set up a neural net for the analysis of a particular kind of data it is necessary to train it with sample data; this process will adjust the "weights" on each variable that the neural net analyses in the ntuple, in order to optimise performance. These weights can then be viewed as a scatter plot in ROOT.

Specifying files as Signal/Background or as real data

The input datasets need to be specified in a number of peripheral files, so that the ANN can distinguish between signal and background MC files or real data files. There is only one place where data files need to be specified differently to MC - FlatAtlastthPhysicsProc1.txt - and if you are running the fit with data and not pseudodata, this is determined through one single flag, set in genemflat - see here. Errors for each process also need to be specified - how this is done is detailed in that section. The relevant files for adding processes are atlastth_histlist_flat-v15.txt, AtlasttHRealTitles.txt, FlatAtlastthPhysicsProc1.txt and FlatSysSetAtlastth1.txt. There are also some files that are produced through the action of At several points in these files, there are common structures for inputting data, relating to ListParameter and ColumnParameter:

ListParameter <tag> <onoff> <colon-separated-parameter-list>

<onoff> - specifies whether this parameter will be taken into consideration (1) or ignored (0) - generally this should be set to 1.
<tag> and <colon-seperated-parameter-list> - varies from process to process, will be explained for individual cases below. There can only be one instance of a <tag> active at any one time (i.e. you can write more than one version, but only one can be taken into consideration).

ColumnParameter <tag> <sequence> <keyword=doubleValue:keyword=doubleValue...>

The expression <tag>:<sequence> must be unique, e.g.

ColumnParameter   File         0 OnOff=0:SorB=0:Process=Data
ColumnParameter   File         1 OnOff=1:SorB=0:Process=Fake

where <tag> is the same, but <sequence> is different. The fact that the <sequence> carries meaning is specific to the implementation. Note that all of the values passed from ColumnParameter will eventually be evaluated as Doubles - any variables where you pass a string (as for 'Process' above), this is not actually passed to the code - these code snippets are to make the code more easily readable by puny humans, who comprehend the meaning of strings more readily than Doubles.


This file provides a map for the ANN, giving it the output file names (and in which directory they are to be stored, relative to ${template_area} - set in genemflat) and the tree structure where the final result of the ANN will be stored in the output (in the example below, the output file is ${template_area}/11602-filter.root, and the result graph will be FlatPlotter/NNScoreAny_0_0_0 0). The number to the left of the file name indicates which process it is - this is also used in the files AtlasttHRealTitles.txt and FlatAtlastthPhysicsProc1.txt, and needs to be consistent with them), and corresponds to the variable my_Eventtype in the input files (this can also be influenced in genemflat). Thus it is possible to assign multiple files to the same process (e.g. a file for the electron stream and muon stream are both assigned to Data), by giving them a common number at the start of the line.

0 116102-filter.root    FlatPlotter/NNScoreAny_0_0_0 0

The file in general expects each line to contain an integer, 2 strings and another integer, separated by spaces. If the integers are less than one, then that line is ignored. Therefore, so long as you are careful to exclude spaces from your strings, and stick to the string/integer formula, it is possible to place comments in this file:

-1 -------------------------------------------------------- x -1
-1 ttH x -1
-1 -------------------------------------------------------- x -1


The list of signal/background processes can be found in AtlasttHRealTitles.txt (where the names are specified and associated with process numbers - these process/number associations need to be the same as in atlastth_hislist_flat-v15.txt and FlatAtlastthPhysicsProc1.txt). At present these are:

Process_0_0 TTjj:Semileptonic
Process_1_0 ttH:Semileptonic
Process_2_0 EWK:Semileptonic
Process_3_0 QCD:Semileptonic

FlatStackInputSteer.txt / FlatStackInputSteerLog.txt

These files create the stacked input plots. FlatStackInputSteer.txt creates plots with a linear y-scale, FlatStackInputSteerLog.txt is with a log y-scale. Comments are within the file to explain the meaning of a few parameters.

genemflat creates the file TMVAsteer.txt, which sets a number of parameters for the running of the ANN - the constraints on the events, the precise Neural Net structure and so on - for establishing the input files, we are interested in only a couple of these parameters:

GeneralParameter  string      1 FileString=my_Eventtype

Indicates the leaf in the input file which shows which process the event belongs to - this is the same number as we specify later in genemflat for steerComputentp.txt - it does not have to be consistent with the process numbers as defined in atlastth_hislist_flat-v15.txt, AtlasttHRealTitles.txt and FlatAtlastthPhysicsProc1.txt.

ColumnParameter   File         1 OnOff=1:SorB=1:Process=tth

The number before the switches (OnOff, SorB, etc - in this case it is 1) corresponds to the number given in AtlasttHRealTitles.txt. The other numbers are self-explanatory - they establish if that file is to be used, if it is signal or background (1=signal. 0=background) and the name of the process. In this instance, the Process name is just a comment for your own elucidation - it is not used itself in the code, so does not necessarily have to correspond to the process names as provided in AtlasttHRealTitles.txt (though of course it is useful for them to be similar).

The other file that is produced by genemflat that specifies the input files for Computentp is steerComputentp.txt

# Specify the known metadata
ListParameter SignalProcessList 1 Alistair_tth
ListParameter  Process:Alistair_tth       1 Filename:${ntuple_area}/ttH-v15.root:File:${mh}:IntLumi:1.0

This is just a list of the various input files, and we specify the integrated luminosity. The 'File' parameter is only used for book-keeping by Computentp, and does not have to correspond to the file numbers used in the ANN steering files (or to my_Eventtype), but for sanity's sake it is probably best to keep things consistent. We make an exception for the signal - we assign it the number ${mh} - so that we can keep track of things if we have different mass Higgs in our signals.

#  Map of input file name to output file name: The ComputentpOutput will have a sed used to get the right mapping.
ListParameter  InputOutputMapName:1  1 ${ntuple_area}/ttH-v15.root:${Computentpoutput}/tth_NNinput.root

The InputOutputMapName is a list of integers - this doesn't have to bear any relevance to any numbers that have gone before - just give each output a unique number. This is followed by the mapping of input file names provided, to the output names that Computentp will produce.


This file contains various parameters:

ColumnParameter BackgroundList 0 tt0j=0
ColumnParameter SignalList     1 ttH=1
ColumnParameter DataList       1  Data=11

Here you specify once again the numbers assigned to the processes by my_Eventtype (for tt0j it equals zero), and list things as BackgroundList, SignalList or DataList. The number after 'BackgroundList' or 'SignalList' is unique for each process (to preserve the uniqueness of <tag>:<sequence>), but it must be sequential, running from 0 to n-1 (where you have n samples) - apart from for DataList entries (as shown above). It also does not need to correspond to my_Eventtype, however, for completeness' sake within this file I have set it as such. The number at the end of this declaration (tt0j=0 in this case) needs to be sequential - it instructs the net of the order in which to process the samples, so it must go from 0 to n-1 (when you have n samples). It must match up with the numbers provided in atlastth_histlist_flat-v16.txt and AtlasttHRealTitles.txt so that processes and data can be matched to the various individual files.

ColumnParameter PseudoDataList 0 tt0j=0

This is simply a restatement of the BackgroundList (as we're looking for exclusion, the pseudodata is background only) - the same numbers in the same place. This list specifies the processes included in the pseudoexperiments, and therefore the signal process is not included in this list.

ListParameter     ProcessLabels:1 1 tt0j:t#bar{t}0j

The number after ProcessLabels again doesn't correspond to my_Eventtype - I have made it the same as the number after BackgroundList/SignalList and PseudoDataList. The important feature from this is that it tells the ANN what to label each of the various processes as in the results plots. The numbers must run from 1 to n.

ColumnParameter UCSDPalette 0    tt0j=19
ColumnParameter PrimaryColorPalette 0    tt0j=0

These two parameters specify the colours used in the plotting for each of the processes (the numbers correspond to those in the Color Wheel of TColor). The numbers after the UCSDPalette and PrimaryColorPalette are the same ones as have been used previously in this file. Whether the plotting uses the colours stated in UCSDPalette or PrimaryColourPalette is determined in the file flatsteerStackNNAtlas.txt by setting the parameter:

GeneralParameter  string       1 Palette=UCSDPalette

The final parameter to be set in FlatAtlastthPhysicsProc1.txt is:

ColumnParameter ProcessOrder 0 tt0j=0

Once again, the number on its own (in this case 0) is the same as the other such instances in this file. The final number (zero in this case) is the order in which this process should be plotted - i.e. in this case, the tt0j sample will be plotted first in the output, with the other samples piled on top of it. This number obviously does not need to correspond to my_Eventtype.


This file contains all the information on the errors that you pass to the ANN so that it can work out how the errors propogate to the final plots and answers, and so most of the details of this file will be covered in that section. The basic format of the file is:

ColumnParameter   Combine:Lumi   0  OnOff=1:Low=-0.11:High=0.11:Channel=1:Process=TTjj

The <sequence> parameter (in this case '0') is there so that you can specify the parameters for a given error for multiple channels, without falling foul of the uniqueness requirement for <tag>:<sequence>. We have chosen it so that it equals my_Eventtype for that process. 'Channel' is present just in case you're considering multiple channels. We're only considering the one channel in this case (SemiLeptonic). The final parameter (Process) is not actually used - the second parameter tells the ANN which errors are which, but this isn't very easily read by you, so feel free to add it in to help you keep track of the various errors! These final few parameters can be placed in any order, so long as they are separated by semicolons.


This file contains parameters to control the loops over events.

GeneralParameter int    1 NEvent=20000000
GeneralParameter int    0 FirstEvent=1
GeneralParameter int    0 LastEvent=10

FirstEvent and LastEvent allow you to specify a range of events to run over - this is liable only to be useful during debugging. (Note that these parameters are currently turned off). NEvent gives the maximum number of events processed for any given sample - take care with this, if you are running a particularly large sample through the code....

Setting which variables to plot and train on

You need to let GlaNtp know where the variables you are interested in are. It is also possible to merely plot some variables you're interested in without adding them to the training just yet. You need to provide GlaNtp with the location of the variables in all cases - this is done in VariableTreeToNTPATLASttHSemiLeptonic-v16.txt and TreeSpecATLAStth-v16_event.txt (or TreeSpecATLAStth-v16_global.txt as applicable).


ListParameter   EvInfoTree:1  1 my_NN_BJetWeight_Jet1:my_NN_BJetWeight_Jet1/my_NN_BJetWeight_Jet1

This information must be provided for every variable you're interested in in any way. It provides the variable name, and a map to that variable name from the input tree. Note that the number after EvInfoTree must be unique for each entry (EvInfoTree:2, EvInfoTree:3, etc)


ListParameter    SpecifyVariable:my_NN_BJetWeight_Jet1  1 Type:double

This is another compulsory piece of information for GlaNtp - telling it which tree the information is in (event or global) and the event type.


ColumnParameter   SpecifyHist:my_NN_BJetWeight_Jet1    0  OnOff=1:Min=-5:Max=10:NBin=25

This is just for the plotting scripts (but if you're training on variables, you should probably want them plotted as well...). The number after the SpecifyHist string (in this case 0) needs to be different for each entry. OnOff decides whether the variable is to be plotted or not, and must be specified. Min and Max specifiy the range of the x-axis (for energy / mass, this is in units of MeV), and unless specified defaults to 0 and 200 respectively. NBin specifies the number of bins in the histogram, with the default of 50.

TMVAvarset.txt (

This is for the templating - a nice and simple list of all the variables you want to train on. Simples.

FlatStackInputSteer.txt / FlatStackInputSteerLog.txt

Parameters for the templating and the making of the stacked plots. Individual paramters are commented within the file itself.


Called by the FlatStack files. It allows for more instructive axes labels.

ListParameter   DiscrToLabel:7  1 my_NN_BJet12_M:M^{BJet}_{12}\(MeV/c^{2}),MeV/c^{2}
The number number following DiscrToLabel must match that given in VariableTreeToNTPATLASttHSemiLeptonic-v16.txt. Then comes the real variable name - the name that the code deals with. Following the semicolon is the x-axis label, written in LaTeX style formatting. The backslash denotes a space in the axis label (the parameter must be one long continuous stream). The bit after the comma is optional, but if used specifies the units for the y-axis (e.g. # events per MeV). These labels are written using Root's LaTeX markup.

Setting Systematic Uncertainties

The fitting code can take into account two different types of systematic uncertainty - rate and shape. The basic method to obtain both these uncertainties is that you should make your input samples for both your nominal sample, and for the two bounds of a given error (e.g. Initial State Radiation, ISR). Repeat this for all of the errors you wish to consider. The rate systematic uncertainty is simply how the number of events change that pass your preselection cuts etc. (you can only consider this, if you like). To obtain the shape uncertainty, you should pass each of the resulting datasets through the ANN (up to and including the templating, so that you have ANN results for both the nominal results, and as a result of varying each background). These ANN outputs can then be used to produce the rate uncertainties based on their integrals, before being normalised to the nominal cross-section so as to find the shape uncertainty - a measure of the percentage change in the bin-by-bin distribution for each error.

The fitting code is passed the relevant information about errors through the use of a number of files, but in the simplest case (when shape uncertainties are not being considered), there are only two: FlatSysSetAtlastth1.txt and SysNamesAtlastth1.txt. The basic call to the fitting code is in

mkdir -p templates/fit
rm -f templates/fit/out_${mh}.log
Fit ${basehistlistname} ${template_area}/ \$sysfile \$steerfile $mh > templates/fit/out_${mh}.log

The final call is rendered in the actual job file (e.g. run114) as

Fit /home/ahgemmell/NNFitter-00-00-09-Edited/NNTraining/atlastth_histlist_flat-v15.txt templates/tth120/ $sysfile $steerfile 120 > templates/fit/out_120.log

If you want to save time, (by not having to run templating for every error you wish to consider), you can instead only consider the rate uncertainties, and provide these as fractional changes to the rate, specified in FlatSysSetAtlastth1.txt. Whether or not you consider shape uncertainties is controlled by a couple of parameters in the steering file FlatFitSteer,txt, (which is created by the action of

GeneralParameter  bool        1 UseShape=0
GeneralParameter  bool        1 UseShapeMean=0

Setting UseShape=1 means shape uncertainties will be taken into account for all the uncertainties that you provide the extra steering files and ANN scores for, UseShapeMean=1 means that the ANN results for your various uncertainties will be used to produce the rate uncertainties based on their integrals, rather than on the numbers provided in FlatSysSetAtlastth1.txt - using the relative sizes of the integrals of the AAN output as an estimator of the rate uncertainty can be useful if you don't want to be subject to statistical variations in the computation of your systematic uncertainties (if UseShapeMean=0, the systematic rate uncertainty is calculated as a fractional change on the nominal rate). Considering shape uncertainties requires more steering files, and this will be detailed in later.


ColumnParameter   Combine:Lumi   0  OnOff=1:Low=-0.11:High=0.11:Channel=1:Process=TTjj

The first parameter consists of two parts in this example: 'Combine' and 'Lumi'. The second part is the name of the uncertainty being considered. The first part 'Combine' (and the associated semicolon between them) is optional. It tells the ANN that the uncertainty thus labelled are independent of each other, and can be added in quadrature. 'OnOff' obviously tells the ANN to consider those uncertainty (1) or not (0). 'Low' and 'High' establish the relevant bounds of the uncertainty as fractions of the total (however, for the ANN these uncertainties are symmetrised, so to save time they are here assumed to be symmetric unless elsewhere stated) - note that these are not the uncertainties on the quantity, but rather the effect of that uncertainty on the rate of your process. Process is not actually read by the ANN, but is there to make the whole thing more human-friendly to read. The current errors, and their bounds are below. If no source for these error bounds is given, then they were the defaults found in the files from time immemorial (where as necessary I assumed that all tt + X errors were the same, as were all ttbb (QCD) errors, as in the original files the only samples considered were ttjj, ttbb(EWK), ttbb(QCD) and ttH - these errors probably originate from the CSC note). If you are only considering rate uncertainties, this is where the fitting code will find the relevant numbers.

Error Combined? Process Upper/Lower Bound Source
Luminosity Yes All 11%
Trigger Yes tt + X 1.5%  
    ttbb (EWK) 1.4%  
    ttbb (QCD) 1.3%  
    ttH 1.5%  
Lepton ID Yes Backgrounds 0.3%  
    Signal 0.6%  
MET No All 1.0%  
NLO Acceptance No tt + X 5.5%  
    Others 10%  
X-Section No All 10%  
PDF No tt + X 1.9%  
    ttbb 2.7%  
    ttH 2.2%  
b-tagging No Backgrounds 20%  
    Signal 16%  
JES No Backgrounds 5.0%  
    Signal 9.0%  


ListParameter   SysInfoToSysMap:1 1  Combine:LumiTrigLepID

The number in the <tag> after SysInfoToSysMap is unique for each error (in this case it goes from one to eight). There is one entry per error considered, apart from the cases where the errors are combined in quadrature (as specified in FlatSysSetAtlastth1.txt), where they are given one entry to share between them. The <colon-separated-parameter-list> provides a map between the name of the errors as considered by FlatSysSetAtlastth1.txt (the errors combined in quadrature are lumped together under the name 'Combine'), and something more human-readable. The human-readable names are what will be written out by the fitting code (which identifies each error based on numbers, rather than the names in FlatSysSetAtlastth1.txt) when it is producing its logfile. Obviously there is often not much change between the two names, apart form in the case of Combined errors.

Including shape uncertainties

For the sake of argument, we shall pretend to only be considering the one error overall. It is possible to consider rate errors independently of shape uncertainties (by setting UseShapeMean=0 in FlatFitSteer.txt) - this might be useful and quicker to run if a given error produces a large rate error, but the change to the shape of the ANN distribution is minimal (you can have a look at the ANN results yourself and make your own judgements). If you are not considering a given shape uncertainty but you are considering the rate uncertainty, all that needs to be done is to not produce the relevant steering files.

In this example, we will already have run three ANN templating steps - run1 (the nominal run), run2 (the results of taking the lower bound of the error) and run3 (the results of taking the higher bound of the error). You can then move into a new directory (e.g. run1_2) in which you want to perform the fitting, and at the very least set UseShape=1 in FlatFitSteer.txt (also perhaps setting UseShapeMean=1). This requires some changes to the call to the fitting code and atlastth_histlist_flat-v15.txt so that the combination of ${template_area} and the filenames given in atlastth_histlist_flat-v15.txt still point toward the ANN template files you wish to consider - as shown in the two lines below (the first from genemflat establishing ${template_area}, the second from atlastth_histlist_flat-v15.txt establishing the filename for the ANN template):

0 116102-filter.root    FlatPlotter/NNScoreAny_0_0_0 0

could become:

0 run1/templates/tth120/116102-filter.root    FlatPlotter/NNScoreAny_0_0_0 0

This ensures that atlastth_histlist_flat-v15.txt will still point toward the ANN templates from the nominal run. You must now create additional steering files to point toward the high and low error ANN templates - their names are of the format:


where HistOutput is atlastth_histlist_flat-v15.txt and errorname is the human-readable error name, as defined in SysNamesAtlastth1.txt. You also need to change the ${basehistlistname} in the call to the fitting code so that it points directly at atlastth_histlist_flat-v15.txt, with no preceding directory structure - the code bases the names of the two extra shape steering files on this argument, and will not take into account any directories in the argument. (So that if ${basehistlistname} was directory/file.txt, the fitting code would look for the extra steering files with the name ShapePos _ISR_directory/file.txt in the case of ISR being our error).


It is possible for the inputs to the ANN to have more events in than those that you want to pass to on for processing. We only want to train the ANN on those samples that would pass our preselection cuts - general cleaning cuts and the like. (There was a previous version of our inputs where we also required 'sensible states' - for each candidate event we required it to reconstruct tops and Ws with vaguely realistic masses. However - this is a Neural Net analysis, so it has been decided to remove these cuts - they will after all in effect be reintroduced by the net itself if they would have been useful, and by not applying them ourselves, we are passing more information to the net.) We therefore have filters so that Computentp and the ANN only look at events of our choosing. These filters take the place of various bitwise tests in TMVAsteer.txt (created in (not currently used, as explained below) and TreeSpecATLAStth_global.txt.


GeneralParameter string 1 FlatTupleVar/cutWord=my_GoodJets_N/my_GoodJets_N

This sets the variable we wish to use in our filter - it interfaces with the cutMask and invertWord as specified in TreeSpecATLAStth.txt. Note that depending on the number of jets you wish to run your analysis on (set as a command line argument during the running of the script), this is edited with genemflat.


In TreeSpecATLAStth.txt, we establish the filters which control what is used for the templating, and Computentp:

ListParameter SpecifyVariable:Higgs:cutMask 1 Type:int:Default:3
ListParameter SpecifyVariable:Higgs:invertWord 1 Type:int:Default:0

InvertWord is used to invert the relevant bits (in this case no bits are inverted) before the cut from cutMask is applied. The cutMask tells the filter which bits we care about (we use a binary filter). So, for example, if cutMask is set to 6 (110 in binary), we are telling the filter that we wish the second and third bit to be equal to one in cutWord - we don't care about the first bit. It is possible to specify multiple options of the cutMask and invertWord in the same file, distinguished by the word after SpecifyVariable (in this case Higgs). Which ones are used are determined by teststeerFlatReaderATLAStthSemileptonic-v16.txt.

TMVAsteer.txt (

GeneralParameter string 1 Constraint=(my_failEvent&3)==3

This controls the events used in the training, using a bitwise comparison. If the constraint is true (i.e. the first two bits are set, and not equal to zero), then the event is used for training. This filter is not used currently, as training of the net takes place based on the Computentp output - this Computentp output only contains sensible states (as specified in the TreeSpecATLAStth.txt file's filter). If further filtering is required, then care must be taken to ensure that my_failEvent (or whatever you wish to base your filter on) is specified in the VariableTreeToNTP file, so that Computentp will copy it into its output.

**If USEHILOSB is set to 1 then && must be appended to cut criteria, e.g. GeneralParameter string 1 Constraint=(my_failEvent&65536)==0&&. This is because USEHILOSB adds more constraints.**


GeneralParameter string 1 ControlRegion=Higgs
Specifies which cutMask and invertWord are to be used from TreeSpecATLAStth_global.txt. This is changed at runtime with one of the parameters


To run the script, first log into the batch system (ppepbs).

The (NNFitter 00-00-21 version) script can be executed with the command (the last argument can be optional):

./ 12 400 1.04 tth 120 120 6 Higgs 00-00-45 /data/atlas07/stdenis/v16-r13/bjet2 srv001 ahgemmell

These options denote:

  • 12 is the run number

  • 400 is the jobstart - this is a potentially redundant parameter to do with the PBS queue.

  • 1.04 is the luminosity that will be normalised to (in fb^-1).

  • tth is the process type - aim to develop this to incorporate other processes, e.g. lbb

  • 120 is the min. Higgs mass

  • 120 is the max. Higgs mass

  • 6 is the number of jets in the events you want to run over (i.e. this is an exclusive 6 jet analysis - events with 7 jets are excluded)

  • Higgs controls which cutMask and invertWord you wish to use, as specified in TreeSpecATLAStth-v16_global.txt. Current options are 'Higgs' and 'NoCuts'

  • 00-00-45 is the release of GlaNtp that you are using for your run

  • /data/atlas07/stdenis/v16-r13/bjet2 is the directory where the input ntuples are located (having my_failEvent bits set for ( 65536 for >0 sensible states) and ( 131072 for 4 tight b-tagged jets)

  • is your email address, so the batch system can let you know when the jobs are done

  • srv001 is the Neurobayes server you want to run (if you're running a TMVA run, this is less important. There are 10 servers, 001-010. Servers 001-005 are on ppepc23, servers 006-010 are on ppepc39. This is related to the last argument you can pass to the script.

  • ahgemmell is your Glasgow username, used for Neurobayes servers

  • is the machine your Neurobayes server is located on. If you don't provide this argument, it defaults to ppepc23.


  • Creates a run12 subdirectory in working directory and makes it the working directory

  • Creates TMVAsteer.txt - writes fitting parameters to it

  • NN structure is set ( H6AONN5MEMLP MLP 1 H:!V:NCycles=1000:HiddenLayers=N+1,N:RandomSeed=9876543). This line sets up two hidden layers with N+1 and N neurons respectively (where N is the number of input variables).

  • Training cycles (1000) and hidden layers - N+1?

  • 4 text steer files are copied into the run directory for templating

  • 2 text steer files are copied into the run directory for stacking plots.

  • 2 lines of text are appended to a temporary copy of flatsteerStackNNAtlas.txt: GeneralParameter string 1 HWW=tth-TMVA
    GeneralParameter double 1 <nop>IntLumi=${lumi}

Contents of training file:

  • jetmin/jetmax - These seem to be redundant. Commented out, effective as of v.3

  • zmin/zmax (1/2) - what is their function?

  • weighting = <nop>TrainWeight - is this redundant?

  • TMVAvarset.txt - input variable set

Other switches to influence the running

At the start of the file there are a number of switches established:

# Flags to limit the scope of the run if desired

These control whether or not various parts of the code are run - the names of the flags are pretty self-explanatory about what parts of the code they control. For example, it is possible to omit the training in subsequent (templating) runs, if it has previously been done. This shortens the run time significantly.

***NOTE*** The flags DoTraining and DoTemplates had previously (until release 00-00-21) been set on the command line. They were moved from the command line when the other flags were introduced.

If you wish the fit to be run using data and not pseudodata, then the flag is set in FlatFitSteer.txt, which is created in genemflat:

GeneralParameter  bool        1 PseudoData=1

If this flag is set to 1 then pseudodata is used, 0 causes data to be used.

teststeerFlatPlotterATLAStthSemileptonic-v16.txt and teststeerFlatReaderATLAStthSemileptonic-v16.txt

GeneralParameter bool   1 LoadGlobalOnEachEvent=0

Determines if you have a separate global tree or not. If you do not, set this equal to one, and the relevant global values will be read out anew for each event from the event tree.

Important notes about running parts of the code (not a complete run - for debugging, replotting etc)

  1. When you want to only make stacked input plots (e.g. when you've decided you want to change the scale of one plot), you must also redo the templating. Just redoing the stacked inputs on their own does nothing new. The histograms for the stacked inputs are booked during the templating.

Where the output is stored

  1. Computentp120.log

    The log file from Computentp -- more information about the information contained within it is found here

  2. stackedinput/StackInput/tth120

    A collection of root files and eps files showing pretty stacked plots of all the input variables.

  3. trees/NNInputs_120.root

    The output from Computentp - it is a copy of all of the input datasets, with the addition of the variables TrainWeight and weight.

  4. computentp_output/

    The same as for trees/NNInputs_120.root, but now each individual input dataset has its own appropriately named file.

  5. weights/TMVTEST2_120_3.root_H6AONN5MEMLP.weights.txt

    Contains the weights of the Neural Net - results of the training. Starts with info about the run: Date, time, etc. followed by various parameters you've set. Then the variables you're considering in the net, with their range. Finally comes the Neural Net structure itself - weights etc.

  6. TMVA2_120_3.root

    1. InputVariable _NoTransform/my_----_B_NoTransform

      The distribution of this variable for the Background (or signal for S). N.B. Has N_sig / 2 entries (other half used for testing, not training). Check this to make sure there are no obvious problems with the input - missing data points, random spikes, etc.

    2. Method_MLP/H6AONN5MEMLP/estimatorHistTrain

      A measure after each iteration of how much tweaking is required to get S=1, B=0. Should settle after a while to a stable number

    3. Method_MLP/H6AONN5MEMLP/estimatorHistTest

      Same, but for the test sample - should look broadly similar. Otherwise we're over/undertraining.


      The final Neural Net Score for the Background (or signal for S) (Test result - not the final result).


      Rick's not sure.


      Comparing Bkg rejection to Signal Efficiency. Ideally want rej=1 with eff=1 (i.e. reject all bkg, accept all signal).

  7. train2.log

    A text summary of the training process. Mentions files used (how many files, numbers of events, if they signal/background, etc). Will refer to file 0, 1... - refer to Atlas ttHRealTitles.txt to work out which file is which. Use this to make sure the right numbers of events are getting though your filters. At the start of the file is a summary listing some relevant information about the input files:

    ParamInfo: File:0
       OnOff : 1
       Process : 2.19001e-314
       SorB : 0
    These reflect the parameters as set in TMVAsteer.txt (created via genemflat_batch_Complete2_SL5_sh). Note that the number following 'Process' is nonsense (and in later releases of GlaNtp is not present) - that parameter is there in the steering file simply to make it more human-readable. However, the code still tries to read it in, but can only handle doubles - the net result varies from run to run, but can always be safely ignored.
  8. templates/out/FlatPlotter${prefix}.out

    (Where ${prefix} is the process name.) The output from the screen from the templating stage, where the Neural Net scores are actually produced, detailing errors etc. It also contains a more detailed breakdown of the event numbers and the yields (event numbers in the specified luminosity) at various points:

    ****** FlatReader Info Start ******
    Entries examined : 2884
    Events:00 Seen : 2884
    Events:00 Seen Any : 2884
    Events:01 Passing Mask Selection for Higgs : 1800
    Events:01 Passing Mask Selection for Higgs Any : 1800
    Events:02 Passing DilPairType Selection for ALL : 1800
    Events:02 Passing DilPairType Selection for ALL Any : 1800
    Events:30 Passing Selection : 1800
    Events:30 Passing Selection Any : 1800
    Events:40 Passing Bad Lepton Energy Filter : 1800
    Events:40 Passing Bad Lepton Energy Filter Any : 1800
    Events:9999 Final Selection : 1800
    Events:9999 Final Selection Any : 1800
    No precomputed weight: Weight Difference not checked : 148336
    UPEvents: 00 Seen : 2884
    UPEvents: 00 SeenAny : 2884
    UPEvents:30 Passing Selection : 1800
    UPEvents:30 Passing SelectionAny : 1800
    UPWtEvents: 00 Seen : -2115.31
    UPWtEvents: 00 SeenAny : -2115.31
    UPWtEvents:30 Passing Selection : -1318.36
    UPWtEvents:30 Passing SelectionAny : -1318.36
    WtEvents:00 Seen : -2115.31
    WtEvents:00 Seen Any : -2115.31
    WtEvents:01 Passing Mask Selection for Higgs : -1318.36
    WtEvents:01 Passing Mask Selection for Higgs Any : -1318.36
    WtEvents:02 Passing DilPairType Selection for ALL : -1318.36
    WtEvents:02 Passing DilPairType Selection for ALL Any : -1318.36
    WtEvents:30 Passing Selection : -1318.36
    WtEvents:30 Passing Selection Any : -1318.36
    WtEvents:40 Passing Bad Lepton Energy Filter : -1318.36
    WtEvents:40 Passing Bad Lepton Energy Filter Any : -1318.36
    WtEvents:9999 Final Selection : -1318.36
    WtEvents:9999 Final Selection Any : -1318.36
    ****** FlatReader Info End ******

    In 'Events', the first number, the entries examined, is the number of entries in the MC sample being passed to the Neural Net. The Final Selection is the number of entries that make it past your cutmask etc to actually be passed to the Neural Net. The intervening numbers are at this stage rather meaningless. 'UPWtEvents' can be safely ignored in its entirety. 'WtEvents' provides the same numbers as in 'Events', but this time with Scale Weights etc applied - these are now the yields.

  9. templates/ttH120/

    Files showing the final results of the ANN. NNScoreAny _0_0_0 is of particular interest - this is the ANN output for that file, weighted to represent real data (i.e. if you simply add all of these together, you should get a realistic 'Signal and Background' result, which is shown in:

  10. stacked/Plots_120_TMVA_Lum_1.0/FlatStack_1.eps


    The final ANN scores of the signal and background, scaled as to real data

  11. stacked/out1_120.log


    Output from making the stacked (combined) plots.

  12. templates/fit/out_120.log

    This is the paydirt - the output to screen from the fitting stage. Look at the end of this file, and you shall see the exclusions generated by all the pseudoexperiments performed on the ANN output, in a little table, also giving +/- 1/2 sigma results.

    From here you can also get a record of the yields of the various processes considered by the Neural Net (the number of events expected for each process after applying all our preselection cuts, cutmasks, etc in our specified luminosity):

    Print out Yield ================
      Channel   & tt       & ttH     & ttbb    & Wlnu     & Wbb     & Wc      & Wcc     & st_Wt   & st_schan & st_tchan & eFake    & Data\\
    SemiLeptonic& 2154.475 &   2.598 &  49.313 & 4140.901 & 160.634 & 498.841 & 322.464 &  83.942 &   3.265  &  18.288  & 4069.197 & 10616.000
           sum  & 2154.475 &   2.598 &  49.313 & 4140.901 & 160.634 & 498.841 & 322.464 &  83.942 &   3.265  &  18.288  & 4069.197 & 10616.000
    NSig 2.59829 NBkg 11501.3 NData 10616
    End Print out

    Just above this is information of the weighted histograms of the various processes (weighted such that the integral equals the yield).

    Channel: SemiLeptonic(0) Process: eFake(10)
    ib= 1 0.88551 1.50073 wgt=      0.88551 wgtE=     1.50073 wgtEsum2= 2.2522
    The first number is the bin number being considered. wgt is the weighted integral of that bin, and all preceding bins (i.e. the total integral up to that point)

    Immediately following this is the record of generating the first pseudoexperiment. It lists the weighted contents of each of the bins of a neural net histogram, assuming background only, with poisson fluctuations. It then gives the integral of this pseudoexperiment:

    Pseudodata Integral: 11506
    For obvious reasons this should be similar to the projected background yield.

    Later on, at the start of the fitting we also have the following:

    = After fit ==========================================
    Parameters fit: 7
    Name        Value        Error
    =========== ============ =========
    LumiTrigLepID : 0.0140496 0.903865
    JES : 0.00324431 0.980996
    Met : 0.00131664 0.99884
    btag : 0.0197434 0.649849
    NLOAccep : -0.0627014 0.727165
    pdf : -0.0106981 0.99073
    xsec : 0.00620052 0.923289
    These values come from a Minuit fit, so should be taken with a pinch of salt. The 'Value' compares the results of the pseudoexperiment for all the various errors, and compares it to what you told it. E.g. if you said you had 1fb-1 for luminosity, but the pseudodata suggested a luminosity of 1.01, then Value would be 0.01 - you are 'out' by 1%. 'Error' says how much of your proposed error you have 'used' - if you say you have a 10% error on your luminosity, but the fit suggests at 1% error, then 'Error' would be 0.10 - you are using 10% of your 'allowed' error. These are calculated against data, seeing how much the various backgrounds are allowed to vary according to the systematics before they are no longer compatible with data.
  13. drivetestFlatFitAtlastth.rootUnscaledTemplates.root.

    The distribution of the exclusions. This is plotted only if

    GeneralParameter bool 1 PlotLikelihood =1
    in teststeerFlatReaderATLAStthSemileptonic.txt. The range and number of bins in this plot can be controlled by editing the following switches:
    GeneralParameter int    1 LikeliPseudoExpNBin=400
    GeneralParameter double 1 LikeliPseudoExpMin=0.
    GeneralParameter double 1 LikeliPseudoExpMax=10.
  14. tr[run number].o[PBS job number]

    This output file is written as soon as the job stops, and contains a summary of the full run - useful for looking for error messages.

  15. TMVAPerf_120_Run[run number]Job2.html

    A useful little html page that one of Rick's scripts creates, showing a number of useful plots - the signal and background Net scores, distributions of input variables and their correlations, and so on.

Information found in log files


After looping over all the events, you will see a table like the one below:

 Process Name                                                                       File Name File     Scale    Events  Integral   IntLumi     Alpha
         ttjj /data/atlas09/ahgemmell/NNInputFiles_v16/mergedfilesProcessed/105200-29Aug.root    0         1       613       613         1  0.907015
          ttH      /data/atlas09/ahgemmell/NNInputFiles_v16/mergedfilesProcessed/ttH-v16.root  120         1       556       556         1         1

Some of the values are established through steerComputentp.txt in the line

ListParameter  Process:ttH       1 Filename:/data/atlas09/ahgemmell/NNInputFiles_v16/mergedfilesProcessed/ttH-v16.root:File:120:IntLumi:1.0

  1. Process Name / File Name : The mapping between these two is established in steerComputentp.txt
  2. File : The number which Computentp uses to differentiate between the various files it's processing - established in steerComputentp.txt
  3. IntLumi : The luminosity which Computentp is aiming to simulate by applying 'weight' to your samples - established in steerComputentp.txt
  4. Integral : The number of events you would expect that sample to have within your desired luminosity
  5. Alpha : This should equal TrainWeight


  • It must also be run on a PBS machine because of the structure of the file (i.e. PBS commands).

  • If USEHILOSB is set to 1 then && must be appended to cut criteria, e.g. GeneralParameter string 1 Constraint=(my_failEvent&65536)==0&&

  • It would be desirable to adapt the code to be able to process different signals, e.g. lbb.

Diagnostic Run

A diagnostic run may be carried out by setting Debug=1 and NEvent=99 in teststeerFlatReaderATLAStthSemileptonic.txt. It is also advisable to cut the run time down by setting the number of training cycles to a low number (e.g. 20) in - this appears as NCycles in the TMVAsteer.txt part of the file.

TMVA Training Plots

There is a macro in the latest NNFitter version which will plot the contents of run12/TMVA2_120_3.root showing the responsiveness of the neural net as a function of the number of training cycles, in order to gauge the optimal number of cycles to use (i.e. avoid the dangers of under- or over-training). The macro must be run in a directory where a neural net run has already been carried out.

  • Type source to export the relevant parameters.

  • Enter ROOT and type .x runTrainTest.C

This will create two .eps output files, one showing the success of signal/background fitting, and the other displaying the sensitivity of the neural net to the number of training cycles used, allowing the speed of convergence to be gauged.

Running analysis & making ntuples

This is the procedure used to remake the ATLAS ntuples, using code from CERN Subversion repositories. These ntuples were then used as input for the neural net.

cd ~
mkdir tth_analysis_making_ntuples_v.13
cd tth_analysis_making_ntuples_v.13
export SVNROOT=svn+ssh://
svn co $SVNROOT/PhysicsAnalysis/HiggsPhys/HiggsAssocTop/HiggsAssocWithTopToBBbar/tags/HiggsAssocWithTopToBBbar-00-00-00-13 <nop>PhysicsAnalysis/HiggsPhys/HiggsAssocTop/HiggsAssocWithTopToBBbar
cd <nop>PhysicsAnalysis/HiggsPhys/HiggsAssocTop/HiggsAssocWithTopToBBbar/NtupleAnalysis/

In a new terminal window: cd /data/atlas07
mkdir gkirby
cd gkirby
mkdir ntuples_sensstatecutword

The script files in the <nop>NtupleAnalysis directory were then altered to output to this new directory. This is the output line for the signal ( tthhbbOptions86581430018061-nn.txt)

OUTPUT /data/atlas07/gkirby/ntuples_sensstatecutword/86581430018061-nn.root

Code changes: The tthhbbClass.cxx file was edited to include a new error code: the following lines were added to it to allow us to exclude events that we do not wish the NN to train/test with.

if (<nop>SensibleStates.size()==0) {

Then the make command was used again in this directory. The tthhbb executable was run with each of the input ("Options") text files to prepare the ntuples.

Another change was also required; the tthhbbClass.cxx file was also edited to include a fail code to allow for events not having '4 tight b-tagged jets' since this was one of the criteria used in the cut-based preselection. The following code was added to tthhbbClass.cxx:

if (<nop>BJets.size()<4) {

This was done so that the Signal to Background ratio could be increased in order for the fit to finish and provide sensible results. Requiring 4 tight b-Jets removes proportionally much more of the ttjj background than the other samples because there are fewer b-jets. The reason the S/B ratio was so low was that a problem was found which meant that the original ratio used in the 'fix_weight' variable in (up to version -06) was in fact incorrect, and when the correct weights were calculated the S/B ratio was very low indeed.

The Neural Net was then configured to exclude events where: (m_failEvent & 196608)==1
with 196608=131072+65536.

Creating plots to review the data

There is a simple shell script included in the running Neural Net code package that can produce a nice html document you can use to review a few plots of interest - To run it, move it into the run directory you want to review, then it's a simple one-line command:

./ 120 <run> <job> 

N.B. This is done automatically by genemflat currently.

Debugging the code

Before trying debugging, you should set up the environment in your terminal (when running the code normally, this is done automatically by tr${run}.job). This can be done by sourcing, which automates setting of the relevant paths - remember to specify the release number of GlaNtp that you have in your area:

source 00-00-32

At any point you can check that the a given steering file can be read by GlaNtp by using testSteerrv5.exe - found inside your GlaNtp package:

testSteerrv5.exe <file to be tested>

Another debugging script checks you have defined the processes correctly:

testFlatProcessInforv5.exe FlatAtlastthPhysicsProc1.txt

The output from the near the end of this is the important bit:

Table of what category each process is falling under
IP :       PN :       LB : B : D : S : P : O
 0 :       tt : t#bar{t} : 1 : 0 : 0 : 1 : 0

IP is something or other (need to ask Rick to remind me), PN is process name, LB is the label of the process. B, D and S are whether or not the process is Background, Data or Signal respectively. P is whether or not that process is included in the manufacture of Pseudoexperiments, and O is the order in which that process is plotted.

To debug the code further, two things need to be done - first, all the debug switches need to be turned on, and then you need to restrict the number of events to ~10 (for a Computentp run this will still manage to generate a 2 GB log file!). All of these switches are found in teststeerFlatReaderATLAStthSemileptonic.txt (the progenitor for all FlatReader files) and steerComputentp.txt (created by genemflat). The debug switches are:

GeneralParameter bool 1 Debug=0
GeneralParameter bool 1 DebugGlobalInfo=0
GeneralParameter bool 1 DebugEvInfo=0
GeneralParameter int  1 ReportInterval=100

In steerComputentp.txt there is also one additional debug option:

GeneralParameter bool   1 DebugFlatTRntp=1

All the debug switches can be set to one (I'm not sure of the exact effect of each individual switch) - the report interval can be adapted depending on how many events are present in your input files and on how large you want your log files to be. To restrict the events you use

# Loop Control
GeneralParameter int  1 NEvent=999999
GeneralParameter int  0 FirstEvent=1
GeneralParameter int  0 LastEvent=10

The easiest switch is to set NEvent=10 - however, if desired you can run over a specified range, by switching of the NEvent switch (changing it to int 0 NEvent) and switching on the other two switches, using them to specify the events you wish to run over.

Then you can run a subset of a complete run, but altering the flags found in genemflat:

# Flags to limit the scope of the run if desired

However, sometimes even this can not produce enough information., so there exist a few other options for checking your code. The first option is

runFlatReader FlatReaderATLAStthNoNN.txt /data/atlas09/ahgemmell/NNInputFiles_v16/mergedfilesProcessed/ttH-v16.root

This produces a lot of printout, so be sure to restrict the number of events as described above! An example of part of the output is

****** FlatReader Info Start ******
Entries examined : 13307
Events:00 Seen : 13307
Events:00 Seen Any : 13307
Events:01 Passing Mask Selection for Higgs : 5049

The numbers correspond to first of all the number of entries in the input MC passing into the FlatReader (e.g. in our preselection we require at least 6 jets). The second number is in reference to the number of entries passing the cutMask (e.g. requiring exactly 6 jets). This is shortly followed by

WtEvents:00 Seen : 6.17741
WtEvents:00 Seen Any : 6.17741
WtEvents:01 Passing Mask Selection for Higgs : 2.59829

These entries correspond to the yields - the numbers of events expected in our specified luminosity.

If you want to get more debugging from Computentp, then run it with another argument (doesn't matter what the argument is - in the example below it's simply 1):

Computentp steerComputentp.txt 1

Some error messages and how to fix them

Double Variable: my_NN_BJet12_M not valid and hence saved : 1

Look at VariableTreeToNTPATLASttHSemiLeptonic-v16.txt - are the names of the variables really consistent?

Various other switches of interest

In FlatReader:

GeneralParameter int 1 LikeliPseudoExpMin=0.
GeneralParameter int 1 LikeliPseudoExpMax=10.

GeneralParameter int 1 LikeliPseudoExpNBin=400

To enable you to specify the range and number of bins in the histogram showing the distribution of the pseudoexperiment exclusions. (Found in drivetestFlatFitAtlastth.rootUnscaledTemplates.root)

TMVAsteer.txt (

H6AONN5MEMLP      MLP         1 !H:!V:NCycles=1000:HiddenLayers=N+1,N:RandomSeed=9876543

If the phrase 'H6AONN5MEMLP' is changed, then this change must also be propogated to the webpage plotter (e-mail from Rick 1 Mar 2011)

  • CutWordListv9.ods: Cutmasks for tth, WH and ZH version 9 – NB this is obsolete with respect to GlaNtp tag 72! Follow bit spec in GlaNtp's steering files, NOT this spreadsheet.
Topic attachments
I Attachment History Action Size Date Who Comment
Unknown file formatods CutWordListv8.ods r1 manage 73.5 K 2012-05-03 - 10:29 AdrianBuzatu CutWordList version 8
Unknown file formatods CutWordListv9.ods r1 manage 73.8 K 2012-05-04 - 09:17 RichardStDenis Cutmasks for tth, WH and ZH version 9
Unknown file formateps Est_12_120.eps r1 manage 16.0 K 2009-07-24 - 12:06 GavinKirby  
Texttxt FlatStackParams.txt r1 manage 4.9 K 2012-03-07 - 16:06 AdrianBuzatu Description of the parameters that can be set in the plotting control.
Unknown file formateps FlatStack_1.eps r1 manage 90.1 K 2009-07-24 - 11:12 GavinKirby  
Unknown file formateps FlatStack_2.eps r1 manage 109.0 K 2009-07-24 - 12:06 GavinKirby  
PDFpdf Report-FINAL.pdf r1 manage 202.3 K 2009-09-29 - 10:47 ChrisCollins  
Unknown file formateps drivetestFlatFitAtlastth.rootSemiLeptonic_lnsb1.eps r1 manage 17.0 K 2009-07-24 - 12:06 GavinKirby  
Unknown file formateps drivetestFlatFitAtlastth.rootSemiLeptonic_lnsb2.eps r1 manage 16.1 K 2009-07-24 - 12:06 GavinKirby  
Unknown file formateps score_12_120.eps r1 manage 28.5 K 2009-07-24 - 12:06 GavinKirby  
Edit | Attach | Watch | Print version | History: r161 < r160 < r159 < r158 < r157 | Backlinks | Raw View | Raw edit | More topic actions
Topic revision: r161 - 2015-07-16 - GavinKirby
This site is powered by the TWiki collaboration platform Powered by PerlCopyright © 2008-2024 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding TWiki? Send feedback