CA Configuration
You can configure the running of Component Accumulator based code using "blocks".
Each block is a Python class that holds configuration keys as constructor arguments, and implements a to_ca()
function that returns a ComponentAccumulator
object.
Blocks are defined in the FTagDumper/python/blocks/
directory, and inherit from the BaseBlock
class.
Blocks are configured inside an optional "ca_blocks": []
list in the top-level JSON configuration.
Each block is a dict with a "block"
key that specifies the block name, and any other keys that are passed to the block constructor.
This page contains automatically generated documentation for the different CA blocks that can be configured with the dumpster.
BTagJetLinker
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Matches a jet collection to a b-tagging collection.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
jet_collection |
str
|
The name of the jet collection to match. If not provided, it will be taken from the dumper config. |
None
|
old_link |
str
|
The name of the old link to match. If not provided, it will be set to |
None
|
new_link |
str
|
The name of the new link to match. If not provided, it will be set to |
None
|
Source code in FTagDumper/python/blocks/BTagJetLinker.py
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 |
|
FixedConeAssociation
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Run a fixed cone track association. The fixed cone association is used for the training of the nntc model.
To run the nntc, you can use the following configuration: EMPFlow_fixedcone.json. The size of the cone is parametrized by the fixedConeRadius parameter.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
radius |
float
|
The radius of the fixed cone. |
0.5
|
Source code in FTagDumper/python/blocks/FixedConeAssociation.py
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 |
|
FoldHashDecorator
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Decorate jets with a hash for kfold training and evaluation.
See FoldDecoratorAlg.cxx and FoldDecoratorConfig.py
Parameters:
Name | Type | Description | Default |
---|---|---|---|
jet_collection |
str
|
Name of the jet collection to decorate. Use the dumper's jet_collection if not provided. |
None
|
prefix |
str
|
Prefix to add to the hash. |
''
|
Source code in FTagDumper/python/blocks/FoldHashDecorator.py
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 |
|
GeneratorWeights
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Add truth weights to EventInfo
Source code in FTagDumper/python/blocks/GeneratorWeights.py
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 |
|
JetMatcher
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Matches jets from a single source collection to jets in a target collection. For more details, see JetMatcherAlg.cxx
Parameters:
Name | Type | Description | Default |
---|---|---|---|
source_jets |
str
|
The name of the source jet collection to match. |
required |
source_name |
str
|
The name of a singular jet in the source collection. Used for generating the variables
|
None
|
target_collection |
str
|
The name of the target jet collection to match to. If not provided, it will be taken from the dumper config. |
None
|
floats_to_copy |
list[str]
|
List of float variables to copy from the source jets to the target jets |
None
|
ints_to_copy |
list[str]
|
List of int variables to copy from the source jets to the target jets |
None
|
pt_priority_with_delta_r |
float
|
The priority of the pt variable when matching jets based on deltaR. Disabled by default. |
-1
|
Source code in FTagDumper/python/blocks/JetMatcher.py
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 |
|
MultifoldTagger
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Run a multifold tagger.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
nn_paths |
list[str]
|
List of paths to the neural network files. |
required |
target |
str
|
Whether to tag the BTagging object or the Jet. |
'BTagging'
|
jet_collection |
str | None
|
Name of the jet collection to decorate. If None, uses the jet collection from the dumper configuration. |
None
|
remap |
dict
|
Remap input and output variable names. |
None
|
decorate_tracks |
bool
|
Whether to decorate the tracks objects directly with the aux outputs. |
True
|
fold_hash_name |
str
|
Name of the fold hash variable. |
'jetFoldHash'
|
constituents |
str
|
Name of the constituent container. |
'InDetTrackParticles'
|
Source code in FTagDumper/python/blocks/MultifoldTagger.py
10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 |
|
ShrinkingConeAssociation
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Associate tracks to jets using the shrinking cone algorithm.
More info: https://ftag.docs.cern.ch/algorithms/taggers/inputs/
Source code in FTagDumper/python/blocks/ShrinkingConeAssociation.py
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 |
|
Trackless
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Source code in FTagDumper/python/blocks/Trackless.py
9 10 11 12 13 14 15 16 17 |
|
TruthLabelling
dataclass
#
Bases: FTagDumper.python.blocks.BaseBlock.BaseBlock
Re-run the FTAG truth labelling.
Detailed FTAG truth variables, such as ftagTruthOriginLabel
and ftagTruthVertexIndex
are normally produced when running derivations. With this block you can reproduce these
labels as part of your dumping job, for example if the labelling code has been updated
and you don't want to run expensive derivations.
For more info see TruthParticleDecoratorAlg
and TrackTruthDecoratorAlg
in Athena,
and this page in the
FTAG docs.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
suffix |
str
|
Suffix to append to the truth labelling variables. |
required |
Examples:
{
"ca_blocks": [{"block": "TruthLabelling", "suffix": "tdd"}]
}
Remember to add the new ftagTruthOriginLabelTDD
etc variables to your output variables,
or alternatively if you want to avoid the presence of new variable names in your h5 you
can remap variable names as follows:
"tracks": [
{
...
"edm_names": {
"ftagTruthOriginLabel": "ftagTruthOriginLabelTDD",
"ftagTruthVertexIndex": "ftagTruthVertexIndexTDD"
}
}
]
Source code in FTagDumper/python/blocks/TruthLabelling.py
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 |
|