|
|
| (6 intermediate revisions by one user not shown) |
| Line 1: |
Line 1: |
| − | = Introduction =
| |
| | | | |
| − | The AuroraMaster package contains python classes providing high level interfaces to the Aurora algorithms and tools. First thing you need to do is instantiate an AuroraMaster object: | + | The AuroraMaster analysis configuration system used in Aurora version 2.x has been removed in the development version (branch [https://git.inp.nsk.su/sctau/aurora/-/tree/dev-gcf1?ref_type=heads dev-gcf1]). |
| − | | + | The analysis is now configured by creation of various analysis tools using usual Python functions and joining them into analysis-tool chain of AnalysisAlgorithm. |
| − | from AuroraMaster.auroramaster import AuroraMaster
| + | Please see the page [[Use_Analysis_package]] for details. |
| − | am = AuroraMaster(purpose='parsim', olvl='info')
| + | |
| − | | + | |
| − | The first argument specifies purpose of the job option. Possible values are:
| + | |
| − | | + | |
| − | * <code>'parsim'</code>
| + | |
| − | * <code>'fullsim'</code>
| + | |
| − | * <code>'evtgen'</code>
| + | |
| − | * <code>'analysis'</code>
| + | |
| − | | + | |
| − | An AuroraMaster object initializes Aurora services corresponding to the job option purpose. The second argument specifies general output level:
| + | |
| − | | + | |
| − | * <code>'debug'</code>
| + | |
| − | * <code>'info'</code>
| + | |
| − | | + | |
| − | A job option must contain only one AuroraMaster object. A job option logic is formed by stacking the predefined components. Each component has corresponding method in the AuroraMaster class. The following example shows a ready-to-use job option for event generation with [https://evtgen.hepforge.org/ EvtGen] and saving them to file in SCT EDM format:
| + | |
| − | | + | |
| − | from AuroraMaster.auroramaster import AuroraMaster, AuroraConfig
| + | |
| − | # Instantiate AuroraMaster
| + | |
| − | am = AuroraMaster('evtgen', 'info')
| + | |
| − | # Plug in component for EvtGen
| + | |
| − | evtgenCfg = AuroraConfig({
| + | |
| − | 'root' : 'psi(3770)',
| + | |
| − | 'dec': './dkpi.dec'
| + | |
| − | })
| + | |
| − | am.add_evtgen(cfg=evtgenCfg)
| + | |
| − | # Plug in component for SCT EDM output
| + | |
| − | edmoutputCfg = AuroraConfig{
| + | |
| − | 'filename': 'parsim.root',
| + | |
| − | 'commands': ['keep *'],
| + | |
| − | })
| + | |
| − | am.add_edmo(cfg=edmoutputCfg)
| + | |
| − | am.run(evtmax=10**4)
| + | |
| − | | + | |
| − | | + | |
| − | The <code>run</code> method should be invoked at the end.
| + | |
| − | | + | |
| − | = Components =
| + | |
| − | | + | |
| − | An AuroraMaster class method <code>add_{component}</code> receives two parameters:
| + | |
| − | | + | |
| − | * <code>cfg</code> - an object of the <code>AuroraConfig</code> class. Default value it None
| + | |
| − | * <code>json</code> - string path to a json file with configuration. Default value it None
| + | |
| − | | + | |
| − | A component set up is dome with three steps:
| + | |
| − | | + | |
| − | # Default configuration
| + | |
| − | # Configuration with passed json file. It overwrites any subset of default parameters. Parameters not specified in json keep the default values
| + | |
| − | # Configuration with <code>AuroraConfig</code> object. It overrides values of the specified parameters leaving other parameters unchanged
| + | |
| − | | + | |
| − | If some parameter is specified in both json file and <code>AuroraConfig</code> object, the final value is taken from the <code>AuroraConfig</code> object.
| + | |
| − | | + | |
| − | = AuroraConfig =
| + | |
| − | | + | |
| − | The AuroraConfig is a data structure very similar to python dict. It can contain nested lists, dicts and other AuroraConfig objects. An example below shows configuration of simple parametric simulation:
| + | |
| − | | + | |
| − | parsimCfg = AuroraConfig({
| + | |
| − | 'Tracker' : {
| + | |
| − | 'deteff': 0.99,
| + | |
| − | 'ptcut': 50e-3,
| + | |
| − | 'bfield': 1.5,
| + | |
| − | 'maxCosth': np.cos(10./180. * np.pi),
| + | |
| − | 'momentumSampler' : {
| + | |
| − | 'mean' : np.zeros(3),
| + | |
| − | 'covar': np.diag(np.ones(3)) * 1.e-3**2
| + | |
| − | },
| + | |
| − | 'vertexSampler' : {
| + | |
| − | 'mean' : np.zeros(3),
| + | |
| − | 'covar': np.diag(np.ones(3)) * 1.e-3**2
| + | |
| − | }
| + | |
| − | },
| + | |
| − | 'PID' : {
| + | |
| − | 'eff' : 0.95,
| + | |
| − | 'sigmaKpi' : 6.,
| + | |
| − | 'sigmaMupi' : 4.,
| + | |
| − | 'sigmaKp' : 3.,
| + | |
| − | 'sigmaE' : 3.,
| + | |
| − | },
| + | |
| − | 'Calorimeter' : {
| + | |
| − | 'deteff': 1.0,
| + | |
| − | 'energyThreshold' : 15e-3,
| + | |
| − | 'maxCosth' : np.cos(10./180. * np.pi),
| + | |
| − | 'sampler' : {
| + | |
| − | 'mean' : np.zeros(3),
| + | |
| − | 'covar': (np.diag(np.ones(3)) * 1.e-2**2).ravel()
| + | |
| − | }
| + | |
| − | }
| + | |
| − | })
| + | |
| − | | + | |
| − | An AuroraConfig object can be serialized to and serialized from json with methods <code>to_json</code> and <code>from_json</code>. It is recommended to create json files with configuration using this interface, and not create json manualy.
| + | |
The AuroraMaster analysis configuration system used in Aurora version 2.x has been removed in the development version (branch dev-gcf1).
The analysis is now configured by creation of various analysis tools using usual Python functions and joining them into analysis-tool chain of AnalysisAlgorithm.
Please see the page Use_Analysis_package for details.
