![]() |
Belle II Software development
|
Struct to keep neurotrigger parameters. More...
#include <NeuroTrigger.h>
Public Attributes | |
unsigned | nMLP = 1 |
Number of networks. | |
std::vector< std::vector< float > > | nHidden = {{3.}} |
Number of nodes in each hidden layer for all networks or factor to multiply with number of inputs. | |
bool | targetZ = true |
train z as output | |
bool | targetTheta = false |
train theta as output | |
bool | multiplyHidden = true |
If true, multiply nHidden with number of input nodes. | |
std::vector< std::vector< float > > | outputScale = {{ -1., 1.}} |
Output scale for all networks. | |
std::vector< std::vector< float > > | phiRange = {{0., 360.}} |
Phi region for which MLP is used in degree for all networks. | |
std::vector< std::vector< float > > | invptRange = {{ -5., 5.}} |
Charge / Pt region for which MLP is used in 1/GeV for all networks. | |
std::vector< std::vector< float > > | thetaRange = {{17., 150.}} |
Theta region for which MLP is used in degree for all networks. | |
std::vector< std::vector< float > > | phiRangeTrain = {{0., 360.}} |
Phi region for which MLP is trained in degree for all networks. | |
std::vector< std::vector< float > > | invptRangeTrain = {{ -5., 5.}} |
Charge / Pt region for which MLP is trained in 1/GeV for all networks. | |
std::vector< std::vector< float > > | thetaRangeTrain = {{17., 150.}} |
Theta region for which MLP is trained in degree for all networks. | |
std::vector< unsigned short > | maxHitsPerSL = {1} |
Maximum number of hits in a single super layer for all networks. | |
std::vector< unsigned long > | SLpattern = {0} |
Super layer pattern for which MLP is trained for all networks. | |
std::vector< unsigned long > | SLpatternMask = {0} |
Super layer pattern mask for which MLP is trained for all networks. | |
unsigned | tMax = 256 |
Maximal drift time, identical for all networks. | |
std::string | et_option = "etf_or_fastestpriority" |
Determine, how the event time should be obtained. | |
Struct to keep neurotrigger parameters.
Contains all information that is needed to initialize several expert MLPs (not including values determined during training).
Definition at line 46 of file NeuroTrigger.h.
std::string et_option = "etf_or_fastestpriority" |
Determine, how the event time should be obtained.
The options are: "etf_only" : only ETF info is used, otherwise an error is thrown. "fastestpriority" : event time is estimated by fastest priority time in selected track segments. if something fails, it is set to 0. "zero" : the event time is set to 0. "etf_or_fastestpriority" : the event time is obtained by the ETF, if not possible, the flag "fastestppriority" is used. "etf_or_zero" : the event time is obtained by the ETF, if not possible, it es set to 0 "min_etf_fastestpriority" : take whichever is smaller
Definition at line 112 of file NeuroTrigger.h.
std::vector<std::vector<float> > invptRange = {{ -5., 5.}} |
Charge / Pt region for which MLP is used in 1/GeV for all networks.
Definition at line 69 of file NeuroTrigger.h.
std::vector<std::vector<float> > invptRangeTrain = {{ -5., 5.}} |
Charge / Pt region for which MLP is trained in 1/GeV for all networks.
Can be larger than use range to avoid edge effects.
Definition at line 77 of file NeuroTrigger.h.
std::vector<unsigned short> maxHitsPerSL = {1} |
Maximum number of hits in a single super layer for all networks.
Definition at line 82 of file NeuroTrigger.h.
bool multiplyHidden = true |
If true, multiply nHidden with number of input nodes.
Definition at line 63 of file NeuroTrigger.h.
std::vector<std::vector<float> > nHidden = {{3.}} |
Number of nodes in each hidden layer for all networks or factor to multiply with number of inputs.
The number of layers is derived from the shape.
Definition at line 57 of file NeuroTrigger.h.
unsigned nMLP = 1 |
Number of networks.
For network specific parameters you can give either a list with values for each network, or a single value that will be used for all. The ranges are also valid if nPhi * nPt * nTheta * nPattern = nMLPs.
Definition at line 52 of file NeuroTrigger.h.
std::vector<std::vector<float> > outputScale = {{ -1., 1.}} |
Output scale for all networks.
Definition at line 65 of file NeuroTrigger.h.
std::vector<std::vector<float> > phiRange = {{0., 360.}} |
Phi region for which MLP is used in degree for all networks.
Definition at line 67 of file NeuroTrigger.h.
std::vector<std::vector<float> > phiRangeTrain = {{0., 360.}} |
Phi region for which MLP is trained in degree for all networks.
Can be larger than use range to avoid edge effects.
Definition at line 74 of file NeuroTrigger.h.
std::vector<unsigned long> SLpattern = {0} |
Super layer pattern for which MLP is trained for all networks.
Binary pattern of 9 * maxHitsPerSL bits (on/off for each hit). 0 in bit <i>: hits from super layer <i> are not used. 1 in bit <i>: hits from super layer <i> are used. SLpattern = 0: use any hits present, don't check the pattern.
Definition at line 88 of file NeuroTrigger.h.
std::vector<unsigned long> SLpatternMask = {0} |
Super layer pattern mask for which MLP is trained for all networks.
Binary pattern of 9 * maxHitsPerSL bits (on/off for each hit). 0 in bit <i>: super layer <i> may or may not have a hit. 1 in bit <i>: super layer <i>
Definition at line 95 of file NeuroTrigger.h.
bool targetTheta = false |
train theta as output
Definition at line 61 of file NeuroTrigger.h.
bool targetZ = true |
train z as output
Definition at line 59 of file NeuroTrigger.h.
std::vector<std::vector<float> > thetaRange = {{17., 150.}} |
Theta region for which MLP is used in degree for all networks.
Definition at line 71 of file NeuroTrigger.h.
std::vector<std::vector<float> > thetaRangeTrain = {{17., 150.}} |
Theta region for which MLP is trained in degree for all networks.
Can be larger than use range to avoid edge effects.
Definition at line 80 of file NeuroTrigger.h.
unsigned tMax = 256 |
Maximal drift time, identical for all networks.
Definition at line 97 of file NeuroTrigger.h.