New features for ALERT Kalman filter+optimized parameters #445
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
* adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5;
* residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java
* affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction;
- removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts.
* added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function
be able to initialize a vertex.
"hit_beam" vertex.
efuchey
changed the title
access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign.
…tion and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign.
…with hit vector and measurement functions that handle hit left/right disambiguation.
…effersonLab-development Included new addition to alert.json with atof hits + residuals for ahdc
…cks on the wrong side.
* if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi;
…e by setting the measurement on the correct "virtual wire" with a larger error.
…of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715.
…t +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively.
…is 0 or if sign is good.
…ent matrix) function with goodsign as an input.
- one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter.
mathieuouillon
previously approved these changes
Mar 27, 2025
Collaborator
mathieuouillon
left a comment
mathieuouillon
left a comment
There was a problem hiding this comment.
Look good for me.
baltzell
approved these changes
Mar 27, 2025
N-Plx
added a commit
that referenced
this pull request
Apr 1, 2025
* DC v2: includes TFLight fix and new dc beta time walk with constants from /calibration/dc/v2 (#494) * fix issue for path length in FD tracking * fix issue for coordinate transformation from global to tilted sector coordinate * add comments to explain transition of starting point for path length * T2D with beta dependence. Fixes in interpolation. * Fix for instances where the t2d function turns over. * Fix in last interpolation step. More validation plots. * Fix table filling * B-field interpolation test interpolation in B instead of B^2. More plots * Added modularity for calibration use * version bump * switch to use new dc/v2 tables * switch to use new dc/v2 for ref_pressure too --------- Co-authored-by: tongtongcao <[email protected]> Co-authored-by: ziegler <ziegler@enpziegler-mac> * Update DC reconstruction at the denoisinng level (#499) * cancel SNR and change limit of total DC hits from raw hits to hits after denoising * remove unused codes * reset limit for prob in DC clustering and cancel requirement that no skipped layer for cluster candidates from splitter (#500) * pass 2 additional values * fix data type * feat: pre-commit hook to update banks README (#505) * Change the option to select the track finding to a enum. Now this option can be setup from the yaml file with the option Mode: ``` ALERT: Mode: "AI_Track_Finding" ``` Add a cut on the number of hits to rely on the conventional track finding with there is more than 300 hits. * Change the default options for DJL: - Use a single thread for the inference. - Remove waring about the graph executor optimization * Modification of the track candidates generation * Improve the function to get all the track candidates. * Update AHDCEngine to initialize model conditionally and adjust prediction threshold * New features for ALERT Kalman filter+optimized parameters (#445) * Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]> * Petiroc timestamps2 (#512) * Added petiroc board timestamps * Adding CI stuff that is for gitlab modified: common-tools/clas-detector/src/main/java/org/jlab/detector/decode/CLASDecoder4.java modified: common-tools/clas-detector/src/main/java/org/jlab/detector/decode/CodaEventDecoder.java modified: common-tools/clas-detector/src/main/java/org/jlab/detector/decode/DetectorDataDgtz.java modified: etc/bankdefs/hipo4/data.json --------- Co-authored-by: Whitney Armstrong <[email protected]> * bump version * version bump * create a method to decode evio events in a single line to simplify usage in mon12 and ced * new DCRB firmware (#509) * implemented decoding of new DCRB banks with ToT and order labeling of DC hit to use the first in time * removed anused code * printingg exception from DC banks decoding --------- Co-authored-by: Nathan Baltzell <[email protected]> * fix: remove `jdtls` files * fix: ignore `jdtls` files * Use pulse time and hide MVTFitter * Change samplingTime * Fix issue * version bump --------- Co-authored-by: raffaelladevita <[email protected]> Co-authored-by: tongtongcao <[email protected]> Co-authored-by: ziegler <ziegler@enpziegler-mac> Co-authored-by: baltzell <[email protected]> Co-authored-by: Christopher Dilks <[email protected]> Co-authored-by: MathieuOuillon <[email protected]> Co-authored-by: efuchey <[email protected]> Co-authored-by: Mathieu Ouillon <[email protected]> Co-authored-by: Whitney Armstrong <[email protected]> Co-authored-by: Whitney Armstrong <[email protected]> Co-authored-by: Felix Touchte Codjo <[email protected]> Co-authored-by: Nathan Baltzell <[email protected]>
c-dilks
pushed a commit
that referenced
this pull request
Sep 15, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
tongtongcao
pushed a commit
that referenced
this pull request
Sep 22, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
tongtongcao
pushed a commit
that referenced
this pull request
Nov 18, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
tongtongcao
pushed a commit
that referenced
this pull request
Nov 21, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
tongtongcao
pushed a commit
that referenced
this pull request
Dec 5, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
tongtongcao
pushed a commit
that referenced
this pull request
Dec 5, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
tongtongcao
pushed a commit
that referenced
this pull request
Dec 31, 2025
* Optimization of Kalman Filter: * adjusted number of filtering iterations from 10 to 5; * adjusted step size dx for calculation of ddoca/dx from 10^8 to 10^5; * Added AHDC hits residuals (post-fit and pre-fit) in the output: * residuals in the AHDC::Hits list in the alert.json file; * filling the hits residuals in the RecoBankWriter; * added residual and residual_prefit in ahdc/Hit/Hit.java * added a identification flag to match ahdc/KalmanFilter/Hit.java to ahdc/Hit/Hit.java * Successfully affected the calculated hit residual to the correct AHDC::Hit. * Fixed and improved the calculation of the post-fit residuals: * affecting the track parameters to the KFTrack right after the fit; * redo a forward indicators pass without correction; * * Cleaning the Kalman filter code: - removed all "cylindrical coordinates" vector and measurement functions; - renamed all preexisting vector and measurement function with their original name. - removed many commented printouts. * Attempt to include hit "sign" / left-right disambiguation: * added "virtual wires" located at the distance-of-closest-approach of the actual wire, on each side of the wire; * added hit sign parameter in KalmanFilter/Hit class; * added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign; * attempt to modify the "h" function to call new distance function * Fix of a parameter modified by mistake. * Added a second definition of BackwardIndicators in AHDC/KalmanFilter to be able to initialize a vertex. * Added a flag setDefinedVertex to AHDC/KalmanFilter and KFitter to define "hit_beam" vertex. * Reset Niter and ddoca step size parameters to 10, 1.e-8 respectively. * Added reading of wire ADC from the AHDC HitReader, and functions to access ADC for AHDC/Hit/Hit and AHDC/KalmanFilter/Hit. Added filtering of two hits on same superlayer/layer based on ADC (largest ADC is kept) and use info to determine the hit sign. * Added an option to build the initial track with just the hits combination and preset fixed parameters in AHDCEngine. Added a function in AHDC/KalmanFilter/Hit.java to calculate the measurement vector if we have a sign. * Substituted call of default hit vector and hit measurement functions with hit vector and measurement functions that handle hit left/right disambiguation. * Started to reintroduce the hit sign. * Save state: back to status quo before revising sign. * Added variable measurement error for hits with sign defined, with tracks on the wrong side. * Implemented varaible measurement error for signed hits: * if track on right side of wire, normal error; * if track on wrong side of wire, inflated error; Ensured reordering of hits by increasing phi; added exception for "rollover" around phi = pi; * fixed once and for all the convention for hit sign: sign >0 if phi_expected state > phi_wire * Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error. * Revert "Tried to introduce a "pull" to the track on the correct sign of a wire by setting the measurement on the correct "virtual wire" with a larger error." This reverts commit 9bf4715. * Fixed the convention for the "virtual wires": wire "minus" ("plus") at +deltaphi (-deltaphi) since wire x, y position depend on -R*sin(phi), -R cos(phi) respectively. * Improved the functions to calculate hit vector: returns doca if sign is 0 or if sign is good. * Added a hit distance function with goodsign as input, and H (measurement matrix) function with goodsign as an input. * Added (commented) calls of functions with sign. * Added a simple handle to disable reading of MC variables. * Rerolled to fitting with no double hit. * Harmonized simulation flag: - one simulation flag is declared in AHDCengine and defined as false; - it is now propagated into KalmanFilter. * Added a check to read MC hits in AHDC_engine. --------- Co-authored-by: Mathieu Ouillon <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Optimized number of iterations and step size for numerical calculation of ddoca/dx:
https://clasweb.jlab.org/wiki/images/7/7b/Alert_progress_20250123.pdf
Redid optimization after introduction of AI track finder;
Added AHDC hit residuals (post-fit and pre-fit) in AHDC hit output:
added a new distance function to KalmanFilter/Hit class calculate the distance of a point to the correct virtual wire depending on the sign;