#include <MuonPathFitter.h>

Inheritance diagram for MuonPathFitter:

Public Member Functions
fit_common_out_t	fit (fit_common_in_t fit_common_in, int XI_WIDTH, int COEFF_WIDTH_T0, int COEFF_WIDTH_POSITION, int COEFF_WIDTH_SLOPE, int PRECISSION_T0, int PRECISSION_POSITION, int PRECISSION_SLOPE, int PROD_RESIZE_T0, int PROD_RESIZE_POSITION, int PROD_RESIZE_SLOPE, int MAX_DRIFT_TDC, int sl)

int	get_rom_addr (MuonPathPtr &inMPath, latcomb lats)

bool	hasPosRF (int wh, int sec)

	MuonPathFitter (const edm::ParameterSet &pset, edm::ConsumesCollector &iC, std::shared_ptr< GlobalCoordsObtainer > &globalcoordsobtainer)

coeffs_t	RomDataConvert (std::vector< int > slv, short COEFF_WIDTH_T0, short COEFF_WIDTH_POSITION, short COEFF_WIDTH_SLOPE, short LOLY, short HILY)

void	setChi2Th (double chi2Th)

void	setTanPhiTh (double tanPhiTh)

	~MuonPathFitter () override

Public Member Functions inherited from MuonPathAnalyzer
virtual void	finish ()

virtual void	initialise (const edm::EventSetup &iEventSetup)

	MuonPathAnalyzer (const edm::ParameterSet &pset, edm::ConsumesCollector &iC)

virtual void	run (edm::Event &iEvent, const edm::EventSetup &iEventSetup, MuonPathPtrs &inMpath, std::vector< cmsdt::metaPrimitive > &metaPrimitives)=0

virtual void	run (edm::Event &iEvent, const edm::EventSetup &iEventSetup, MuonPathPtrs &inMpath, std::vector< lat_vector > &lateralities, std::vector< cmsdt::metaPrimitive > &metaPrimitives)=0

virtual void	run (edm::Event &iEvent, const edm::EventSetup &iEventSetup, std::vector< cmsdt::metaPrimitive > &inMPaths, std::vector< cmsdt::metaPrimitive > &outMPaths)=0

virtual void	run (edm::Event &iEvent, const edm::EventSetup &iEventSetup, MuonPathPtrs &inMpath, MuonPathPtrs &outMPath)=0

virtual	~MuonPathAnalyzer ()

Public Attributes
double	chi2Th_

const bool	debug_

DTGeometry const *	dtGeo_

edm::ESGetToken< DTGeometry, MuonGeometryRecord >	dtGeomH

std::shared_ptr< GlobalCoordsObtainer >	globalcoordsobtainer_

int	max_drift_tdc = -1

edm::FileInPath	maxdrift_filename_

int	maxdriftinfo_ [5][4][14]

edm::FileInPath	shift_filename_

std::map< int, float >	shiftinfo_

double	tanPhiTh_

Detailed Description

Definition at line 52 of file MuonPathFitter.h.

Constructor & Destructor Documentation

◆ MuonPathFitter()

MuonPathFitter::MuonPathFitter	(	const edm::ParameterSet &	pset,
		edm::ConsumesCollector &	iC,
		std::shared_ptr< GlobalCoordsObtainer > &	globalcoordsobtainer
	)

Definition at line 12 of file MuonPathFitter.cc.

References edm::BeginRun, debug_, dtGeomH, edm::ConsumesCollector::esConsumes(), Exception, edm::FileInPath::fullPath(), globalcoordsobtainer_, LogDebug, maxdrift_filename_, maxdriftinfo_, muonDTDigis_cfi::pset, nano_mu_digi_cff::rawId, edm::shift, shift_filename_, and shiftinfo_.

     : MuonPathAnalyzer(pset, iC), debug_(pset.getUntrackedParameter<bool>("debug")) {
   if (debug_)
     LogDebug("MuonPathFitter") << "MuonPathAnalyzer: constructor";
 
   //shift phi
   int rawId;
   shift_filename_ = pset.getParameter<edm::FileInPath>("shift_filename");
   std::ifstream ifin3(shift_filename_.fullPath());
   double shift;
   if (ifin3.fail()) {
     throw cms::Exception("Missing Input File")
         << "MuonPathFitter::MuonPathFitter() -  Cannot find " << shift_filename_.fullPath();
   }
   while (ifin3.good()) {
     ifin3 >> rawId >> shift;
     shiftinfo_[rawId] = shift;
   }
 
   int wh, st, se, maxdrift;
   maxdrift_filename_ = pset.getParameter<edm::FileInPath>("maxdrift_filename");
   std::ifstream ifind(maxdrift_filename_.fullPath());
   if (ifind.fail()) {
     throw cms::Exception("Missing Input File")
         << "MPSLFilter::MPSLFilter() -  Cannot find " << maxdrift_filename_.fullPath();
   }
   while (ifind.good()) {
     ifind >> wh >> st >> se >> maxdrift;
     maxdriftinfo_[wh][st][se] = maxdrift;
   }
 
   dtGeomH = iC.esConsumes<DTGeometry, MuonGeometryRecord, edm::Transition::BeginRun>();
   globalcoordsobtainer_ = globalcoordsobtainer;
 }

◆ ~MuonPathFitter()

MuonPathFitter::~MuonPathFitter ( )

override

Definition at line 49 of file MuonPathFitter.cc.

References debug_, and LogDebug.

                                 {
   if (debug_)
     LogDebug("MuonPathFitter") << "MuonPathAnalyzer: destructor";
 }

Member Function Documentation

◆ fit()

fit_common_out_t MuonPathFitter::fit	(	fit_common_in_t	fit_common_in,
		int	XI_WIDTH,
		int	COEFF_WIDTH_T0,
		int	COEFF_WIDTH_POSITION,
		int	COEFF_WIDTH_SLOPE,
		int	PRECISSION_T0,
		int	PRECISSION_POSITION,
		int	PRECISSION_SLOPE,
		int	PROD_RESIZE_T0,
		int	PROD_RESIZE_POSITION,
		int	PROD_RESIZE_SLOPE,
		int	MAX_DRIFT_TDC,
		int	sl
	)

Definition at line 62 of file MuonPathFitter.cc.

References funct::abs(), cmsdt::CELL_SEMILENGTH, fit_common_out_t::chi2, nano_mu_local_reco_cff::chi2, chi2Th_, fit_common_in_t::coarse_bctr, fit_common_in_t::coarse_wirepos, fit_common_in_t::coeffs, fit_common_in_t::hits, fit_common_in_t::hits_valid, mps_fire::i, createfilelist::int, fit_common_in_t::lateralities, NUM_LAYERS, coeffs_t::position, fit_common_out_t::position, position, funct::pow(), slope, coeffs_t::slope, fit_common_out_t::slope, coeffs_t::t0, fit_common_out_t::t0, FrontierCondition_GT_autoExpress_cfi::t0, cmsdt::T0_CUT_TOLERANCE, fit_common_out_t::valid_fit, vhdl_int_to_signed(), vhdl_int_to_unsigned(), vhdl_resize_signed(), vhdl_resize_signed_ok(), vhdl_resize_unsigned(), vhdl_resize_unsigned_ok(), vhdl_signed_to_int(), cmsdt::WIDTH_COARSED_TIME, cmsdt::WIDTH_DIFBX, cmsdt::WIDTH_FULL_TIME, cmsdt::WIREPOS_NORM_LSB_IGNORED, and cmsdt::WIREPOS_WIDTH.

Referenced by MuonPathSLFitter::analyze(), MuonPathCorFitter::analyze(), and trackingPlots.Iteration::modules().

                                              {
   const int PARTIALS_PRECISSION = 4;
   const int PARTIALS_SHR_T0 = PRECISSION_T0 - PARTIALS_PRECISSION;
   const int PARTIALS_SHR_POSITION = PRECISSION_POSITION - PARTIALS_PRECISSION;
   const int PARTIALS_SHR_SLOPE = PRECISSION_SLOPE - PARTIALS_PRECISSION;
   const int PARTIALS_WIDTH_T0 = PROD_RESIZE_T0 - PARTIALS_SHR_T0;
   const int PARTIALS_WIDTH_POSITION = PROD_RESIZE_POSITION - PARTIALS_SHR_POSITION;
   const int PARTIALS_WIDTH_SLOPE = PROD_RESIZE_SLOPE - PARTIALS_SHR_SLOPE;
 
   const int WIDTH_TO_PREC = 11 + PARTIALS_PRECISSION;
   const int WIDTH_SLOPE_PREC = 14 + PARTIALS_PRECISSION;
   const int WIDTH_POSITION_PREC = WIDTH_SLOPE_PREC + 1;
 
   const int SEMICHAMBER_H_PRECISSION = 13 + PARTIALS_PRECISSION;
   const float SEMICHAMBER_H_REAL = ((235. / 2.) / (16. * 6.5)) * std::pow(2, SEMICHAMBER_H_PRECISSION);
   const int SEMICHAMBER_H = (int)SEMICHAMBER_H_REAL;  // signed(SEMICHAMBER_H_WIDTH-1 downto 0)
 
   const int SEMICHAMBER_RES_SHR = SEMICHAMBER_H_PRECISSION;
 
   const int LYRANDAHALF_RES_SHR = 4;
 
   const int CHI2_CALC_RES_BITS = 7;
 
   /*******************************
             clock cycle 1
   *******************************/
   std::vector<int> normalized_times;
   std::vector<int> normalized_wirepos;
 
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     // normalized times
     // this should be resized to an unsigned of 10 bits (max drift time ~508 TDC counts, using 9+1 to include tolerance)
     // leaving it as an integer for now
     // we are obtaining the difference as the difference in BX + the LS bits from the hit time
 
     if (fit_common_in.hits_valid[i] == 1) {
       int dif_bx = (fit_common_in.hits[i].ti >> (WIDTH_FULL_TIME - WIDTH_COARSED_TIME)) - fit_common_in.coarse_bctr;
 
       int tmp_norm_time = (dif_bx << (WIDTH_FULL_TIME - WIDTH_COARSED_TIME)) +
                           (fit_common_in.hits[i].ti % (int)std::pow(2, WIDTH_FULL_TIME - WIDTH_COARSED_TIME));
       // resize test
       // this has implications in the FW (reducing number of bits).
       // we keep here the int as it is, but we do the same check done in the fw
       std::vector<int> tmp_dif_bx_vector;
       vhdl_int_to_unsigned(dif_bx, tmp_dif_bx_vector);
       vhdl_resize_unsigned(tmp_dif_bx_vector, 12);
       if (!vhdl_resize_unsigned_ok(tmp_dif_bx_vector, WIDTH_DIFBX))
         return fit_common_out_t();
 
       normalized_times.push_back(tmp_norm_time);
       int tmp_wirepos = fit_common_in.hits[i].wp - (fit_common_in.coarse_wirepos << WIREPOS_NORM_LSB_IGNORED);
       // resize test
       std::vector<int> tmp_wirepos_vector;
       vhdl_int_to_signed(tmp_wirepos, tmp_wirepos_vector);
       vhdl_resize_signed(tmp_wirepos_vector, WIREPOS_WIDTH);
       if (!vhdl_resize_signed_ok(tmp_wirepos_vector, XI_WIDTH))
         return fit_common_out_t();
 
       normalized_wirepos.push_back(tmp_wirepos);
     } else {  // dummy hit
       normalized_times.push_back(-1);
       normalized_wirepos.push_back(-1);
     }
   }
 
   /*******************************
             clock cycle 2
   *******************************/
 
   std::vector<int> xi_arr;
   // min and max times are computed throught several clk cycles in the fw,
   // here we compute it at once
   int min_hit_time = 999999, max_hit_time = 0;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 1) {
       // calculate xi array
       auto tmp_xi_incr = normalized_wirepos[i];
       tmp_xi_incr += (-1 + 2 * fit_common_in.lateralities[i]) * normalized_times[i];
 
       // resize test
       std::vector<int> tmp_xi_incr_vector;
       vhdl_int_to_signed(tmp_xi_incr, tmp_xi_incr_vector);
       vhdl_resize_signed(tmp_xi_incr_vector, XI_WIDTH + 1);
       if (!vhdl_resize_signed_ok(tmp_xi_incr_vector, XI_WIDTH))
         return fit_common_out_t();
       xi_arr.push_back(tmp_xi_incr);
 
       // calculate min and max times
       if (normalized_times[i] < min_hit_time) {
         min_hit_time = normalized_times[i];
       }
       if (normalized_times[i] > max_hit_time) {
         max_hit_time = normalized_times[i];
       }
     } else {
       xi_arr.push_back(-1);
     }
   }
 
   /*******************************
             clock cycle 3
   *******************************/
 
   std::vector<int> products_t0;
   std::vector<int> products_position;
   std::vector<int> products_slope;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 0) {
       products_t0.push_back(-1);
       products_position.push_back(-1);
       products_slope.push_back(-1);
     } else {
       products_t0.push_back(xi_arr[i] * vhdl_signed_to_int(fit_common_in.coeffs.t0[i]));
       products_position.push_back(xi_arr[i] * vhdl_signed_to_int(fit_common_in.coeffs.position[i]));
       products_slope.push_back(xi_arr[i] * vhdl_signed_to_int(fit_common_in.coeffs.slope[i]));
     }
   }
 
   /*******************************
             clock cycle 4
   *******************************/
   // Do the 8 element sums
   int t0_prec = 0, position_prec = 0, slope_prec = 0;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 0) {
       continue;
     } else {
       t0_prec += products_t0[i] >> PARTIALS_SHR_T0;
       position_prec += products_position[i] >> PARTIALS_SHR_POSITION;
       slope_prec += products_slope[i] >> PARTIALS_SHR_SLOPE;
     }
   }
 
   /*******************************
             clock cycle 5
   *******************************/
   // Do resize tests for the computed sums with full precision
   std::vector<int> t0_prec_vector, position_prec_vector, slope_prec_vector;
   vhdl_int_to_signed(t0_prec, t0_prec_vector);
 
   vhdl_resize_signed(t0_prec_vector, PARTIALS_WIDTH_T0);
   if (!vhdl_resize_signed_ok(t0_prec_vector, WIDTH_TO_PREC))
     return fit_common_out_t();
 
   vhdl_int_to_signed(position_prec, position_prec_vector);
   vhdl_resize_signed(position_prec_vector, PARTIALS_WIDTH_POSITION);
   if (!vhdl_resize_signed_ok(position_prec_vector, WIDTH_POSITION_PREC))
     return fit_common_out_t();
 
   vhdl_int_to_signed(slope_prec, slope_prec_vector);
   vhdl_resize_signed(slope_prec_vector, PARTIALS_WIDTH_SLOPE);
   if (!vhdl_resize_signed_ok(slope_prec_vector, WIDTH_SLOPE_PREC))
     return fit_common_out_t();
 
   /*******************************
             clock cycle 6
   *******************************/
   // Round the fitting parameters to the final resolution;
   // in vhdl something more sofisticated is done, here we do a float division, round
   // and cast again to integer
 
   int norm_t0 = ((t0_prec >> (PARTIALS_PRECISSION - 1)) + 1) >> 1;
   int norm_position = ((position_prec >> (PARTIALS_PRECISSION - 1)) + 1) >> 1;
   int norm_slope = ((slope_prec >> (PARTIALS_PRECISSION - 1)) + 1) >> 1;
 
   // Calculate the (-xi) + pos (+/-) t0, which only is lacking the slope term to become the residuals
   std::vector<int> res_partials_arr;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 0) {
       res_partials_arr.push_back(-1);
     } else {
       int tmp_position_prec = position_prec - (xi_arr[i] << PARTIALS_PRECISSION);
       // rounding
       tmp_position_prec += std::pow(2, PARTIALS_PRECISSION - 1);
 
       tmp_position_prec += (-1 + 2 * fit_common_in.lateralities[i]) * t0_prec;
       res_partials_arr.push_back(tmp_position_prec);
     }
   }
 
   // calculate the { slope x semichamber, slope x 1.5 layers, slope x 0.5 layers }
   // these 3 values are later combined with different signs to get the slope part
   // of the residual for each of the layers.
   int slope_x_halfchamb = (((long int)slope_prec * (long int)SEMICHAMBER_H)) >> SEMICHAMBER_RES_SHR;
   if (sl == 2)
     slope_x_halfchamb = 0;
   int slope_x_3semicells = (slope_prec * 3) >> LYRANDAHALF_RES_SHR;
   int slope_x_1semicell = (slope_prec * 1) >> LYRANDAHALF_RES_SHR;
 
   /*******************************
             clock cycle 7
   *******************************/
   // Complete the residuals calculation by constructing the slope term (1/2)
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 1) {
       if (i % 4 == 0)
         res_partials_arr[i] -= slope_x_3semicells;
       else if (i % 4 == 1)
         res_partials_arr[i] -= slope_x_1semicell;
       else if (i % 4 == 2)
         res_partials_arr[i] += slope_x_1semicell;
       else
         res_partials_arr[i] += slope_x_3semicells;
     }
   }
 
   /*******************************
             clock cycle 8
   *******************************/
   // Complete the residuals calculation by constructing the slope term (2/2)
   std::vector<int> residuals, position_prec_arr;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 0) {
       residuals.push_back(-1);
       position_prec_arr.push_back(-1);
     } else {
       int tmp_position_prec = res_partials_arr[i];
       tmp_position_prec += (-1 + 2 * (int)(i >= NUM_LAYERS)) * slope_x_halfchamb;
       position_prec_arr.push_back(tmp_position_prec);
       residuals.push_back(abs(tmp_position_prec >> PARTIALS_PRECISSION));
     }
   }
 
   // minimum and maximum fit t0
   int min_t0 = max_hit_time - MAX_DRIFT_TDC - T0_CUT_TOLERANCE;
   int max_t0 = min_hit_time + T0_CUT_TOLERANCE;
 
   /*******************************
             clock cycle 9
   *******************************/
   // Prepare addition of coarse_offset to T0 (T0 de-normalization)
   int t0_fine = norm_t0 & (int)(std::pow(2, 5) - 1);
   int t0_bx_sign = ((int)(norm_t0 < 0)) * 1;
   int t0_bx_abs = abs(norm_t0 >> 5);
 
   // De-normalize Position and slope
   int position = (fit_common_in.coarse_wirepos << WIREPOS_NORM_LSB_IGNORED) + norm_position;
   int slope = norm_slope;
 
   // Apply T0 cuts
   if (norm_t0 < min_t0)
     return fit_common_out_t();
   if (norm_t0 > max_t0)
     return fit_common_out_t();
 
   // square the residuals
   std::vector<int> squared_residuals;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 0) {
       squared_residuals.push_back(-1);
     } else {
       squared_residuals.push_back(residuals[i] * residuals[i]);
     }
   }
 
   // check for residuals overflow
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 1) {
       std::vector<int> tmp_vector;
       int tmp_position_prec = (position_prec_arr[i] >> PARTIALS_PRECISSION);
       vhdl_int_to_signed(tmp_position_prec, tmp_vector);
       vhdl_resize_signed(tmp_vector, WIDTH_POSITION_PREC);
       if (!vhdl_resize_signed_ok(tmp_vector, CHI2_CALC_RES_BITS + 1))
         return fit_common_out_t();
       // Commented for now, maybe later we need to do something here
       // if ((tmp_position_prec / (int) std::pow(2, CHI2_CALC_RES_BITS)) > 0)
       // return fit_common_out_t();
     }
   }
 
   /*******************************
         clock cycle 10, 11, 12
   *******************************/
   int t0 = t0_fine;
   t0 += (fit_common_in.coarse_bctr - (-1 + 2 * t0_bx_sign) * t0_bx_abs) * (int)std::pow(2, 5);
 
   int chi2 = 0;
   for (int i = 0; i < 2 * NUM_LAYERS; i++) {
     if (fit_common_in.hits_valid[i] == 1) {
       chi2 += squared_residuals[i];
     }
   }
 
   // Impose the thresholds
 
   if (chi2 / 16 >= (int)round(chi2Th_ * (std::pow((float)MAX_DRIFT_TDC / ((float)CELL_SEMILENGTH / 10.), 2)) / 16.))
     return fit_common_out_t();
 
   fit_common_out_t fit_common_out;
   fit_common_out.position = position;
   fit_common_out.slope = slope;
   fit_common_out.t0 = t0;
   fit_common_out.chi2 = chi2;
   fit_common_out.valid_fit = 1;
 
   return fit_common_out;
 }

◆ get_rom_addr()

int MuonPathFitter::get_rom_addr	(	MuonPathPtr &	inMPath,
		latcomb	lats
	)

◆ hasPosRF()

bool MuonPathFitter::hasPosRF	(	int	wh,
		int	sec
	)

inline

Definition at line 70 of file MuonPathFitter.h.

References fileinputsource_cfi::sec.

Referenced by MuonPathSLFitter::analyze(), and MuonPathCorFitter::analyze().

70 { return wh > 0 || (wh == 0 && sec % 4 > 1); };

fileinputsource_cfi.sec

sec

Definition: fileinputsource_cfi.py:88

◆ RomDataConvert()

coeffs_t MuonPathFitter::RomDataConvert	(	std::vector< int >	slv,
		short	COEFF_WIDTH_T0,
		short	COEFF_WIDTH_POSITION,
		short	COEFF_WIDTH_SLOPE,
		short	LOLY,
		short	HILY
	)

Definition at line 372 of file MuonPathFitter.cc.

References dqmiodatasetharvest::ctr, cmsdt::GENERIC_COEFF_WIDTH, mps_fire::i, vhdl_resize_unsigned(), and vhdl_slice().

Referenced by MuonPathSLFitter::analyze(), and MuonPathCorFitter::analyze().

                                                     {
   coeffs_t res;
   int ctr = 0;
   for (int i = LOLY; i <= HILY; i++) {
     res.t0[i] = vhdl_slice(slv, COEFF_WIDTH_T0 + ctr - 1, ctr);
     vhdl_resize_unsigned(res.t0[i], GENERIC_COEFF_WIDTH);
     res.t0[i] = vhdl_slice(res.t0[i], COEFF_WIDTH_T0 - 1, 0);
     ctr += COEFF_WIDTH_T0;
   }
   for (int i = LOLY; i <= HILY; i++) {
     res.position[i] = vhdl_slice(slv, COEFF_WIDTH_POSITION + ctr - 1, ctr);
     vhdl_resize_unsigned(res.position[i], GENERIC_COEFF_WIDTH);
     res.position[i] = vhdl_slice(res.position[i], COEFF_WIDTH_POSITION - 1, 0);
     ctr += COEFF_WIDTH_POSITION;
   }
   for (int i = LOLY; i <= HILY; i++) {
     res.slope[i] = vhdl_slice(slv, COEFF_WIDTH_SLOPE + ctr - 1, ctr);
     vhdl_resize_unsigned(res.slope[i], GENERIC_COEFF_WIDTH);
     res.slope[i] = vhdl_slice(res.slope[i], COEFF_WIDTH_SLOPE - 1, 0);
     ctr += COEFF_WIDTH_SLOPE;
   }
   return res;
 }