1 |
< |
/* A simple jet-finding analyzer */ |
1 |
> |
/* A JetFitAnalyzer that makes histograms with smearing */ |
2 |
|
|
3 |
|
#include "UserCode/JetFitAnalyzer/interface/JetFitAnalyzer.h" |
4 |
|
|
5 |
– |
#include "fastjet/ClusterSequence.hh" |
5 |
|
#include "FWCore/ServiceRegistry/interface/Service.h" |
6 |
|
#include "FWCore/MessageLogger/interface/MessageLogger.h" |
7 |
|
#include "PhysicsTools/UtilAlgos/interface/TFileService.h" |
8 |
|
|
9 |
< |
#include "SimDataFormats/HepMCProduct/interface/HepMCProduct.h" |
11 |
< |
#include "DataFormats/Candidate/interface/Particle.h" |
12 |
< |
#include "DataFormats/ParticleFlowCandidate/interface/PFCandidateFwd.h" |
13 |
< |
#include "DataFormats/ParticleFlowCandidate/interface/PFCandidate.h" |
14 |
< |
#include "DataFormats/ParticleFlowReco/interface/PFBlock.h" |
15 |
< |
|
16 |
< |
#include <map> |
17 |
< |
#include <vector> |
18 |
< |
#include <limits> |
19 |
< |
#include <cmath> |
20 |
< |
#include <cstdlib> |
21 |
< |
#include <fstream> |
9 |
> |
#include <iostream> |
10 |
|
#include <sstream> |
11 |
|
|
24 |
– |
#include "TFormula.h" |
12 |
|
#include "TF2.h" |
13 |
+ |
#include "TNtuple.h" |
14 |
|
|
15 |
|
#define PI 3.141593 |
16 |
|
|
17 |
|
using namespace std; |
30 |
– |
using namespace fastjet; |
31 |
– |
|
32 |
– |
// Class to represent a "generic" particle, whether raw or reconstructed |
33 |
– |
class GenericParticle { |
34 |
– |
public: |
35 |
– |
GenericParticle(double __px, double __py, double __pz, double __e, |
36 |
– |
double __charge = 0.0) |
37 |
– |
: _px(__px), _py(__py), _pz(__pz), _e(__e), _charge(__charge) { |
38 |
– |
} |
39 |
– |
GenericParticle(const HepMC::GenParticle &genParticle) |
40 |
– |
: _charge(0.0) { |
41 |
– |
_px = genParticle.momentum().px(); |
42 |
– |
_py = genParticle.momentum().py(); |
43 |
– |
_pz = genParticle.momentum().pz(); |
44 |
– |
_e = genParticle.momentum().e(); |
45 |
– |
} |
46 |
– |
GenericParticle(const reco::PFCandidate &pfCandidate) |
47 |
– |
: _charge(0.0) { |
48 |
– |
_px = pfCandidate.px(); |
49 |
– |
_py = pfCandidate.py(); |
50 |
– |
_pz = pfCandidate.pz(); |
51 |
– |
_e = pfCandidate.energy(); |
52 |
– |
} |
53 |
– |
double px() { |
54 |
– |
return _px; |
55 |
– |
} |
56 |
– |
double py() { |
57 |
– |
return _py; |
58 |
– |
} |
59 |
– |
double pz() { |
60 |
– |
return _pz; |
61 |
– |
} |
62 |
– |
double e() { |
63 |
– |
return _e; |
64 |
– |
} |
65 |
– |
double charge() { |
66 |
– |
return _charge; |
67 |
– |
} |
68 |
– |
double eta() { |
69 |
– |
double theta = acos(_pz/sqrt(_px*_px + _py*_py + _pz*_pz)); |
70 |
– |
return -log(tan(theta*0.5)); |
71 |
– |
} |
72 |
– |
double phi() { |
73 |
– |
double phi0 = acos(_px/sqrt(_px*_px + _py*_py)); |
74 |
– |
return _py > 0.0 ? phi0 : -phi0; |
75 |
– |
} |
76 |
– |
double m() { |
77 |
– |
return sqrt(_e*_e - _px*_px - _py*_py - _pz*_pz); |
78 |
– |
} |
79 |
– |
private: |
80 |
– |
double _px; |
81 |
– |
double _py; |
82 |
– |
double _pz; |
83 |
– |
double _e; |
84 |
– |
double _charge; |
85 |
– |
}; |
18 |
|
|
19 |
|
class JetFinderAnalyzer : public JetFitAnalyzer { |
20 |
|
public: |
89 |
– |
struct jet { |
90 |
– |
double energy; |
91 |
– |
double eta; |
92 |
– |
double phi; |
93 |
– |
}; |
94 |
– |
|
21 |
|
explicit JetFinderAnalyzer( const edm::ParameterSet&); |
22 |
|
~JetFinderAnalyzer() {} |
23 |
|
|
24 |
|
private: |
25 |
< |
static map<TH2 *, vector< vector<jet> > > unique_jets; |
26 |
< |
|
27 |
< |
static double phi_cutoff_; |
102 |
< |
|
103 |
< |
static double g2int(double xlo, double xhi, double ylo, double yhi, |
104 |
< |
double *pval) { |
105 |
< |
double sum1 = 0.0; |
106 |
< |
double sum2 = 0.0; |
107 |
< |
double xmid = 0.5 * (xlo + xhi); |
108 |
< |
double ymid = 0.5 * (ylo + yhi); |
109 |
< |
double xstep = (xhi - xlo) / 50.0; |
110 |
< |
double ystep = (yhi - ylo) / 50.0; |
111 |
< |
for (int i = 0; i < 50; i++) { |
112 |
< |
double x = (static_cast<double>(i) + 0.5) * xstep + xlo; |
113 |
< |
sum1 += xstep * jetfit::fit_fcn(x, ymid, pval); |
114 |
< |
} |
115 |
< |
for (int i = 0; i < 50; i++) { |
116 |
< |
double y = (static_cast<double>(i) + 0.5) * ystep + ylo; |
117 |
< |
sum2 += ystep * jetfit::fit_fcn(xmid, y, pval); |
118 |
< |
} |
119 |
< |
return sum1 * sum2; |
120 |
< |
} |
121 |
< |
|
122 |
< |
static void jetfinder(TMinuit *gMinuit, TH2 *hist, int ngauss) { |
123 |
< |
double dist_sq = numeric_limits<double>::infinity(); |
124 |
< |
unique_jets[hist].resize(ngauss); |
125 |
< |
int nbinsX = hist->GetXaxis()->GetNbins(); |
126 |
< |
int nbinsY = hist->GetYaxis()->GetNbins(); |
127 |
< |
double XbinSize = (hist->GetXaxis()->GetXmax() |
128 |
< |
- hist->GetXaxis()->GetXmin()) |
129 |
< |
/ static_cast<double>(nbinsX); |
130 |
< |
double YbinSize = (hist->GetYaxis()->GetXmax() |
131 |
< |
- hist->GetYaxis()->GetXmin()) |
132 |
< |
/ static_cast<double>(nbinsY); |
133 |
< |
for (int i = 0; i < ngauss; i++) { |
134 |
< |
double N, mu_x, mu_y, sig, err, lo, hi; |
135 |
< |
int iuint; |
136 |
< |
TString name; |
137 |
< |
gMinuit->mnpout(4*i, name, N, err, lo, hi, iuint); |
138 |
< |
gMinuit->mnpout(4*i + 1, name, mu_x, err, lo, hi, iuint); |
139 |
< |
gMinuit->mnpout(4*i + 2, name, mu_y, err, lo, hi, iuint); |
140 |
< |
gMinuit->mnpout(4*i + 3, name, sig, err, lo, hi, iuint); |
141 |
< |
for (int j = 0; j < i; j++) { |
142 |
< |
double N2, mu_x2, mu_y2, sig2; |
143 |
< |
gMinuit->mnpout(4*j, name, N2, err, lo, hi, iuint); |
144 |
< |
gMinuit->mnpout(4*j + 1, name, mu_x2, err, lo, hi, iuint); |
145 |
< |
gMinuit->mnpout(4*j + 2, name, mu_y2, err, lo, hi, iuint); |
146 |
< |
gMinuit->mnpout(4*j + 3, name, sig2, err, lo, hi, iuint); |
147 |
< |
double _dist_sq = (mu_x2 - mu_x)*(mu_x2 - mu_x) |
148 |
< |
+ (mu_y2 - mu_y)*(mu_y2 - mu_y); |
149 |
< |
if (_dist_sq < dist_sq) |
150 |
< |
dist_sq = _dist_sq; |
151 |
< |
} |
152 |
< |
|
153 |
< |
jet j; |
154 |
< |
j.energy = N; |
155 |
< |
j.eta = mu_x; j.phi = mu_y; |
156 |
< |
unique_jets[hist][ngauss-1].push_back(j); |
157 |
< |
} |
158 |
< |
} |
159 |
< |
|
160 |
< |
virtual void beginJob(const edm::EventSetup&); |
161 |
< |
virtual TH2D* make_histo(const edm::Event&, const edm::EventSetup&); |
162 |
< |
virtual jetfit::model_def& make_model_def(const edm::Event&, |
163 |
< |
const edm::EventSetup&, |
164 |
< |
TH2 *); |
165 |
< |
virtual void analyze_results(jetfit::results r, |
166 |
< |
std::vector<jetfit::trouble> t, TH2 *); |
167 |
< |
vector<GenericParticle *> get_particles(const edm::Event&); |
168 |
< |
void fetchCandidateCollection(edm::Handle<reco::PFCandidateCollection>&, |
169 |
< |
const edm::InputTag&, const edm::Event&) const; |
170 |
< |
|
171 |
< |
fstream ofs; |
172 |
< |
edm::InputTag inputTagPFCandidates_; |
173 |
< |
int info_type_; |
174 |
< |
double smear_; |
175 |
< |
int smear_coord_; |
25 |
> |
virtual void beginJob(const edm::EventSetup &es); |
26 |
> |
virtual void analyze_results(HistoFitter::FitResults, std::vector<HistoFitter::Trouble>, |
27 |
> |
TH2 *); |
28 |
|
}; |
29 |
|
|
178 |
– |
map<TH2 *, vector< vector< JetFinderAnalyzer::jet > > > |
179 |
– |
JetFinderAnalyzer::unique_jets; |
180 |
– |
|
30 |
|
JetFinderAnalyzer::JetFinderAnalyzer(const edm::ParameterSet &pSet) |
31 |
|
: JetFitAnalyzer(pSet) // this is important! |
32 |
|
{ |
184 |
– |
info_type_ = pSet.getUntrackedParameter("info_type", 0); |
185 |
– |
|
186 |
– |
if (info_type_ == 1) { |
187 |
– |
inputTagPFCandidates_ = pSet.getParameter<edm::InputTag>("PFCandidates"); |
188 |
– |
} |
189 |
– |
|
190 |
– |
smear_ = pSet.getUntrackedParameter("smear", 0.02); |
191 |
– |
smear_coord_ = pSet.getUntrackedParameter("smear_coord", 0); |
192 |
– |
// 0 = eta-phi smear |
193 |
– |
// 1 = proper angle smear |
194 |
– |
set_user_minuit(jetfinder); |
195 |
– |
} |
196 |
– |
|
197 |
– |
void |
198 |
– |
JetFinderAnalyzer::fetchCandidateCollection(edm::Handle<reco::PFCandidateCollection>& c, |
199 |
– |
const edm::InputTag& tag, |
200 |
– |
const edm::Event& iEvent) const { |
201 |
– |
|
202 |
– |
bool found = iEvent.getByLabel(tag, c); |
203 |
– |
|
204 |
– |
if(!found ) { |
205 |
– |
ostringstream err; |
206 |
– |
err<<" cannot get PFCandidates: " |
207 |
– |
<<tag<<endl; |
208 |
– |
edm::LogError("PFCandidates")<<err.str(); |
209 |
– |
throw cms::Exception( "MissingProduct", err.str()); |
210 |
– |
} |
211 |
– |
|
212 |
– |
} |
213 |
– |
|
214 |
– |
vector<GenericParticle *> JetFinderAnalyzer::get_particles(const edm::Event &evt) { |
215 |
– |
// fill unreduced histo |
216 |
– |
edm::Handle<edm::HepMCProduct> hRaw; |
217 |
– |
edm::Handle<reco::PFCandidateCollection> hPFlow; |
218 |
– |
if (info_type_ == 0) { |
219 |
– |
evt.getByLabel("source", hRaw); |
220 |
– |
} |
221 |
– |
if (info_type_ == 1) { |
222 |
– |
fetchCandidateCollection(hPFlow, |
223 |
– |
inputTagPFCandidates_, |
224 |
– |
evt); |
225 |
– |
} |
226 |
– |
|
227 |
– |
vector<GenericParticle *> particles; |
228 |
– |
|
229 |
– |
switch (info_type_) { |
230 |
– |
case 0: |
231 |
– |
const HepMC::GenEvent *hmcEvt = hRaw->GetEvent(); |
232 |
– |
for (HepMC::GenEvent::particle_const_iterator |
233 |
– |
pit = hmcEvt->particles_begin(); pit != hmcEvt->particles_end(); |
234 |
– |
pit++) { |
235 |
– |
if ((*pit)->status() == 1) { |
236 |
– |
particles.push_back(new GenericParticle(**pit)); |
237 |
– |
} |
238 |
– |
} |
239 |
– |
|
240 |
– |
break; |
241 |
– |
case 1: |
242 |
– |
for (unsigned i = 0; i < hPFlow->size(); i++) { |
243 |
– |
particles.push_back(new GenericParticle((*hPFlow)[i])); |
244 |
– |
} |
245 |
– |
break; |
246 |
– |
default: |
247 |
– |
cerr << "Unknown event type" << endl; // TODO use MessageLogger |
248 |
– |
} |
249 |
– |
|
250 |
– |
return particles; |
251 |
– |
} |
252 |
– |
|
253 |
– |
TH2D * JetFinderAnalyzer::make_histo(const edm::Event &evt, const edm::EventSetup&) { |
254 |
– |
ostringstream oss; |
255 |
– |
oss << "eta_phi_energy_unred"<<evt.id().event() << flush; |
256 |
– |
TH2D *unred_histo = new TH2D(oss.str().c_str(), oss.str().c_str(), |
257 |
– |
600, -2.5, 2.5, 600, -PI, PI); |
258 |
– |
|
259 |
– |
vector<GenericParticle *> particles = get_particles(evt); |
260 |
– |
for (int i = 0; i < particles.size(); i++) { |
261 |
– |
unred_histo->Fill(particles[i]->eta(), |
262 |
– |
particles[i]->phi(), |
263 |
– |
particles[i]->e()); |
264 |
– |
} |
265 |
– |
|
266 |
– |
// reduce histo |
267 |
– |
ostringstream oss2; |
268 |
– |
oss2 << "eta_phi_energy_red"<<evt.id().event() << flush; |
269 |
– |
edm::Service<TFileService> fs; |
270 |
– |
// draw cone of radius 0.5 around highest energy bin, reduce |
271 |
– |
double maxE = 0.0; |
272 |
– |
int max_i = 29, max_j = 29; |
273 |
– |
for (int i = 0; i < unred_histo->GetNbinsX(); i++) { |
274 |
– |
for (int j = 0; j < unred_histo->GetNbinsY(); j++) { |
275 |
– |
double E = unred_histo->GetBinContent(i+1, j+1); |
276 |
– |
if (E > maxE) { |
277 |
– |
maxE = E; |
278 |
– |
max_i = i; |
279 |
– |
max_j = j; |
280 |
– |
} |
281 |
– |
} |
282 |
– |
} |
283 |
– |
|
284 |
– |
double rcone = 0.5; |
285 |
– |
double Xlo = unred_histo->GetXaxis()->GetXmin(); |
286 |
– |
double Xhi = unred_histo->GetXaxis()->GetXmax(); |
287 |
– |
double Ylo = unred_histo->GetYaxis()->GetXmin(); |
288 |
– |
double Yhi = unred_histo->GetYaxis()->GetXmax(); |
289 |
– |
double XbinSize = (Xhi - Xlo) / |
290 |
– |
static_cast<double>(unred_histo->GetXaxis()->GetNbins()); |
291 |
– |
double YbinSize = (Yhi - Ylo) / |
292 |
– |
static_cast<double>(unred_histo->GetYaxis()->GetNbins()); |
293 |
– |
double max_x = (static_cast<double>(max_i) + 0.5) * XbinSize + Xlo; |
294 |
– |
double max_y = (static_cast<double>(max_j) + 0.5) * YbinSize + Ylo; |
295 |
– |
TH2D *histo = fs->make<TH2D>(oss2.str().c_str(), oss2.str().c_str(), |
296 |
– |
60, max_x-rcone, max_x+rcone, |
297 |
– |
60, max_y-rcone, max_y+rcone); |
298 |
– |
|
299 |
– |
// create an unsmeared reduced histo |
300 |
– |
TH2D *histo_unsmeared = fs->make<TH2D>((oss2.str()+"_unsmeared").c_str(), |
301 |
– |
(oss2.str()+"_unsmeared").c_str(), |
302 |
– |
60, max_x-rcone, max_x+rcone, |
303 |
– |
60, max_y-rcone, max_y+rcone); |
304 |
– |
for (int i = 0; i < particles.size(); i++) { |
305 |
– |
double N = particles[i]->e(); |
306 |
– |
double x = particles[i]->eta(); |
307 |
– |
double y = particles[i]->phi(); |
308 |
– |
histo_unsmeared->Fill(x, y, N); |
309 |
– |
} |
310 |
– |
|
311 |
– |
// create a smeared reduced histo |
312 |
– |
// create a temporary 2D vector for smeared energies |
313 |
– |
XbinSize = (histo->GetXaxis()->GetXmax() |
314 |
– |
- histo->GetXaxis()->GetXmin()) / |
315 |
– |
static_cast<double>(histo->GetXaxis()->GetNbins()); |
316 |
– |
YbinSize = (histo->GetYaxis()->GetXmax() |
317 |
– |
- histo->GetYaxis()->GetXmin()) / |
318 |
– |
static_cast<double>(histo->GetYaxis()->GetNbins()); |
319 |
– |
vector< vector<double> > smeared(60, vector<double>(60, 0.0) ); |
320 |
– |
switch (smear_coord_) { |
321 |
– |
case 1: |
322 |
– |
for (int i = 0; i < particles.size(); i++) { |
323 |
– |
double N = particles[i]->e(); |
324 |
– |
double x = particles[i]->eta(); |
325 |
– |
double y = particles[i]->phi(); |
326 |
– |
// loop over bins and add Gaussian in proper angle to smeared |
327 |
– |
for (vector< vector<double> >::size_type i2 = 0; i2 < 60; i2++) { |
328 |
– |
for (vector< double >::size_type j2 = 0; j2 < 60; j2++) { |
329 |
– |
double eta = static_cast<double>((signed int)i2) * XbinSize + |
330 |
– |
max_x - rcone - x; |
331 |
– |
double phi = acos(cos(static_cast<double>((signed int)j2) * YbinSize + |
332 |
– |
max_y - rcone - y)); |
333 |
– |
phi = sin(phi) > 0 ? phi : -phi; |
334 |
– |
|
335 |
– |
// transform eta, phi to proper angle |
336 |
– |
double theta = 2.0*atan(exp(-eta)); |
337 |
– |
double iota = asin(sin(theta)*sin(phi)); |
338 |
– |
|
339 |
– |
smeared[i2][j2] += (N*XbinSize*YbinSize/(2.0*PI*smear_*smear_)) |
340 |
– |
* exp(-0.5*(theta*theta + iota*iota)/(smear_*smear_)); |
341 |
– |
} |
342 |
– |
} |
343 |
– |
} |
344 |
– |
break; |
345 |
– |
case 0: |
346 |
– |
default: |
347 |
– |
for (int i = 0; i < particles.size(); i++) { |
348 |
– |
double N = particles[i]->e(); |
349 |
– |
double x = particles[i]->eta(); |
350 |
– |
double y = particles[i]->phi(); |
351 |
– |
// loop over bins and add Gaussian to smeared |
352 |
– |
for (vector< vector<double> >::size_type i2 = 0; i2 < 60; i2++) { |
353 |
– |
for (vector< double >::size_type j2 = 0; j2 < 60; j2++) { |
354 |
– |
double eta = static_cast<double>((signed int)i2) * XbinSize |
355 |
– |
+ max_x - rcone - x; |
356 |
– |
double phi = acos(cos(static_cast<double>((signed int)j2) * YbinSize |
357 |
– |
+ max_y - rcone - y)); |
358 |
– |
phi = sin(phi) > 0 ? phi : -phi; |
359 |
– |
smeared[i2][j2] += (N*XbinSize*YbinSize/(2.0*PI*smear_*smear_)) |
360 |
– |
* exp(-0.5*(eta*eta + phi*phi)/(smear_*smear_)); |
361 |
– |
} |
362 |
– |
} |
363 |
– |
} |
364 |
– |
} |
365 |
– |
// set histogram to match smear vector |
366 |
– |
for (int i = 1; i <= 60; i++) { |
367 |
– |
for (int j = 1; j <= 60; j++) { |
368 |
– |
histo->SetBinContent(i, j, smeared[i-1][j-1]); |
369 |
– |
} |
370 |
– |
} |
371 |
– |
|
372 |
– |
return histo; |
373 |
– |
} |
374 |
– |
|
375 |
– |
void seed_with_CA(vector<GenericParticle *> gParticles, TH2 *histo, |
376 |
– |
jetfit::model_def &_mdef) { |
377 |
– |
// create a PseudoJet vector |
378 |
– |
vector<PseudoJet> particles; |
379 |
– |
for (unsigned i = 0; i < gParticles.size(); i++) { |
380 |
– |
double x_max = (histo->GetXaxis()->GetXmax() |
381 |
– |
+ histo->GetXaxis()->GetXmin()) / 2.0; |
382 |
– |
double y_max = (histo->GetYaxis()->GetXmax() |
383 |
– |
+ histo->GetYaxis()->GetXmin()) / 2.0; |
384 |
– |
valarray<double> pmom(4); |
385 |
– |
pmom[0] = gParticles[i]->px(); |
386 |
– |
pmom[1] = gParticles[i]->py(); |
387 |
– |
pmom[2] = gParticles[i]->pz(); |
388 |
– |
pmom[3] = gParticles[i]->e(); |
389 |
– |
double eta = gParticles[i]->eta(); |
390 |
– |
double phi = gParticles[i]->phi(); |
391 |
– |
if ((eta - x_max)*(eta - x_max) + (phi - y_max)*(phi - y_max) < 0.25) { |
392 |
– |
PseudoJet j(pmom); |
393 |
– |
particles.push_back(j); |
394 |
– |
} |
395 |
– |
} |
396 |
– |
|
397 |
– |
// choose a jet definition |
398 |
– |
double R = 0.2; |
399 |
– |
JetDefinition jet_def(cambridge_algorithm, R); |
400 |
– |
|
401 |
– |
// run clustering and extract the jets |
402 |
– |
ClusterSequence cs(particles, jet_def); |
403 |
– |
vector<PseudoJet> jets = cs.inclusive_jets(); |
404 |
– |
|
405 |
– |
double XbinSize = (histo->GetXaxis()->GetXmax() |
406 |
– |
- histo->GetXaxis()->GetXmin()) / |
407 |
– |
static_cast<double>(histo->GetXaxis()->GetNbins()); |
408 |
– |
double YbinSize = (histo->GetYaxis()->GetXmax() |
409 |
– |
- histo->GetYaxis()->GetXmin()) / |
410 |
– |
static_cast<double>(histo->GetYaxis()->GetNbins()); |
411 |
– |
|
412 |
– |
// seed with C-A jets |
413 |
– |
int ijset = 0; |
414 |
– |
for (unsigned ij = 0; ij < jets.size(); ij++) { |
415 |
– |
double N = jets[ij].e(); |
416 |
– |
if (N > 50.0) { |
417 |
– |
_mdef.set_special_par(ijset, 0, N, _mdef.chisquare_error(N)*0.1, |
418 |
– |
0.0, 1.0e6); |
419 |
– |
_mdef.set_special_par(ijset, 1, jets[ij].eta(), 0.01, |
420 |
– |
0.0, 0.0); |
421 |
– |
double mdef_phi = jets[ij].phi() > PI ? jets[ij].phi() - 2*PI |
422 |
– |
: jets[ij].phi(); |
423 |
– |
_mdef.set_special_par(ijset, 2, mdef_phi, 0.01, |
424 |
– |
0.0, 0.0); |
425 |
– |
_mdef.set_special_par(ijset, 3, 0.1, 0.001, 0.0, 0.0); |
426 |
– |
ijset++; |
427 |
– |
} |
428 |
– |
} |
429 |
– |
} |
430 |
– |
|
431 |
– |
jetfit::model_def& JetFinderAnalyzer::make_model_def(const edm::Event& evt, |
432 |
– |
const edm::EventSetup&, |
433 |
– |
TH2 *histo) { |
434 |
– |
class jf_model_def : public jetfit::model_def { |
435 |
– |
public: |
436 |
– |
virtual double chisquare_error(double E) { |
437 |
– |
return 0.97*E + 14.0; |
438 |
– |
// study from 08-27-09 |
439 |
– |
} |
440 |
– |
}; |
441 |
– |
|
442 |
– |
jf_model_def *_mdef = new jf_model_def(); |
443 |
– |
TFormula *formula = new TFormula("gaus2d", |
444 |
– |
"[0]*exp(-0.5*((x-[1])**2 + (y-[2])**2)/([3]**2))/(2*pi*[3]**2)"); |
445 |
– |
_mdef->set_formula(formula); |
446 |
– |
_mdef->set_indiv_max_E(0); |
447 |
– |
_mdef->set_indiv_max_x(1); |
448 |
– |
_mdef->set_indiv_max_y(2); |
449 |
– |
_mdef->set_indiv_par(0, string("N"), 0.0, 0.0, 0.0, 1.0e6); |
450 |
– |
_mdef->set_indiv_par(1, string("mu_x"), 0.0, 0.0, 0.0, 0.0); |
451 |
– |
_mdef->set_indiv_par(2, string("mu_y"), 0.0, 0.0, 0.0, 0.0); |
452 |
– |
_mdef->set_indiv_par(3, string("sig"), 0.1, 0.001, 0.0, 0.0); |
453 |
– |
|
454 |
– |
seed_with_CA(get_particles(evt), histo, *_mdef); |
455 |
– |
|
456 |
– |
jetfit::set_model_def(_mdef); |
457 |
– |
|
458 |
– |
// generate initial fit histogram |
459 |
– |
edm::Service<TFileService> fs; |
460 |
– |
TH2D *init_fit_histo = fs->make<TH2D>(("init_fit_"+string(histo->GetName())) |
461 |
– |
.c_str(), |
462 |
– |
("Initial fit for " |
463 |
– |
+string(histo->GetName())).c_str(), |
464 |
– |
histo->GetXaxis()->GetNbins(), |
465 |
– |
histo->GetXaxis()->GetXmin(), |
466 |
– |
histo->GetXaxis()->GetXmax(), |
467 |
– |
histo->GetXaxis()->GetNbins(), |
468 |
– |
histo->GetXaxis()->GetXmin(), |
469 |
– |
histo->GetXaxis()->GetXmax()); |
470 |
– |
double XbinSize = (histo->GetXaxis()->GetXmax() |
471 |
– |
- histo->GetXaxis()->GetXmin()) / |
472 |
– |
static_cast<double>(histo->GetXaxis()->GetNbins()); |
473 |
– |
double YbinSize = (histo->GetYaxis()->GetXmax() |
474 |
– |
- histo->GetYaxis()->GetXmin()) / |
475 |
– |
static_cast<double>(histo->GetYaxis()->GetNbins()); |
476 |
– |
double Xlo = histo->GetXaxis()->GetXmin(); |
477 |
– |
double Xhi = histo->GetXaxis()->GetXmax(); |
478 |
– |
double Ylo = histo->GetYaxis()->GetXmin(); |
479 |
– |
double Yhi = histo->GetYaxis()->GetXmax(); |
480 |
– |
|
481 |
– |
for (int i = 0; i < 60; i++) { |
482 |
– |
for (int j = 0; j < 60; j++) { |
483 |
– |
double x = (static_cast<double>(i) + 0.5)*XbinSize + Xlo; |
484 |
– |
double y = (static_cast<double>(j) + 0.5)*YbinSize + Ylo; |
485 |
– |
double pval[256]; |
486 |
– |
if (_mdef->get_n_special_par_sets() > 64) { |
487 |
– |
cerr << "Parameter overload" << endl; |
488 |
– |
return *_mdef; |
489 |
– |
} |
490 |
– |
else { |
491 |
– |
for (int is = 0; is < _mdef->get_n_special_par_sets(); is++) { |
492 |
– |
for (int ii = 0; ii < 4; ii++) { |
493 |
– |
double spval, sperr, splo, sphi; |
494 |
– |
_mdef->get_special_par(is, ii, spval, sperr, splo, sphi); |
495 |
– |
pval[4*is + ii] = spval; |
496 |
– |
} |
497 |
– |
} |
498 |
– |
} |
499 |
– |
jetfit::set_ngauss(_mdef->get_n_special_par_sets()); |
500 |
– |
init_fit_histo->SetBinContent(i+1, j+1, |
501 |
– |
jetfit::fit_fcn(x, y, pval)); |
502 |
– |
} |
503 |
– |
} |
504 |
– |
|
505 |
– |
return *_mdef; |
33 |
|
} |
34 |
|
|
35 |
|
void JetFinderAnalyzer::beginJob(const edm::EventSetup &es) { |
509 |
– |
ofs.open("jetfindlog.txt", ios::out); |
510 |
– |
if (ofs.fail()) { |
511 |
– |
cerr << "Opening jetfindlog.txt FAILED" << endl; |
512 |
– |
} |
513 |
– |
ofs << "Jetfinder log" << endl |
514 |
– |
<< "=============" << endl << endl; |
515 |
– |
} |
36 |
|
|
37 |
< |
ostream& operator<<(ostream &out, jetfit::trouble t) { |
518 |
< |
string action, error_string; |
519 |
< |
|
520 |
< |
if (t.istat != 3) { |
521 |
< |
switch(t.occ) { |
522 |
< |
case jetfit::T_NULL: |
523 |
< |
action = "Program"; break; |
524 |
< |
case jetfit::T_SIMPLEX: |
525 |
< |
action = "SIMPLEX"; break; |
526 |
< |
case jetfit::T_MIGRAD: |
527 |
< |
action = "MIGRAD"; break; |
528 |
< |
case jetfit::T_MINOS: |
529 |
< |
action = "MINOS"; break; |
530 |
< |
default: |
531 |
< |
action = "Program"; break; |
532 |
< |
} |
533 |
< |
|
534 |
< |
switch (t.istat) { |
535 |
< |
case 0: |
536 |
< |
error_string = "Unable to calculate error matrix"; break; |
537 |
< |
case 1: |
538 |
< |
error_string = "Error matrix a diagonal approximation"; break; |
539 |
< |
case 2: |
540 |
< |
error_string = "Error matrix not positive definite"; break; |
541 |
< |
case 3: |
542 |
< |
error_string = "Converged successfully"; break; |
543 |
< |
default: |
544 |
< |
ostringstream oss; |
545 |
< |
oss<<"Unknown status code "<<t.istat << flush; |
546 |
< |
error_string = oss.str(); break; |
547 |
< |
} |
37 |
> |
} |
38 |
|
|
39 |
< |
if (t.occ != jetfit::T_NULL) |
40 |
< |
out << action<<" trouble: "<<error_string; |
41 |
< |
else |
42 |
< |
out << "Not calculated" << endl; |
39 |
> |
double evalFitFunction(HistoFitter::FitResults r, double x, double y) { |
40 |
> |
unsigned nFits = r.pars.size(); |
41 |
> |
unsigned nGauss = r.pars[nFits-1].size() / 4; |
42 |
> |
double fitVal = 0.0; |
43 |
> |
for (unsigned i = 0; i < nGauss; i++) { |
44 |
> |
double N = r.pval[nFits-1][4*i]; |
45 |
> |
double mu_x = r.pval[nFits-1][4*i + 1]; |
46 |
> |
double mu_y = r.pval[nFits-1][4*i + 2]; |
47 |
> |
double sig = r.pval[nFits-1][4*i + 3]; |
48 |
> |
|
49 |
> |
double rel_x = x - mu_x; double rel_y = y - mu_y; |
50 |
> |
fitVal += (N / 2.0 / M_PI / sig / sig) |
51 |
> |
* exp(-(rel_x * rel_x + rel_y * rel_y)/2.0/sig/sig); |
52 |
|
} |
53 |
< |
|
555 |
< |
return out; |
53 |
> |
return fitVal; |
54 |
|
} |
55 |
|
|
56 |
< |
void JetFinderAnalyzer::analyze_results(jetfit::results r, |
57 |
< |
std::vector<jetfit::trouble> t, |
56 |
> |
void JetFinderAnalyzer::analyze_results(HistoFitter::FitResults r, |
57 |
> |
std::vector<HistoFitter::Trouble> t, |
58 |
|
TH2 *hist_orig) { |
59 |
< |
ofs << "Histogram "<<hist_orig->GetName() << endl; |
562 |
< |
for (int i = 0; i < unique_jets[hist_orig].size(); i++) { |
563 |
< |
ofs << "For "<<i+1<<" gaussians: " << endl |
564 |
< |
<< t.at(i) << endl |
565 |
< |
<< unique_jets[hist_orig][i].size()<<" unique jets found" << endl; |
566 |
< |
for (int j = 0; j < unique_jets[hist_orig][i].size(); j++) { |
567 |
< |
jet _jet = unique_jets[hist_orig][i][j]; |
568 |
< |
ofs << "Jet "<<j<<": Energy = "<<_jet.energy<<", eta = "<<_jet.eta |
569 |
< |
<< ", phi = "<<_jet.phi << endl; |
570 |
< |
} |
571 |
< |
ofs << endl; |
572 |
< |
} |
573 |
< |
ofs << endl; |
574 |
< |
|
575 |
< |
// save fit function histograms to root file |
59 |
> |
// perform analysis of fit results |
60 |
|
edm::Service<TFileService> fs; |
61 |
< |
for (vector< vector<double> >::size_type i = 0; |
62 |
< |
i < r.pval.size(); i++) { |
63 |
< |
jetfit::set_ngauss(r.pval[i].size() / 4); |
64 |
< |
TF2 *tf2 = new TF2("fit_func", jetfit::fit_fcn_TF2, |
65 |
< |
hist_orig->GetXaxis()->GetXmin(), |
66 |
< |
hist_orig->GetXaxis()->GetXmax(), |
67 |
< |
hist_orig->GetYaxis()->GetXmin(), |
68 |
< |
hist_orig->GetYaxis()->GetXmax(), |
69 |
< |
r.pval[i].size()); |
70 |
< |
for (vector<double>::size_type j = 0; j < r.pval[i].size(); j++) { |
71 |
< |
tf2->SetParameter(j, r.pval[i][j]); |
72 |
< |
} |
73 |
< |
ostringstream fit_histo_oss; |
74 |
< |
fit_histo_oss << hist_orig->GetName()<<"_fit_"<<i << flush; |
75 |
< |
tf2->SetNpx(hist_orig->GetXaxis()->GetNbins()); |
76 |
< |
tf2->SetNpy(hist_orig->GetYaxis()->GetNbins()); |
77 |
< |
TH2D *fit_histo = fs->make<TH2D>(fit_histo_oss.str().c_str(), |
78 |
< |
fit_histo_oss.str().c_str(), |
79 |
< |
hist_orig->GetXaxis()->GetNbins(), |
80 |
< |
hist_orig->GetXaxis()->GetXmin(), |
81 |
< |
hist_orig->GetXaxis()->GetXmax(), |
82 |
< |
hist_orig->GetYaxis()->GetNbins(), |
599 |
< |
hist_orig->GetYaxis()->GetXmin(), |
600 |
< |
hist_orig->GetYaxis()->GetXmax()); |
601 |
< |
TH1 *tf2_histo = tf2->CreateHistogram(); |
602 |
< |
double XbinSize = (fit_histo->GetXaxis()->GetXmax() |
603 |
< |
- fit_histo->GetXaxis()->GetXmin()) |
604 |
< |
/ static_cast<double>(fit_histo->GetXaxis()->GetNbins()); |
605 |
< |
double YbinSize = (fit_histo->GetYaxis()->GetXmax() |
606 |
< |
- fit_histo->GetYaxis()->GetXmin()) |
607 |
< |
/ static_cast<double>(fit_histo->GetYaxis()->GetNbins()); |
608 |
< |
for (int ih = 0; ih < tf2->GetNpx(); ih++) { |
609 |
< |
for (int jh = 0; jh < tf2->GetNpy(); jh++) { |
610 |
< |
fit_histo->SetBinContent(ih+1, jh+1, |
611 |
< |
tf2_histo->GetBinContent(ih+1, jh+1) |
612 |
< |
* XbinSize * YbinSize); |
613 |
< |
} |
61 |
> |
TH2D *fitHisto = fs->make<TH2D>((std::string(hist_orig->GetName())+"_fit").c_str(), |
62 |
> |
("Fitted distribution to " |
63 |
> |
+std::string(hist_orig->GetName())).c_str(), |
64 |
> |
hist_orig->GetNbinsX(), |
65 |
> |
hist_orig->GetXaxis()->GetXmin(), |
66 |
> |
hist_orig->GetXaxis()->GetXmax(), |
67 |
> |
hist_orig->GetNbinsY(), |
68 |
> |
hist_orig->GetYaxis()->GetXmin(), |
69 |
> |
hist_orig->GetYaxis()->GetXmax()); |
70 |
> |
|
71 |
> |
double Xlo = fitHisto->GetXaxis()->GetXmin(); |
72 |
> |
double Xhi = fitHisto->GetXaxis()->GetXmax(); |
73 |
> |
double Ylo = fitHisto->GetYaxis()->GetXmin(); |
74 |
> |
double Yhi = fitHisto->GetYaxis()->GetXmax(); |
75 |
> |
double XbinSize = (Xhi - Xlo) / static_cast<double>(fitHisto->GetNbinsX()); |
76 |
> |
double YbinSize = (Yhi - Ylo) / static_cast<double>(fitHisto->GetNbinsY()); |
77 |
> |
|
78 |
> |
for (int i = 1; i <= fitHisto->GetNbinsX(); i++) { |
79 |
> |
for (int j = 1; j <= fitHisto->GetNbinsY(); j++) { |
80 |
> |
double x = (static_cast<double>(i) - 0.5) * XbinSize + Xlo; |
81 |
> |
double y = (static_cast<double>(j) - 0.5) * YbinSize + Ylo; |
82 |
> |
fitHisto->SetBinContent(i, j, evalFitFunction(r, x, y) * XbinSize * YbinSize); |
83 |
|
} |
84 |
|
} |
85 |
|
|
86 |
< |
// save results to file |
87 |
< |
ostringstream res_tree_oss, rt_title_oss; |
88 |
< |
res_tree_oss << hist_orig->GetName()<<"_results" << flush; |
89 |
< |
rt_title_oss << "Fit results for "<<hist_orig->GetName() << flush; |
86 |
> |
// save fit results to an ntuple |
87 |
> |
TNtuple *rNtuple = fs->make<TNtuple>((std::string(hist_orig->GetName())+"_results").c_str(), |
88 |
> |
("Fit results for "+std::string(hist_orig->GetName())).c_str(), |
89 |
> |
"N:mu_x:mu_y:sigma"); |
90 |
> |
unsigned nFits = r.pval.size(); |
91 |
> |
unsigned nGauss = r.pval[nFits-1].size() / 4; |
92 |
> |
for (unsigned i = 0; i < nGauss; i++) { |
93 |
> |
rNtuple->Fill(r.pval[nFits-1][4*i], r.pval[nFits-1][4*i+1], r.pval[nFits-1][4*i+2], |
94 |
> |
r.pval[nFits-1][4*i+3]); |
95 |
> |
} |
96 |
> |
|
97 |
> |
// save chisquares to ntuple |
98 |
> |
for (unsigned i = 0; i < r.chisquare.size(); i++) { |
99 |
> |
ostringstream csNtupleName, csNtupleTitle; |
100 |
> |
csNtupleName << hist_orig->GetName() << "_chi2_" << i << flush; |
101 |
> |
csNtupleTitle << "Chisquare "<<i<<" for histo "<<hist_orig->GetName() |
102 |
> |
<< flush; |
103 |
> |
TNtuple *csNtuple = fs->make<TNtuple>(csNtupleName.str().c_str(), |
104 |
> |
csNtupleTitle.str().c_str(), |
105 |
> |
"chisq"); |
106 |
> |
csNtuple->Fill(r.chisquare[i]); |
107 |
> |
} |
108 |
|
} |
109 |
|
|
110 |
|
DEFINE_FWK_MODULE(JetFinderAnalyzer); |