1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
|
//
// CMSCGEN.cc version 3.0 Thomas Hebbeker 2007-05-15
//
// implemented in CMSSW by P. Biallass 2007-05-28
// see header for documentation and CMS internal note 2007 "Improved Parametrization of the Cosmic Muon Flux for the generator CMSCGEN" by Biallass + Hebbeker
//
#include <CLHEP/Random/RandomEngine.h>
#include <CLHEP/Random/JamesRandom.h>
#include "GeneratorInterface/CosmicMuonGenerator/interface/CMSCGEN.h"
CMSCGEN::CMSCGEN() : initialization(0), RanGen2(nullptr), delRanGen(false) {}
CMSCGEN::~CMSCGEN() {
if (delRanGen)
delete RanGen2;
}
void CMSCGEN::setRandomEngine(CLHEP::HepRandomEngine *v) {
if (delRanGen)
delete RanGen2;
RanGen2 = v;
delRanGen = false;
}
int CMSCGEN::initialize(double pmin_in,
double pmax_in,
double thetamin_in,
double thetamax_in,
CLHEP::HepRandomEngine *rnd,
bool TIFOnly_constant,
bool TIFOnly_linear) {
if (delRanGen)
delete RanGen2;
RanGen2 = rnd;
delRanGen = false;
//set bools for TIFOnly options (E<2GeV with unphysical energy dependence)
TIFOnly_const = TIFOnly_constant;
TIFOnly_lin = TIFOnly_linear;
// units: GeV
// WARNING: coordinate system:
// - to outside world define z axis downwards, i.e.
// muon coming from above, vertically: cos = 1
// (used for backward compatibility)
// - internally use frame with z axis upwards, i.e.
// muon coming from above, vertically: cos = -1
// (corresponds to CMS note definition)
//set cmin and cmax, here convert between coordinate systems:
cmin_in = -TMath::Cos(thetamin_in); //input angle already converted from Deg to Rad!
cmax_in = -TMath::Cos(thetamax_in); //input angle already converted from Deg to Rad!
//allowed energy range
pmin_min = 3.;
//pmin_max = 100.;
pmin_max = 3000.;
pmax = 3000.;
//allowed angular range
//cmax_max = -0.1,
cmax_max = -0.01, cmax_min = -0.9999;
if (TIFOnly_const == true || TIFOnly_lin == true)
pmin_min = 0.; //forTIF
// set pmin
if (pmin_in < pmin_min || pmin_in > pmin_max) {
std::cout << " >>> CMSCGEN.initialize <<< warning: illegal pmin_in =" << pmin_in;
return (-1);
} else if (pmax_in > pmax) {
std::cout << " >>> CMSCGEN.initialize <<< warning: illegal pmax_in =" << pmax_in;
return (-1);
} else {
pmin = pmin_in;
pmax = pmax_in;
xemax = 1. / (pmin * pmin);
xemin = 1. / (pmax * pmax);
}
// set cmax and cmin
if (cmax_in < cmax_min || cmax_in > cmax_max) {
std::cout << " >>> CMSCGEN.initialize <<< warning: illegal cmax_in =" << cmax_in;
return (-1);
} else {
cmax = cmax_in;
cmin = cmin_in;
}
initialization = 1;
if (TIFOnly_const == true || TIFOnly_lin == true)
pmin_min = 3.; //forTIF
// Lmin = log10(pmin_min);
Lmin = log10(pmin);
Lmax = log10(pmax);
Lfac = 100. / (Lmax - Lmin);
//
// +++ coefficients for energy spectrum
//
pe[0] = -1.;
pe[1] = 6.22176;
pe[2] = -13.9404;
pe[3] = 18.1643;
pe[4] = -9.22784;
pe[5] = 1.99234;
pe[6] = -0.156434;
pe[7] = 0.;
pe[8] = 0.;
//
// +++ coefficients for cos theta distribution
//
b0c[0] = 0.6639;
b0c[1] = -0.9587;
b0c[2] = 0.2772;
b1c[0] = 5.820;
b1c[1] = -6.864;
b1c[2] = 1.367;
b2c[0] = 10.39;
b2c[1] = -8.593;
b2c[2] = 1.547;
//
// +++ calculate correction table for different cos theta dependence!
// reference range: c1 to c2
//
// explanation: the parametrization of the energy spectrum as used above
// is the integral over the c = cos(zenith angle) range -1...-0.1
// since the c distribution depends on energy, the integrated energy
// spectrum depends on this range. Here a correction factor is determined,
// based on the linear c dependence of the c distribution.
// The correction is calculated for 100 bins in L = log10(energy).
//
// +++ in same loop calculate integrated flux
// (integrated over angles and momentum)
c1 = -1.;
c2 = -0.1;
double cemax0 = 1.0;
double L, L2;
double s;
double p, p1, p2;
double integral_here, integral_ref;
double c_cut;
integrated_flux = 0.;
for (int k = 1; k <= 100; k++) {
L = Lmin + (k - 0.5) / Lfac;
L2 = L * L;
p = pow(10, L);
p1 = pow(10, L - 0.5 / Lfac);
p2 = pow(10, L + 0.5 / Lfac);
b0 = b0c[0] + b0c[1] * L + b0c[2] * L2;
b1 = b1c[0] + b1c[1] * L + b1c[2] * L2;
b2 = b2c[0] + b2c[1] * L + b2c[2] * L2;
// cut out explicitly regions of zero flux
// (for low momentum and near horizontal showers)
// since parametrization for z distribution doesn't work here
// (can become negative)
c_cut = -0.42 + L * 0.35;
if (c_cut > c2)
c_cut = c2;
integral_ref =
b0 * (c_cut - c1) + b1 / 2. * (c_cut * c_cut - c1 * c1) + b2 / 3. * (c_cut * c_cut * c_cut - c1 * c1 * c1);
if (c_cut > cmax)
c_cut = cmax;
integral_here = b0 * (c_cut - cmin) + b1 / 2. * (c_cut * c_cut - cmin * cmin) +
b2 / 3. * (c_cut * c_cut * c_cut - cmin * cmin * cmin);
corr[k] = integral_here / integral_ref;
s = (((((((pe[8] * L + pe[7]) * L + pe[6]) * L + pe[5]) * L + pe[4]) * L + pe[3]) * L + pe[2]) * L + pe[1]) * L +
pe[0];
integrated_flux += 1. / pow(p, 3) * s * corr[k] * (p2 - p1);
/*
std::cout << k << " "
<< corr[k] << " "
<< p << " "
<< s << " "
<< p1 << " "
<< p2 << " "
<< integrated_flux << " "
<< std::endl;
*/
// std::cout << k << " " << corr[k] << " " << std::endl;
}
integrated_flux *= 1.27E3;
std::cout << " >>> CMSCGEN.initialize <<< "
<< " Integrated flux = " << integrated_flux << " /m**2/s " << std::endl;
// find approximate peak value, for Monte Carlo sampling
// peak is near L = 2
double ce;
ce = (((((((pe[8] * 2. + pe[7]) * 2. + pe[6]) * 2. + pe[5]) * 2. + pe[4]) * 2. + pe[3]) * 2. + pe[2]) * 2. + pe[1]) *
2. +
pe[0];
// normalize to 0.5 (not 1) to have some margin if peak is not at L=2
//
ce = 0.5 / ce;
for (int k = 0; k < 9; k++) {
pe[k] = pe[k] * ce;
}
cemax = cemax0 * corr[50];
return initialization;
}
int CMSCGEN::initialize(double pmin_in,
double pmax_in,
double thetamin_in,
double thetamax_in,
int RanSeed,
bool TIFOnly_constant,
bool TIFOnly_linear) {
CLHEP::HepRandomEngine *rnd = new CLHEP::HepJamesRandom;
//set seed for Random Generator (seed can be controled by config-file), P.Biallass 2006
rnd->setSeed(RanSeed, 0);
delRanGen = true;
return initialize(pmin_in, pmax_in, thetamin_in, thetamax_in, rnd, TIFOnly_constant, TIFOnly_linear);
}
int CMSCGEN::generate() {
if (initialization == 0) {
std::cout << " >>> CMSCGEN <<< warning: not initialized" << std::endl;
return -1;
}
// note: use historical notation (fortran version l3cgen.f)
//
// +++ determine x = 1/e**2
//
// explanation: the energy distribution
// dn/d(1/e**2) = dn/de * e**3 = dn/dlog10(e) * e**2
// is parametrized by a polynomial. accordingly xe = 1/e**2 is sampled
// and e calculated
//
// need precise random variable with high precison since for
// emin = 3 GeV energies around 3000 GeV are very rare!
//
double r1, r2, r3;
double xe, e, ce, L, L2;
int k;
double prob;
double c_max;
double z, z_max;
while (true) {
prob = RanGen2->flat();
r1 = double(prob);
prob = RanGen2->flat();
r2 = double(prob);
xe = xemin + r1 * (xemax - xemin);
if ((1. / sqrt(xe) < 3) &&
TIFOnly_const == true) { //generate constant energy dependence for E<2GeV, only used for TIF
//compute constant to match to CMSCGEN spectrum
e = 3.;
L = log10(e);
L2 = L * L;
ce = (((((((pe[8] * L + pe[7]) * L + pe[6]) * L + pe[5]) * L + pe[4]) * L + pe[3]) * L + pe[2]) * L + pe[1]) * L +
pe[0];
k = int((L - Lmin) * Lfac + 1.);
k = TMath::Max(1, TMath::Min(k, 100));
ce = ce * corr[k];
e = 1. / sqrt(xe);
if (r2 < (e * e * e * ce / (cemax * 3. * 3. * 3.)))
break;
} else if ((1. / sqrt(xe) < 3) &&
TIFOnly_lin == true) { //generate linear energy dependence for E<2GeV, only used for TIF
//compute constant to match to CMSCGEN spectrum
e = 3.;
L = log10(e);
L2 = L * L;
ce = (((((((pe[8] * L + pe[7]) * L + pe[6]) * L + pe[5]) * L + pe[4]) * L + pe[3]) * L + pe[2]) * L + pe[1]) * L +
pe[0];
k = int((L - Lmin) * Lfac + 1.);
k = TMath::Max(1, TMath::Min(k, 100));
ce = ce * corr[k];
e = 1. / sqrt(xe);
if (r2 < (e * e * e * e * ce / (cemax * 3. * 3. * 3. * 3.)))
break;
} else { //this is real CMSCGEN energy-dependence
e = 1. / sqrt(xe);
L = log10(e);
L2 = L * L;
ce = (((((((pe[8] * L + pe[7]) * L + pe[6]) * L + pe[5]) * L + pe[4]) * L + pe[3]) * L + pe[2]) * L + pe[1]) * L +
pe[0];
k = int((L - Lmin) * Lfac + 1.);
k = TMath::Max(1, TMath::Min(k, 100));
ce = ce * corr[k];
if (cemax * r2 < ce)
break;
} //end of CMSCGEN energy-dependence
} //end of while
pq = e;
//
// +++ charge ratio 1.280
//
prob = RanGen2->flat();
r3 = double(prob);
double charg = 1.;
if (r3 < 0.439)
charg = -1.;
pq = pq * charg;
//
// +++ determine cos(angle)
//
// simple trial and rejection method
//
// first calculate energy dependent coefficients b_i
if (TIFOnly_const == true && e < 3.) { //forTIF (when E<2GeV use angles of 2GeV cosmic)
L = log10(3.);
L2 = L * L;
}
if (TIFOnly_lin == true && e < 3.) { //forTIF (when E<2GeV use angles of 2GeV cosmic)
L = log10(3.);
L2 = L * L;
}
b0 = b0c[0] + b0c[1] * L + b0c[2] * L2;
b1 = b1c[0] + b1c[1] * L + b1c[2] * L2;
b2 = b2c[0] + b2c[1] * L + b2c[2] * L2;
//
// need to know the maximum of z(c)
//
// first calculate c for which c distribution z(c) = maximum
//
// (note: maximum of curve is NOT always at c = -1, but never at c = -0.1)
//
// try extremal value (from z'(c) = 0), but only if z''(c) < 0
//
// z'(c) = b1 + b2 * c => at c_max = - b1 / (2 b_2) is z'(c) = 0
//
// z''(c) = b2
c_max = -1.;
if (b2 < 0.) {
c_max = -0.5 * b1 / b2;
if (c_max < -1.)
c_max = -1.;
if (c_max > -0.1)
c_max = -0.1;
}
z_max = b0 + b1 * c_max + b2 * c_max * c_max;
// again cut out explicitly regions of zero flux
double c_cut = -0.42 + L * 0.35;
if (c_cut > cmax)
c_cut = cmax;
// now we throw dice:
while (true) {
prob = RanGen2->flat();
r1 = double(prob);
prob = RanGen2->flat();
r2 = double(prob);
c = cmin + (c_cut - cmin) * r1;
z = b0 + b1 * c + b2 * c * c;
if (z > z_max * r2)
break;
}
return 0;
}
double CMSCGEN::momentum_times_charge() {
if (initialization == 1) {
return pq;
} else {
std::cout << " >>> CMSCGEN <<< warning: not initialized" << std::endl;
return -9999.;
}
}
double CMSCGEN::cos_theta() {
if (initialization == 1) {
// here convert between coordinate systems:
return -c;
} else {
std::cout << " >>> CMSCGEN <<< warning: not initialized" << std::endl;
return -0.9999;
}
}
double CMSCGEN::flux() {
if (initialization == 1) {
return integrated_flux;
} else {
std::cout << " >>> CMSCGEN <<< warning: not initialized" << std::endl;
return -0.9999;
}
}
int CMSCGEN::initializeNuMu(double pmin_in,
double pmax_in,
double thetamin_in,
double thetamax_in,
double Enumin_in,
double Enumax_in,
double Phimin_in,
double Phimax_in,
double ProdAlt_in,
CLHEP::HepRandomEngine *rnd) {
if (delRanGen)
delete RanGen2;
RanGen2 = rnd;
delRanGen = false;
ProdAlt = ProdAlt_in;
Rnunubar = 1.2;
sigma = (0.72 * Rnunubar + 0.09) / (1 + Rnunubar) * 1.e-38; //cm^2GeV^-1
AR = (0.69 + 0.06 * Rnunubar) / (0.09 + 0.72 * Rnunubar);
//set smin and smax, here convert between coordinate systems:
pmin = pmin_in;
pmax = pmax_in;
cmin = TMath::Cos(thetamin_in); //input angle already converted from Deg to Rad!
cmax = TMath::Cos(thetamax_in); //input angle already converted from Deg to Rad!
enumin = (Enumin_in < 10.) ? 10. : Enumin_in; //no nu's below 10GeV
enumax = Enumax_in;
//do initial run of flux rate to determine Maximum
integrated_flux = 0.;
dNdEmudEnuMax = 0.;
negabs = 0.;
negfrac = 0.;
int trials = 100000;
for (int i = 0; i < trials; ++i) {
double ctheta = cmin + (cmax - cmin) * RanGen2->flat();
double Emu = pmin + (pmax - pmin) * RanGen2->flat();
double Enu = enumin + (enumax - enumin) * RanGen2->flat();
double rate = dNdEmudEnu(Enu, Emu, ctheta);
//std::cout << "trial=" << i << " ctheta=" << ctheta << " Emu=" << Emu << " Enu=" << Enu
// << " rate=" << rate << std::endl;
//std::cout << "cmin=" << cmin << " cmax=" << cmax
// << " pmin=" << pmin << " pmax=" << pmax
// << " enumin=" << enumin << " enumax=" << enumax << std::endl;
if (rate > 0.) {
integrated_flux += rate;
if (rate > dNdEmudEnuMax)
dNdEmudEnuMax = rate;
} else
negabs++;
}
negfrac = negabs / trials;
integrated_flux /= trials;
std::cout << "CMSCGEN::initializeNuMu: After " << trials << " trials:" << std::endl;
std::cout << "dNdEmudEnuMax=" << dNdEmudEnuMax << std::endl;
std::cout << "negfrac=" << negfrac << std::endl;
//multiply by phase space boundaries
integrated_flux *= (cmin - cmax);
integrated_flux *= (Phimax_in - Phimin_in);
integrated_flux *= (pmax - pmin);
integrated_flux *= (enumax - enumin);
//remove negative phase space areas which do not contribute anything
integrated_flux *= (1. - negfrac);
std::cout << " >>> CMSCGEN.initializeNuMu <<< "
<< " Integrated flux = " << integrated_flux << " units??? " << std::endl;
initialization = 1;
return initialization;
}
int CMSCGEN::initializeNuMu(double pmin_in,
double pmax_in,
double thetamin_in,
double thetamax_in,
double Enumin_in,
double Enumax_in,
double Phimin_in,
double Phimax_in,
double ProdAlt_in,
int RanSeed) {
CLHEP::HepRandomEngine *rnd = new CLHEP::HepJamesRandom;
//set seed for Random Generator (seed can be controled by config-file), P.Biallass 2006
rnd->setSeed(RanSeed, 0);
delRanGen = true;
return initializeNuMu(
pmin_in, pmax_in, thetamin_in, thetamax_in, Enumin_in, Enumax_in, Phimin_in, Phimax_in, ProdAlt_in, rnd);
}
double CMSCGEN::dNdEmudEnu(double Enu, double Emu, double ctheta) {
double cthetaNu = 1. + ctheta; //swap cos(theta) from down to up range
double thetas = asin(sin(acos(cthetaNu)) * (Rearth - SurfaceOfEarth) / (Rearth + ProdAlt));
double costhetas = cos(thetas);
double dNdEnudW =
0.0286 * pow(Enu, -2.7) *
(1. / (1. + (6. * Enu * costhetas) / 115.) + 0.213 / (1. + (1.44 * Enu * costhetas) / 850.)); //cm^2*s*sr*GeV
double dNdEmudEnu = N_A * sigma / alpha * dNdEnudW * 1. / (1. + Emu / epsilon) *
(Enu - Emu + AR / 3 * (Enu * Enu * Enu - Emu * Emu * Emu) / (Enu * Enu));
return dNdEmudEnu;
}
int CMSCGEN::generateNuMu() {
if (initialization == 0) {
std::cout << " >>> CMSCGEN <<< warning: not initialized" << std::endl;
return -1;
}
double ctheta, Emu;
while (true) {
ctheta = cmin + (cmax - cmin) * RanGen2->flat();
Emu = pmin + (pmax - pmin) * RanGen2->flat();
double Enu = enumin + (enumax - enumin) * RanGen2->flat();
double rate = dNdEmudEnu(Enu, Emu, ctheta);
if (rate > dNdEmudEnuMax * RanGen2->flat())
break;
}
c = -ctheta; //historical sign convention
pq = Emu;
//
// +++ nu/nubar ratio (~1.2)
//
double charg = 1.; //nubar -> mu+
if (RanGen2->flat() > Rnunubar / (1. + Rnunubar))
charg = -1.; //neutrino -> mu-
pq = pq * charg;
//int flux += this event rate
return 1;
}
|