void Neuron::excite(double clock){
for (std::vector<std::vector<double> >::const_iterator i = spikes.begin(); i != spikes.end(); i++){
double w = (*i)[0];
double t = (*i)[1];
v += w * decay(clock - t);
}
}
forceinline void
FloatVarBranch::expand(Home home, const FloatVarArgs& x) {
switch (select()) {
case SEL_AFC_MIN: case SEL_AFC_MAX:
case SEL_AFC_SIZE_MIN: case SEL_AFC_SIZE_MAX:
if (!_afc.initialized())
_afc = FloatAFC(home,x,decay());
break;
case SEL_ACTIVITY_MIN: case SEL_ACTIVITY_MAX:
case SEL_ACTIVITY_SIZE_MIN: case SEL_ACTIVITY_SIZE_MAX:
if (!_act.initialized())
_act = FloatActivity(home,x,decay());
break;
default: ;
}
}
QImage makeShadow( const QPixmap& textPixmap, const QColor &bgColor )
{
const int w = textPixmap.width();
const int h = textPixmap.height();
const int bgr = bgColor.red();
const int bgg = bgColor.green();
const int bgb = bgColor.blue();
int alphaShadow;
// This is the source pixmap
QImage img = textPixmap.toImage();
QImage result( w, h, QImage::Format_ARGB32 );
result.fill( 0 ); // fill with black
static const int M = 5;
for( int i = M; i < w - M; i++) {
for( int j = M; j < h - M; j++ )
{
alphaShadow = (int) decay( img, i, j );
result.setPixel( i,j, qRgba( bgr, bgg , bgb, qMin( MAX_OPACITY, alphaShadow ) ) );
}
}
return result;
}
QImage ShadowEngine::makeShadow(const QImage& textImage, const QColor &bgColor)
{
// create a new image for for the shaddow
int w = textImage.width();
int h = textImage.height();
QImage result(w, h, QImage::Format_ARGB32);
// avoid calling these methods for every pixel
int bgRed = bgColor.red();
int bgGreen = bgColor.green();
int bgBlue = bgColor.blue();
double alphaShadow;
result.fill(0); // transparent
for (int i = thickness_; i < w - thickness_; i++)
{
for (int j = thickness_; j < h - thickness_; j++)
{
alphaShadow = decay(textImage, i, j);
alphaShadow = (alphaShadow > 180.0) ? 180.0 : alphaShadow;
// update the shadow's i,j pixel.
result.setPixel(i,j, qRgba(bgRed, bgGreen , bgBlue, (int) alphaShadow));
}
}
return result;
}
static unsigned long
phosphor_draw (Display *dpy, Window window, void *closure)
{
p_state *state = (p_state *) closure;
update_display (state, True);
decay (state);
drain_input (state);
return state->delay;
}
void GuiDrawLine::ageLine() {
// cout << "ageLine() \n";
int milliseconds = lifeSpan*1000;
int stages = 5;
QTimer *timer = new QTimer(this);
connect(timer, SIGNAL(timeout()), this, SLOT(decay()));
if (age == 5) {
timer->start(milliseconds/stages);
} else {
// timer->stop();
}
}
void ADSREnvelopeChip::update()
{
mEnvelopeValueFloat += mEnvelopeDeltaValueFloat;
mEnvelopeValueFloat = fmaxf(0,mEnvelopeValueFloat);
if (!mNoteOn && noteOnInput.getInputBit())
attack();
else if (mNoteOn && !noteOnInput.getInputBit())
release();
if (mEnvelopeValueFloat>=1 && mEnvelopeDeltaValueFloat>0)
{
mEnvelopeValueFloat = 1;
decay();
}
else if (mEnvelopeValueFloat<=sLevel && mEnvelopeDeltaValueFloat<0 && mStage == 2)
sustain();
}
QImage ShadowEngine::makeShadow(const QPixmap& textPixmap, const QColor &bgColor)
{
QImage result;
// create a new image for for the shaddow
int w = textPixmap.width();
int h = textPixmap.height();
// avoid calling these methods for every pixel
int bgRed = bgColor.red();
int bgGreen = bgColor.green();
int bgBlue = bgColor.blue();
float alphaShadow;
/*
* This is the source pixmap
*/
QImage img = textPixmap.convertToImage().convertDepth(32);
/*
* Resize the image if necessary
*/
if ((result.width() != w) || (result.height() != h))
{
result.create(w, h, 32);
}
result.fill(0); // all black
result.setAlphaBuffer(true);
for (int i = thickness_; i < w - thickness_; i++)
{
for (int j = thickness_; j < h - thickness_; j++)
{
alphaShadow = decay(img, i, j);
alphaShadow = (alphaShadow > 180.0) ? 180.0 : alphaShadow;
// update the shadow's i,j pixel.
result.setPixel(i,j, qRgba(bgRed, bgGreen , bgBlue, (int) alphaShadow));
}
}
return result;
}
bool EVector::update(ssi_size_t framework_time, ssi_real_t *baseline){
//calculate lifetime
set_lifetime(framework_time);
subtract_baseline(baseline);
//calculate norms
norm_values();
//decay
decay();
add_baseline(baseline);
//calculate new norms
norm_values();
return true;
}
lbfgsfloatval_t RAE::_evaluate(const lbfgsfloatval_t* x, lbfgsfloatval_t* g, const int n, const lbfgsfloatval_t step)
{
lbfgsfloatval_t fx = 0;
int RAEThreadNum = atoi(para->getPara("RAEThreadNum").c_str());
RAEThreadPara* threadpara = new RAEThreadPara[RAEThreadNum];
int batchsize = trainingData.size() / RAEThreadNum;
updateWeights(x);
for(int i = 0; i < RAEThreadNum; i++)
{
threadpara[i].cRAE = this->copy();
threadpara[i].g = lbfgs_malloc(getRAEWeightSize());
for(int j = 0; j < getRAEWeightSize(); j++)
{
threadpara[i].g[j] = 0;
}
if(i == RAEThreadNum-1)
{
map<string, int>::iterator s;
s = trainingData.begin();
for(int d = 0; d < i*batchsize; d++)
{
s++;
}
threadpara[i].cRAE->trainingData.insert(s, trainingData.end());
threadpara[i].instance_num = trainingData.size()%batchsize;
}
else
{
map<string, int>::iterator s, e;
s = trainingData.begin();
e = trainingData.begin();
for(int d = 0; d < i*batchsize; d++)
{
s++;
}
for(int d = 0; d < (i+1)*batchsize; d++)
{
e++;
}
threadpara[i].cRAE->trainingData.insert(s, e);
threadpara[i].instance_num = batchsize;
}
}
pthread_t* pt = new pthread_t[RAEThreadNum];
for (int a = 0; a < RAEThreadNum; a++) pthread_create(&pt[a], NULL, RAELBFGS::deepThread, (void *)(threadpara + a));
for (int a = 0; a < RAEThreadNum; a++) pthread_join(pt[a], NULL);
for(int i = 0; i < RAEThreadNum; i++)
{
fx += threadpara[i].lossVal;
for(int elem = 0; elem < getRAEWeightSize(); elem++)
{
g[elem] += threadpara[i].g[elem];
}
}
/*
int internal_node_num = 0;
for(map<string, int>::iterator it = trainingData.begin(); it != trainingData.end(); it++)
{
internal_node_num += getInternalNode(it->first);
}*/
fx /= trainingData.size();
fx += ZETA * decay();
for(int elem = 0; elem < getRAEWeightSize(); elem++)
{
g[elem] /= trainingData.size();
}
delWeight1 += ZETA * weights1;
delWeight1_b.setZero();
delWeight2 += ZETA * weights2;
delWeight2_b.setZero();
update(g);
delete pt;
pt = NULL;
for(int i = 0; i < RAEThreadNum; i++)
{
lbfgs_free(threadpara[i].g);
delete threadpara[i].cRAE;
}
delete threadpara;
threadpara = NULL;
return fx;
}
void operate(N n){ //do what needs to be done at this neuron
checkAction(n); //check for action potential
decay(n); //connections naturally decay, mechanism of pruning;
return;
}
int main(int argc, char** argv){
if(tomahawk::utility::IsBigEndian()){
std::cerr << tomahawk::utility::timestamp("ERROR") << "Tomahawk does not support big endian systems..." << std::endl;
return(1);
}
if(argc == 1){
tomahawk::ProgramMessage();
tomahawk::ProgramHelpDetailed();
return(1);
}
// Literal string input line
tomahawk::LITERAL_COMMAND_LINE = tomahawk::TOMAHAWK_PROGRAM_NAME;
for(int i = 1; i < argc; ++i)
tomahawk::LITERAL_COMMAND_LINE += " " + std::string(&argv[i][0]);
if(strcmp(&argv[1][0], "import") == 0){
return(import(argc, argv));
}
else if(strcmp(&argv[1][0], "calc") == 0){
return(calc(argc, argv));
} else if(strcmp(&argv[1][0], "calc-single") == 0 || strcmp(&argv[1][0], "scalc") == 0){
return(scalc(argc, argv));
}
else if(strcmp(&argv[1][0], "view") == 0){
return(view(argc, argv));
}
else if(strncmp(&argv[1][0], "sort", 4) == 0){
return(sort(argc, argv));
}
else if(strncmp(&argv[1][0], "concat", 6) == 0){
return(concat(argc, argv));
}
else if(strncmp(&argv[1][0], "stats", 5) == 0){
return(stats(argc, argv));
}
else if(strncmp(&argv[1][0], "aggregate", 8) == 0){
return(aggregate(argc, argv));
}
else if(strncmp(&argv[1][0], "haplotype", 9) == 0){
return(haplotype(argc, argv));
}
else if(strncmp(&argv[1][0], "relationship", 9) == 0){
return(relationship(argc, argv));
}
else if(strncmp(&argv[1][0], "decay", 5) == 0){
return(decay(argc, argv));
}
else if(strcmp(&argv[1][0], "--version") == 0 || strcmp(&argv[1][0], "version") == 0){
tomahawk::ProgramMessage(false);
return(0);
} else if(strcmp(&argv[1][0], "--help") == 0 || strcmp(&argv[1][0], "help") == 0){
tomahawk::ProgramMessage();
tomahawk::ProgramHelpDetailed();
return(0);
} else {
tomahawk::ProgramMessage();
tomahawk::ProgramHelpDetailed();
std::cerr << tomahawk::utility::timestamp("ERROR") << "Illegal command" << std::endl;
return(1);
}
return(1);
}
void buildModel(RooWorkspace& w,int chooseFitParams, int chooseSample,int whatBin, int signalModel, int bkgdModel, int doRap, int doPt,int doCent,int useRef,float muonPtMin, int fixFSR){
// C r e a t e m o d e l
int nt=100000;
// cout << "you're building a model for the quarkonium resonance of mass = "<< M1S <<" GeV/c^{2},"endl;
RooRealVar *nsig1f = new RooRealVar("N_{ #varUpsilon(1S)}","nsig1S",0,nt*10);
RooRealVar* mass = new RooRealVar("invariantMass","#mu#mu mass",mass_l,mass_h,"GeV/c^{2}");
RooRealVar *nsig2f = NULL;
RooRealVar *nsig3f = NULL;
switch (chooseFitParams)
{
case 0://use the YIELDs of 2S and 3S as free parameters
//minor modif here: 3S forced positive.
nsig2f = new RooRealVar("N_{ #varUpsilon(2S)}","nsig2S", nt*0.25,-200,10*nt);
nsig3f = new RooRealVar("N_{ #varUpsilon(3S)}","nsig3S", nt*0.25,-200,10*nt);
cout << "you're fitting to extract yields, "<< endl;
break;
default:
cout<<"Make a pick from chooseFitParams!!!"<<endl;
break;
}
RooRealVar *mean = new RooRealVar("m_{ #varUpsilon(1S)}","#Upsilon mean",M1S,M1S-0.2,M1S+0.2);
RooConstVar *rat2 = new RooConstVar("rat2", "rat2", M2S/M1S);
RooConstVar *rat3 = new RooConstVar("rat3", "rat3", M3S/M1S);
// scale mean and resolution by mass ratio
RooFormulaVar *mean1S = new RooFormulaVar("mean1S","@0",RooArgList(*mean));
RooFormulaVar *mean2S = new RooFormulaVar("mean2S","@0*@1", RooArgList(*mean,*rat2));
RooFormulaVar *mean3S = new RooFormulaVar("mean3S","@0*@1", RooArgList(*mean,*rat3));
// //detector resolution ?? where is this coming from?
RooRealVar *sigma1 = new RooRealVar("#sigma_{CB1}","#sigma_{CB1}",sigma_min[whatBin],sigma_max[whatBin]); //
RooFormulaVar *sigma1S = new RooFormulaVar("sigma1S","@0" ,RooArgList(*sigma1));
RooFormulaVar *sigma2S = new RooFormulaVar("sigma2S","@0*@1",RooArgList(*sigma1,*rat2));
RooFormulaVar *sigma3S = new RooFormulaVar("sigma3S","@0*@1",RooArgList(*sigma1,*rat3));
RooRealVar *alpha = new RooRealVar("#alpha_{CB}","tail shift",alpha_min[whatBin],alpha_max[whatBin]); // MC 5tev 1S pol2
RooRealVar *npow = new RooRealVar("n_{CB}","power order",npow_min[whatBin],npow_max[whatBin]); // MC 5tev 1S pol2
RooRealVar *sigmaFraction = new RooRealVar("sigmaFraction","Sigma Fraction",0.,1.);
// scale the sigmaGaus with sigma1S*scale=sigmaGaus now.
RooRealVar *scaleWidth = new RooRealVar("#sigma_{CB2}/#sigma_{CB1}","scaleWidth",1.,2.5);
RooFormulaVar *sigmaGaus = new RooFormulaVar("sigmaGaus","@0*@1", RooArgList(*sigma1,*scaleWidth));
RooFormulaVar *sigmaGaus2 = new RooFormulaVar("sigmaGaus","@0*@1*@2", RooArgList(*sigma1,*scaleWidth,*rat2));
RooFormulaVar *sigmaGaus3 = new RooFormulaVar("sigmaGaus","@0*@1*@2", RooArgList(*sigma1,*scaleWidth,*rat3));
RooGaussian* gauss1 = new RooGaussian("gaus1s","gaus1s",
*nsig1f,
*mass, //mean
*sigmaGaus); //sigma
// RooGaussian* gauss1b = new RooGaussian("gaus1sb","gaus1sb",
// *nsig1f,
// *m, //mean
// *sigma1); //sigma
switch(signalModel){
case 1: //crystal boule
RooCBShape *sig1S = new RooCBShape ("cb1S_1", "FSR cb 1s",
*mass,*mean1S,*sigma1,*alpha,*npow);
RooCBShape *sig2S = new RooCBShape ("cb2S_1", "FSR cb 1s",
*mass,*mean2S,*sigma2S,*alpha,*npow);
RooCBShape *sig3S = new RooCBShape ("cb3S_1", "FSR cb 1s",
*mass,*mean3S,*sigma3S,*alpha,*npow);
cout << "you're fitting each signal peak with a Crystal Ball function"<< endl;
break;
case 2: //Gaussein
RooAbsPdf *sig1S = new RooGaussian ("g1", "gaus 1s",
*mass,*mean1S,*sigma1);
cout << "you're fitting 1 signal peak with a Gaussian function"<< endl;
break;
case 3: //Gaussein + crystal boule
RooCBShape *cb1S_1 = new RooCBShape ("cb1S_1", "FSR cb 1s",
*mass,*mean1S,*sigma1,*alpha,*npow);
RooAddPdf *sig1S = new RooAddPdf ("cbg", "cbgaus 1s",
RooArgList(*gauss1,*cb1S_1),*sigmaFraction);
cout << "you're fitting 1 signal peak with a sum of a Gaussian and a Crystal Ball function"<< endl;
break;
case 4: //crystal boules
RooCBShape *cb1S_1 = new RooCBShape ("cb1S_1", "FSR cb 1s",
*mass,*mean1S,*sigma1,*alpha,*npow);
RooCBShape *cb1S_2 = new RooCBShape ("cb1S_2", "FSR cb 1s",
*mass,*mean1S,*sigmaGaus,*alpha,*npow);
RooAddPdf *sig1S = new RooAddPdf ("cbcb","1S mass pdf",
RooArgList(*cb1S_1,*cb1S_2),*sigmaFraction);
// /// Upsilon 2S
RooCBShape *cb2S_1 = new RooCBShape ("cb2S_1", "FSR cb 2s",
*mass,*mean2S,*sigma2S,*alpha,*npow);
RooCBShape *cb2S_2 = new RooCBShape ("cb2S_2", "FSR cb 2s",
*mass,*mean2S,*sigmaGaus2,*alpha,*npow);
RooAddPdf *sig2S = new RooAddPdf ("sig2S","2S mass pdf",
RooArgList(*cb2S_1,*cb2S_2),*sigmaFraction);
// /// Upsilon 3S
RooCBShape *cb3S_1 = new RooCBShape ("cb3S_1", "FSR cb 3s",
*mass,*mean3S,*sigma3S,*alpha,*npow);
RooCBShape *cb3S_2 = new RooCBShape ("cb3S_2", "FSR cb 3s",
*mass,*mean3S,*sigmaGaus3,*alpha,*npow);
RooAddPdf *sig3S = new RooAddPdf ("sig3S","3S mass pdf",
RooArgList(*cb3S_1,*cb3S_2),*sigmaFraction); // = cb3S1*sigmaFrac + cb3S2*(1-sigmaFrac)
cout << "you're fitting each signal peak with a Double Crystal Ball function"<< endl;
break;
case 5: //deux Gausseins
RooAddPdf *sig1S = new RooAddPdf ("cb1S_1", "cbgaus 1s",
RooArgList(*gauss1,*gauss1b),*sigmaFraction);
cout << "you're fitting each signal peak with a Double Gaussian function"<< endl;
break;
}
// bkg Chebychev
RooRealVar *nbkgd = new RooRealVar("n_{Bkgd}","nbkgd",0,nt);
RooRealVar *bkg_a1 = new RooRealVar("a1_bkg", "bkg_{a1}", 0, -5, 5);
RooRealVar *bkg_a2 = new RooRealVar("a2_Bkg", "bkg_{a2}", 0, -2, 2);
RooRealVar *bkg_a3 = new RooRealVar("a3_Bkg", "bkg_{a3}", 0, -0.9, 2);
// likesign
RooRealVar *nLikesignbkgd = new RooRealVar("NLikesignBkg","nlikesignbkgd",nt*0.75,0,10*nt);
// *************************************************** bkgModel
RooRealVar turnOn("turnOn","turnOn", turnOn_minCent[whatBin],turnOn_maxCent[whatBin]);
RooRealVar width("width","width",width_minCent[whatBin],width_maxCent[whatBin]);// MB 2.63
RooRealVar decay("decay","decay",decay_minCent[whatBin],decay_maxCent[whatBin]);// MB: 3.39
if (doRap && !doPt)
{
RooRealVar turnOn("turnOn","turnOn", turnOn_minRap[whatBin],turnOn_maxRap[whatBin]);
RooRealVar width("width","width",width_minRap[whatBin],width_maxRap[whatBin]);// MB 2.63
RooRealVar decay("decay","decay",decay_minRap[whatBin],decay_maxRap[whatBin]);// MB: 3.39
}
if (doPt && !doRap)
{
RooRealVar turnOn("turnOn","turnOn", turnOn_minPt[whatBin],turnOn_maxPt[whatBin]);
RooRealVar width("width","width",width_minPt[whatBin],width_maxPt[whatBin]);// MB 2.63
RooRealVar decay("decay","decay",decay_minPt[whatBin],decay_maxPt[whatBin]);// MB: 3.39
}
width.setConstant(false);
decay.setConstant(false);
turnOn.setConstant(false);
switch (useRef)// no reference
{
case 0: // forcing sigma and fsr to be left free.
fixSigma1 = 0;
fixFSR = 0;
break;
case 1:
//using data-driven estimates
int dataRef=1;
cout<<"You're using the debug mode based on data parameters. So you must not take this result as the central one."<<endl;
break;
case 2:
cout << "doCent="<<doCent << endl;
//using MC-driven estimates
int dataRef=2;
if(doCent) //MB values, assumed to be the same with all centralities...
{
if(muonPtMin <4){
gROOT->LoadMacro("dataTable_loose.h");
}else if(muonPtMin > 3.5){
gROOT->LoadMacro("dataTable_tight.h");
}
npow->setVal(npow_rapBins[8]);
alpha->setVal(alpha_rapBins[8]);
sigma1->setVal(sigma1_rapBins[8]);
scaleWidth->setVal(scale_rapBins[8]);
sigmaFraction->setVal(pdFrac_rapBins[8]);
cout<< whatBin << endl;
}
if(doRap && !doPt)
{
if(muonPtMin <4){
gROOT->LoadMacro("dataTable_loose.h");
}else if(muonPtMin > 3.5){
gROOT->LoadMacro("dataTable_tight.h");
}
npow->setVal(npow_rapBins[whatBin]);
alpha->setVal(alpha_rapBins[whatBin]);
sigma1->setVal(sigma1_rapBins[whatBin]);
scaleWidth->setVal(scale_rapBins[whatBin]);
sigmaFraction->setVal(pdFrac_rapBins[whatBin]);
cout<< whatBin << endl;
}
if(doPt && !doRap)
{
// cout << "we're here" << endl;
if(muonPtMin <4){
gROOT->LoadMacro("dataTable_loose.h");
}else if(muonPtMin > 3.5){
gROOT->LoadMacro("dataTable_tight.h");
}
cout << " ok ... " <<endl;
npow->setVal(npow_ptBins[whatBin]);
alpha->setVal(alpha_ptBins[whatBin]);
sigma1->setVal(sigma1_ptBins[whatBin]);
scaleWidth->setVal(scale_ptBins[whatBin]);
sigmaFraction->setVal(pdFrac_ptBins[whatBin]);
}
cout<<"You're using MC parameters. So you may use this result as the central one, according to the LLR test outcome."<<endl;
break;
default: break;
}
//
cout << "npow tried=" << npow->getVal();
if(fixFSR==3 || fixFSR==1) cout << " constant!" << endl; else cout << " floating!" << endl;
cout << "alpha tried=" << alpha->getVal();
if(fixFSR==2 || fixFSR==1) cout << " constant!" << endl; else cout << " floating!" << endl;
cout << "sigma1 tried=" << sigma1->getVal();
if(fixFSR==4 || fixFSR==1) cout << " constant!" << endl; else cout << " floating!" << endl;
cout << "scale tried=" << scaleWidth->getVal();
if(fixFSR==4 || fixFSR==1) cout << " constant!" << endl; else cout << " floating!" << endl;
cout << "normalisation tried=" << sigmaFraction->getVal();
if(fixFSR==5 || fixFSR==1) cout << " constant!" << endl; else cout << " floating!" << endl;
switch (fixFSR) // 0: free; 1: both fixed 2: alpha fixed 3: npow fixed
{
case 0:// all free
alpha->setConstant(false);
npow->setConstant(false);
sigma1->setConstant(false);
scaleWidth->setConstant(false);
sigmaFraction->setConstant(false);
break;
case 1:// all fixed
alpha->setConstant(true);
npow ->setConstant(true);
sigma1->setConstant(true);
scaleWidth->setConstant(true);
sigmaFraction->setConstant(true);
break;
case 2: // release alpha
alpha->setConstant(false);
npow ->setConstant(true);
sigma1->setConstant(true);
scaleWidth->setConstant(true);
sigmaFraction->setConstant(true);
break;
case 3:// npow released
alpha->setConstant(true);
npow->setConstant(false);
sigma1->setConstant(true);
scaleWidth->setConstant(true);
sigmaFraction->setConstant(true);
break;
case 4:// width+ sF +scale released
alpha->setConstant(true);
npow->setConstant(true);
sigma1->setConstant(false);
scaleWidth->setConstant(true);
sigmaFraction->setConstant(true);
break;
case 5:// scale +sF
alpha->setConstant(true);
npow->setConstant(true);
sigma1->setConstant(true);
scaleWidth->setConstant(false);
sigmaFraction->setConstant(true);
break;
case 6:// scale +sF
alpha->setConstant(true);
npow->setConstant(true);
sigma1->setConstant(true);
scaleWidth->setConstant(true);
sigmaFraction->setConstant(false);
break;
default:
cout<<"Donno this choice! Pick somehting for FSR parameters that I know"<<endl;
break;
}
//thisPdf: form of the bkg pdf
//pdf_combinedbkgd; // total bkg pdf. usually form*normalization (so that you can do extended ML fits)
switch (bkgdModel)
{
case 1 : //(erf*exp ) to fit the SS, then fix the shape and fit OS, in case of constrain option
bkg_a3->setConstant(true);
RooGenericPdf *ErrPdf = new RooGenericPdf("ErrPdf","ErrPdf",
"exp(-@0/decay)*(TMath::Erf((@0-turnOn)/width)+1)",
RooArgList(*mass,turnOn,width,decay));
RooFitResult* fit_1st = ErrPdf->fitTo(*likesignData,Save(),NumCPU(4)) ; // likesign data
if (doTrkRot) fit_1st = thisPdf->fitTo(*TrkRotData,Save(),NumCPU(4)) ;
if (doConstrainFit)
{ // allow parameters to vary within cenral value from above fit + their sigma
turnOn_constr = new RooGaussian("turnOn_constr","turnOn_constr",
turnOn,
RooConst(turnOn.getVal()),
RooConst(turnOn.getError()));
width_constr = new RooGaussian("width_constr","width_constr",
width, RooConst(width.getVal()),
RooConst(width.getError()));
decay_constr = new RooGaussian("decay_constr","decay_constr",
decay,
RooConst(decay.getVal()),
RooConst(decay.getError()));
}
else
{
turnOn.setConstant(kTRUE);
width.setConstant(kTRUE);
decay.setConstant(kTRUE);
}
RooRealVar *fLS =new RooRealVar("R_{SS/OS}","Empiric LS/SS ratio",0.,1.);
RooAbsPdf *ChebPdf = new RooChebychev("ChebPdf","ChebPdf",
*mass, RooArgList(*bkg_a1,*bkg_a2));
RooAbsPdf *pdf_combinedbkgd = new RooAddPdf ("bkgPdf","total combined background pdf",
RooArgList(*ErrPdf,*ChebPdf),
RooArgList(*fLS));
break;
case 2 : //us eRooKeysPdf to smooth the SS, then fit OS with pol+keys
bkg_a3->setConstant(true);
RooRealVar *fLS =new RooRealVar("R_{SS/OS}","Empiric LS/SS ratio",0.,1.);
RooKeysPdf *KeysPdf = new RooKeysPdf("KeysPdf","KeysPdf",*mass,*likesignData,
RooKeysPdf::MirrorBoth, 1.4);
RooAbsPdf *ChebPdf = new RooChebychev("ChebPdf","ChebPdf",
*mass, RooArgList(*bkg_a1,*bkg_a2));
if (doTrkRot) thisPdf = new RooKeysPdf("thisPdf","thisPdf",*mass,*TrkRotData,
RooKeysPdf::MirrorBoth, 1.4);
RooAbsPdf *pdf_combinedbkgd = new RooAddPdf ("bkgPdf","total combined background pdf",
RooArgList(*KeysPdf,*ChebPdf),
RooArgList(*fLS));
break;
case 3 : //use error function to fit the OS directly
bkg_a3->setConstant(true);
RooAbsPdf *pdf_combinedbkgd = new RooGenericPdf("bkgPdf","bkgPdf",
"exp(-@0/decay)*(TMath::Erf((@0-turnOn)/width)+1)",
RooArgList(*mass,turnOn,width,decay));
break;
case 4 : //use pol 2+ErfExp to fit the OS directly
RooRealVar *fPol = new RooRealVar("F_{pol}","fraction of polynomial distribution",0.0,1);
RooAbsPdf *ChebPdf = new RooChebychev("ChebPdf","ChebPdf",
*mass, RooArgList(*bkg_a1,*bkg_a2));
RooGenericPdf *ErrPdf = new RooGenericPdf("ErrPdf","ErrPdf",
"exp(-@0/decay)*(TMath::Erf((@0-turnOn)/width)+1)",
RooArgList(*mass,turnOn,width,decay));
RooAbsPdf *pdf_combinedbkgd = new RooAddPdf ("bkgPdf","total combined background pdf",
RooArgList(*ChebPdf,*ErrPdf),
RooArgList(*fPol));
break;
case 5 : //use ( error function + polynomial 1) to fit the OS directly
bkg_a3->setConstant(true);
bkg_a2->setConstant(true);
RooRealVar *fPol = new RooRealVar("F_{pol}","fraction of polynomial distribution",0.0,1);
RooAbsPdf *ChebPdf = new RooChebychev("ChebPdf","ChebPdf",
*mass, RooArgList(*bkg_a1,*bkg_a2,*bkg_a3));
RooGenericPdf *ErrPdf = new RooGenericPdf("ErrPdf","ErrPdf",
"exp(-@0/decay)*(TMath::Erf((@0-turnOn)/width)+1)",
RooArgList(*mass,turnOn,width,decay));
RooAbsPdf *pdf_combinedbkgd = new RooAddPdf ("bkgPdf","total combined background pdf",
RooArgList(*ChebPdf,*ErrPdf),
RooArgList(*fPol));
break;
case 6: // NOT WORKING
RooRealVar *fPol = new RooRealVar("F_{pol}","fraction of polynomial distribution",0.0,1);
RooAbsPdf *ChebPdf = new RooChebychev("ChebPdf","ChebPdf",
*mass, RooArgList(*bkg_a1,*bkg_a2));
RooGenericPdf *ExpPdf = new RooGenericPdf("ExpPdf","ExpPdf",
"exp(-@0/decay)",
RooArgList(*mass,decay));
RooAbsPdf *pdf_combinedbkgd = new RooAddPdf ("bkgPdf","total combined background pdf",
RooArgList(*ChebPdf,*ExpPdf),
RooArgList(*fPol));
break;
default :
cout<<"Donno what you are talking about! Pick another fit option!"<<endl;
break;
}
//###### the nominal fit with default pdf
// RooAbsPdf *pdf; // nominal PDF
if(chooseSample==8)
{
// bkg_a1->setVal(0);// can be turned on at convenience
// bkg_a1->setConstant();
// bkg_a2->setVal(0);
// bkg_a2->setConstant();
// bkg_a3->setVal(0);
// bkg_a3->setConstant();
// RooAbsPdf *pdf = new RooAddPdf ("pdf","total p.d.f.",
// RooArgList(*sig1S,*pdf_combinedbkgd),
// RooArgList(*nsig1f,*nbkgd));
RooAbsPdf *pdf = new RooAddPdf ("pdf","total p.d.f.",*sig1S,*nsig1f);
} else if(chooseSample!=8)
{
// can remove the double crystal ball in pbpb: just commenting out and copying an appropriate version
RooAbsPdf *pdf = new RooAddPdf ("pdf","total p.d.f.",
RooArgList(*sig1S,*sig2S,*sig3S,*pdf_combinedbkgd),
RooArgList(*nsig1f,*nsig2f,*nsig3f,*nbkgd));
// nsig3f->setVal(0); nsig3f->setConstant();
}
w.import(*pdf);
w.Print();
}