/* Corresponds to Rule 6 of Trickle */
static void
on_interval_expired(void *ptr)
{
clock_time_t half_interval_size;
if(trickle_doublings < IMAX) {
trickle_doublings++;
PRINTF("akes-trickle: Doubling interval size\n");
}
half_interval_size = interval_size() / 2;
new_nbrs_count = 0;
counter = 0;
ctimer_set(&trickle_timer,
half_interval_size + ((half_interval_size * random_rand()) / RANDOM_RAND_MAX),
on_timeout,
NULL);
PRINTF("akes-trickle: I=%lus t=%lus\n",
interval_size()/CLOCK_SECOND,
trickle_timer.etimer.timer.interval/CLOCK_SECOND);
}
dirac_sequence generate_dirac_sequence(double speed_of_sound,
double room_volume,
double sample_rate,
double max_time) {
const auto constant_mean_occurrence =
constant_mean_event_occurrence(speed_of_sound, room_volume);
std::default_random_engine engine{std::random_device{}()};
util::aligned::vector<float> ret(std::ceil(max_time * sample_rate), 0);
for (auto t = t0(constant_mean_occurrence); t < max_time;
t += interval_size(
engine, mean_event_occurrence(constant_mean_occurrence, t))) {
const auto sample_index = t * sample_rate;
const size_t twice = 2 * sample_index;
const bool negative = (twice % 2) != 0;
ret[sample_index] = negative ? -1 : 1;
}
return {ret, sample_rate};
}
/* Corresponds to Rule 4 of Trickle */
static void
on_timeout(void *ptr)
{
if(counter < REDUNDANCY_CONSTANT) {
PRINTF("akes-trickle: Broadcasting HELLO\n");
akes_broadcast_hello();
reset_trickle_info();
} else {
PRINTF("akes-trickle: Suppressed HELLO\n");
}
ctimer_set(&trickle_timer,
round_up(interval_size() - trickle_timer.etimer.timer.interval),
on_interval_expired,
NULL);
ctimer_set(&hello_timer,
(AKES_HELLO_DURATION * CLOCK_SECOND),
on_hello_done,
NULL);
}
int main(int argc, char*argv[]){
try{
defaultUnitsc = &stdPhysUnits;
fittable_gaussianfn mygauss;
mygauss.rename_var("x_0", "t_0")
.rename_var("A", "\\rho_0")
// .var("\\sigma", "\\sigma")
.rename_var("x", "t");
unsigned npeaks = 1;
// cout << "Arguments given: " << argc << endl;
if (argc == 2) {
cout << "Requested number of peaks: " << argv[1] << endl;
npeaks = argv[1][0] - '0';
}
measureseq ps(npeaks);
QTeXdiagmaster outgr("somegaussian.qda");
tr1::variate_generator<tr1::mt19937, tr1::normal_distribution<> >
cdice (std::tr1::mt19937(time(NULL)), std::tr1::normal_distribution<>(0, 1));
physquantity noiselvl = 1./8 * volts, noisenormale = (37*millivolts) (volts),
peaksheightsortof = 2*volts;
tr1::variate_generator<tr1::mt19937, tr1::normal_distribution<> >
noise (std::tr1::mt19937(time(NULL)), std::tr1::normal_distribution<>(0, (noiselvl/noisenormale).dbl()));
int c=3;
measureseq fnplot(0), fnpplot(0), frr, fsum;
physquantity biggestt=0, smallestt=0, smallestrho = 15*tonnes;
// while (c < 4) {
for (measureseq::access p = ps.begin(); p!=ps.end(); ++p) {
p["\\sigma"] = abs((cdice() + 1.6) * .4*seconds);
p["t_0"] = cdice() * npeaks*seconds;
p["t_{0p}"] = p["t_0"] + cdice() * p["\\sigma"];
p["\\rho_0"] = abs(cdice() * peaksheightsortof/2) + abs(cdice() * peaksheightsortof);
biggestt.push_upto(p["t_0"]);
smallestt.push_downto(p["t_0"]);
}
cout << "Given:\n" << ps << endl;
fnplot.clear();
measureseq::access p = ps.begin();
smallestrho = noiselvl * p["\\rho_0"];
smallestt -= 13*p["\\sigma"];
biggestt += 13*p["\\sigma"];
physquantity rastert = (20*milliseconds) (seconds);
for (p["t"] = smallestt; p["t"]<biggestt; p["t"]+=rastert){
p["\\rho"] = 0;
p["\\rho_\\text{noisefree}"] = 0;
for (measureseq::access q = p; q!=ps.end(); ++q) {
q["t"] = p["t"];
p["\\rho"] += q["\\rho_0"]/(1+((q["t"]-q["t_0"])/q["\\sigma"]).squared().dbl()) + noise()*noisenormale;
p["\\rho_\\text{noisefree}"] += q["\\rho_0"]/(1+((q["t"]-q["t_0"])/q["\\sigma"]).squared().dbl());
// p["\\rho"] += mygauss(*q) + noise()*noisenormale;
}
// p["\\rho"].push_upto(0);
fnplot.push_back(measure(p["\\rho"], p["t"]));
fnpplot.push_back(measure(p["\\rho_\\text{noisefree}"], p["t"]));
}
smallestt -= 1*p["\\sigma"];
biggestt += 1*p["\\sigma"];
outgr.insertCurve(fnplot, captfinder("t"), captfinder("\\rho"), /*"gaussianssum.qcv",*/ QTeXgrcolors::blue);
outgr.insertCurve(fnpplot, captfinder("t"), captfinder("\\rho_\\text{noisefree}"), QTeXgrcolors::i_red);
outgr.nextcolor(QTeXgrcolors::green);
// cout << "The correct values:\n" << p << endl;
fittable_multigaussianfn fgfn(npeaks);
fgfn//.rename_var("x", "t")
.rename_var("x, x_i, A_i", "t, t_i, \\rho_i", LaTeXindex("i").from(0).unto(npeaks))
//.rename_var("A_i", "\\rho_i", LaTeXindex("i").from(0).unto(npeaks))
;
captfinder rhofind("\\rho");
fitdist_fntomeasures d(&fnplot, &fgfn, &rhofind);
cout << "t_i in [" << smallestt << ", " << biggestt << "]\n";
evolution_minimizer fitthisback(d,
enforce_each_difference_smaller_than( "t_i" | LaTeXindex("i").from(0).unto(npeaks),
"\\sigma_i" | LaTeXindex("i").from(0).unto(npeaks),
maxval(fnplot, captfinder("t")) )
&& enforce_each_sum_bigger_than( "t_i" | LaTeXindex("i").from(0).unto(npeaks),
"\\sigma_i" | LaTeXindex("i").from(0).unto(npeaks),
minval(fnplot, captfinder("t")) )
&& enforce_each_smaller_than( "\\sigma_i" | LaTeXindex("i").from(0).unto(npeaks), abs(biggestt-smallestt) )
&& enforce_each_bigger_than( "\\sigma_i" | LaTeXindex("i").from(0).unto(npeaks), 2*rastert )
&& enforce_each_bigger_than( "\\rho_i" | LaTeXindex("i").from(0).unto(npeaks), noiselvl*2 ),
evolution_minimizer::solutioncertaintycriteria::doublevalue()
);
measure fttt = fitthisback.result();
measureseq fttps = fttt.pack_subscripted();
cout << "Result:\n" << fttps << endl;
phq_interval totalfitdomain;
physquantity &ldomb = totalfitdomain.l(), &rdomb=totalfitdomain.r();
for (measureseq::access fttp=fttps.begin(); fttp!=fttps.end(); ++fttp){
fttp["\\sigma"] = abs(fttp["\\sigma"]);
fttp["t"].label("t_0");
fttp["\\rho_0"] = physquantity(fttp["\\rho"]);
phq_interval thisgaussdomain(
centered_about(fttp["t_0"]),
interval_size(3*fttp["\\sigma"]),
"t" );
totalfitdomain.widen_to_include(thisgaussdomain);
outgr.plot_phmsq_function(mygauss, *fttp, thisgaussdomain);
#if 0
frr.clear();
cout << "Plot from " << fttp["t_0"]-3*fttp["\\sigma"] << " to " << fttp["t_0"]+3*fttp["\\sigma"] << endl;
for (fttp["t"] = fttp["t_0"]-3*fttp["\\sigma"]; fttp["t"]<fttp["t_0"]+3*fttp["\\sigma"]; fttp["t"]+=.01*fttp["\\sigma"]){
ldomb.push_downto(fttp["t"]);
rdomb.push_upto(fttp["t"]);
fttp["\\rho"] = mygauss(*fttp);
frr.push_back(*fttp);
}
/* outgr.insertCurve(frr, captfinder("t"), captfinder("\\rho"), "gaussianfitted"+std::string(1,char(c+44))+".qcv", */c++/* % 8 + 1)*/;
outgr.insertCurve(frr, captfinder("t"), captfinder("\\rho")/*, "gaussianfitted"+std::string(1,char(c+44))+".qcv", QTeXgrcolors::defaultsequence(c++)*/);
#endif
}
frr.clear();
#if 1
// outgr.plot_phmsq_function(fgfn, fttt, totalfitdomain, QTeXgrcolors::i_red);
#else
cout << "Plot sum from " << ldomb << " to " << rdomb << endl;
for (physquantity t=ldomb; t<rdomb; t += (rdomb-ldomb)/1024) {
frr.push_back(measure(t));
frr.back().push_back(physquantity(0).wlabel("\\sum\\rho"));
for (measureseq::access fttp=fttps.begin(); fttp!=fttps.end(); ++fttp){
fttp["t"] = t;
frr.back()["\\sum\\rho"] += mygauss(*fttp);
}
}
outgr.insertCurve(frr, captfinder("t"), captfinder("\\sum\\rho"), /*"gaussianfittedsum.qcv",*/ QTeXgrcolors::i_red);
#endif
// }
}
catch(unitconversion Utrouble) {
cerr << "Strange stuff with units " << Utrouble.Dim[0]->uName << " and " << Utrouble.Dim[1]->uName << endl; cerr.flush();
abort();
}
catch(phUnit phU) {
cerr << "Strange stuff with unit " << phU.uName << endl; cerr.flush();
abort();
}
catch(std::string abscent) {
cerr << "Name " << abscent << " seems abscent.\n"; cerr.flush();
}
return (0);
}
