void create_accelobs(accelobs * obs, infodata * idata, Cmdline * cmd, int usemmap)
{
int ii, rootlen, input_shorts = 0;
{
int hassuffix = 0;
char *suffix;
hassuffix = split_root_suffix(cmd->argv[0], &(obs->rootfilenm), &suffix);
if (hassuffix) {
if (strcmp(suffix, "fft") != 0 &&
strcmp(suffix, "dat") != 0 && strcmp(suffix, "sdat") != 0) {
printf("\nInput file ('%s') must be an '.fft' or '.[s]dat' file!\n\n",
cmd->argv[0]);
free(suffix);
exit(0);
}
/* If the input file is a time series */
if (strcmp(suffix, "dat") == 0 || strcmp(suffix, "sdat") == 0) {
obs->dat_input = 1;
obs->mmap_file = 0;
if (strcmp(suffix, "sdat") == 0)
input_shorts = 1;
} else {
obs->dat_input = 0;
}
free(suffix);
} else {
printf("\nInput file ('%s') must be an '.fft' or '.[s]dat' file!\n\n",
cmd->argv[0]);
exit(0);
}
}
if (cmd->noharmpolishP)
obs->use_harmonic_polishing = 0;
else
obs->use_harmonic_polishing = 1; // now default
/* Read the info file */
readinf(idata, obs->rootfilenm);
if (idata->object) {
printf("Analyzing %s data from '%s'.\n\n",
remove_whitespace(idata->object), cmd->argv[0]);
} else {
printf("Analyzing data from '%s'.\n\n", cmd->argv[0]);
}
/* Prepare the input time series if required */
if (obs->dat_input) {
FILE *datfile;
long long filelen;
float *ftmp;
printf("Reading and FFTing the time series...");
fflush(NULL);
datfile = chkfopen(cmd->argv[0], "rb");
/* Check the length of the file to see if we can handle it */
filelen = chkfilelen(datfile, sizeof(float));
if (input_shorts)
filelen *= 2;
if (filelen > 67108864) { /* Small since we need memory for the templates */
printf("\nThe input time series is too large. Use 'realfft' first.\n\n");
exit(0);
}
/* Read the time series into a temporary buffer */
/* Note: The padding allows us to search very short time series */
/* using correlations without having to worry about */
/* accessing data before or after the valid FFT freqs. */
if (input_shorts) {
short *stmp = gen_svect(filelen);
ftmp = gen_fvect(filelen+2*ACCEL_PADDING);
for (ii = 0; ii < ACCEL_PADDING; ii++) {
ftmp[ii] = 0.0;
ftmp[ii+filelen+ACCEL_PADDING] = 0.0;
}
chkfread(stmp, sizeof(short), filelen, datfile);
for (ii = 0; ii < filelen; ii++)
ftmp[ii+ACCEL_PADDING] = (float) stmp[ii];
free(stmp);
} else {
ftmp = read_float_file(datfile, -ACCEL_PADDING, filelen+2*ACCEL_PADDING);
}
/* Now, offset the pointer so that we are pointing at the first */
/* bits of valid data. */
ftmp += ACCEL_PADDING;
fclose(datfile);
/* FFT it */
realfft(ftmp, filelen, -1);
obs->fftfile = NULL;
obs->fft = (fcomplex *) ftmp;
obs->numbins = filelen / 2;
printf("done.\n");
/* De-redden it */
printf("Removing red-noise...");
deredden(obs->fft, obs->numbins);
printf("done.\n\n");
}
/* Determine the output filenames */
rootlen = strlen(obs->rootfilenm) + 25;
obs->candnm = (char *) calloc(rootlen, 1);
obs->accelnm = (char *) calloc(rootlen, 1);
obs->workfilenm = (char *) calloc(rootlen, 1);
sprintf(obs->candnm, "%s_ACCEL_%d.cand", obs->rootfilenm, cmd->zmax);
sprintf(obs->accelnm, "%s_ACCEL_%d", obs->rootfilenm, cmd->zmax);
sprintf(obs->workfilenm, "%s_ACCEL_%d.txtcand", obs->rootfilenm, cmd->zmax);
/* Open the FFT file if it exists appropriately */
if (!obs->dat_input) {
obs->fftfile = chkfopen(cmd->argv[0], "rb");
obs->numbins = chkfilelen(obs->fftfile, sizeof(fcomplex));
if (usemmap) {
fclose(obs->fftfile);
obs->fftfile = NULL;
printf("Memory mapping the input FFT. This may take a while...\n");
obs->mmap_file = open(cmd->argv[0], O_RDONLY);
if (obs->mmap_file == -1) {
perror("\nError in open() in accel_utils.c");
printf("\n");
exit(-1);
}
obs->fft = (fcomplex *) mmap(0, sizeof(fcomplex) * obs->numbins, PROT_READ,
MAP_SHARED, obs->mmap_file, 0);
if (obs->fft == MAP_FAILED) {
perror("\nError in mmap() in accel_utils.c");
printf("Falling back to a non-mmaped approach\n");
obs->fftfile = chkfopen(cmd->argv[0], "rb");
obs->mmap_file = 0;
}
} else {
obs->mmap_file = 0;
}
}
/* Determine the other parameters */
if (cmd->zmax % ACCEL_DZ)
cmd->zmax = (cmd->zmax / ACCEL_DZ + 1) * ACCEL_DZ;
if (!obs->dat_input)
obs->workfile = chkfopen(obs->workfilenm, "w");
obs->N = (long long) idata->N;
if (cmd->photonP) {
if (obs->mmap_file || obs->dat_input) {
obs->nph = obs->fft[0].r;
} else {
obs->nph = get_numphotons(obs->fftfile);
}
printf("Normalizing powers using %.0f photons.\n\n", obs->nph);
} else {
obs->nph = 0.0;
/* For short FFTs insure that we don't pick up the DC */
/* or Nyquist component as part of the interpolation */
/* for higher frequencies. */
if (cmd->locpowP) {
obs->norm_type = 1;
printf("Normalizing powers using local-power determination.\n\n");
} else if (cmd->medianP) {
obs->norm_type = 0;
printf("Normalizing powers using median-blocks.\n\n");
} else {
obs->norm_type = 0;
printf("Normalizing powers using median-blocks (default).\n\n");
}
if (obs->dat_input) {
obs->fft[0].r = 1.0;
obs->fft[0].i = 1.0;
}
}
obs->lobin = cmd->lobin;
if (obs->lobin > 0) {
obs->nph = 0.0;
if (cmd->lobin > obs->numbins - 1) {
printf("\n'lobin' is greater than the total number of\n");
printf(" frequencies in the data set. Exiting.\n\n");
exit(1);
}
}
if (cmd->numharm != 1 &&
cmd->numharm != 2 &&
cmd->numharm != 4 && cmd->numharm != 8 && cmd->numharm != 16) {
printf("\n'numharm' = %d must be a power-of-two! Exiting\n\n", cmd->numharm);
exit(1);
}
obs->numharmstages = twon_to_index(cmd->numharm) + 1;
obs->dz = ACCEL_DZ;
obs->numz = cmd->zmax * 2 + 1;
obs->numbetween = ACCEL_NUMBETWEEN;
obs->dt = idata->dt;
obs->T = idata->dt * idata->N;
if (cmd->floP) {
obs->rlo = floor(cmd->flo * obs->T);
if (obs->rlo < obs->lobin)
obs->rlo = obs->lobin;
if (obs->rlo > obs->numbins - 1) {
printf("\nLow frequency to search 'flo' is greater than\n");
printf(" the highest available frequency. Exiting.\n\n");
exit(1);
}
} else {
if (cmd->rloP)
obs->rlo = cmd->rlo;
else
obs->rlo = 1.0;
if (obs->rlo < obs->lobin)
obs->rlo = obs->lobin;
if (obs->rlo > obs->numbins - 1) {
printf("\nLow frequency to search 'rlo' is greater than\n");
printf(" the available number of points. Exiting.\n\n");
exit(1);
}
}
obs->highestbin = obs->numbins - 1;
if (cmd->fhiP) {
obs->highestbin = ceil(cmd->fhi * obs->T);
if (obs->highestbin > obs->numbins - 1)
obs->highestbin = obs->numbins - 1;
obs->rhi = obs->highestbin;
if (obs->highestbin < obs->rlo) {
printf("\nHigh frequency to search 'fhi' is less than\n");
printf(" the lowest frequency to search 'flo'. Exiting.\n\n");
exit(1);
}
} else if (cmd->rhiP) {
obs->highestbin = cmd->rhi;
if (obs->highestbin > obs->numbins - 1)
obs->highestbin = obs->numbins - 1;
obs->rhi = obs->highestbin;
if (obs->highestbin < obs->rlo) {
printf("\nHigh frequency to search 'rhi' is less than\n");
printf(" the lowest frequency to search 'rlo'. Exiting.\n\n");
exit(1);
}
}
obs->dr = ACCEL_DR;
obs->zhi = cmd->zmax;
obs->zlo = -cmd->zmax;
obs->sigma = cmd->sigma;
obs->powcut = (float *) malloc(obs->numharmstages * sizeof(float));
obs->numindep = (long long *) malloc(obs->numharmstages * sizeof(long long));
for (ii = 0; ii < obs->numharmstages; ii++) {
if (obs->numz == 1)
obs->numindep[ii] = (obs->rhi - obs->rlo) / index_to_twon(ii);
else
/* The numz+1 takes care of the small amount of */
/* search we get above zmax and below zmin. */
obs->numindep[ii] = (obs->rhi - obs->rlo) * (obs->numz + 1) *
(obs->dz / 6.95) / index_to_twon(ii);
obs->powcut[ii] = power_for_sigma(obs->sigma,
index_to_twon(ii), obs->numindep[ii]);
}
obs->numzap = 0;
/*
if (zapfile!=NULL)
obs->numzap = get_birdies(cmd->zapfile, obs->T, obs->baryv,
&(obs->lobins), &(obs->hibins));
else
obs->numzap = 0;
*/
}
void search_minifft(fcomplex * minifft, int numminifft,
double min_orb_p, double max_orb_p,
rawbincand * cands, int numcands, int numharmsum,
int numbetween, double numfullfft, double timefullfft,
double lorfullfft, presto_interptype interptype,
presto_checkaliased checkaliased)
/* This routine searches a short FFT (usually produced using the */
/* MiniFFT binary search method) and returns a candidte vector */
/* containing information about the best binary candidates found. */
/* The routine uses either interbinning or interpolation as well */
/* as harmonic summing during the search. */
/* Arguments: */
/* 'minifft' is the FFT to search (complex valued) */
/* 'numminifft' is the number of complex points in 'minifft' */
/* 'min_orb_p' is the minimum orbital period (s) to search */
/* 'max_orb_p' is the maximum orbital period (s) to search */
/* 'cands' is a pre-allocated vector of rawbincand type in which */
/* the sorted (in decreasing sigma) candidates are returned */
/* 'numcands' is the length of the 'cands' vector */
/* 'numharmsum' the number of harmonics to sum during the search */
/* 'numbetween' the points to interpolate per bin */
/* 'numfullfft' the number of points in the original long FFT */
/* 'timefullfft' the duration of the original time series (s) */
/* 'lorfullfft' the 1st bin of the long FFT that was miniFFT'd */
/* 'interptype' is either INTERBIN or INTERPOLATE. */
/* INTERBIN = (interbinning) is fast but less sensitive. */
/* NOTE: INTERBINNING is conducted by this routine! */
/* INTERPOLATE = (Fourier interpolation) is slower but more */
/* sensitive. */
/* NOTE: The interpolation is assumed to ALREADY have been */
/* completed by the calling function! The easiest */
/* way is by zero-padding to 2*numminifft and FFTing. */
/* If you use this method, make sure numminifft is the*/
/* original length rather than the interpolated */
/* length and also make sure numbetween is correct. */
/* 'checkaliased' is either CHECK_ALIASED or NO_CHECK_ALIASED. */
/* NO_CHECK_ALIASED = harmonic summing does not include */
/* aliased freqs making it faster but less sensitive. */
/* CHECK_ALIASED = harmonic summing includes aliased freqs */
/* making it slower but more sensitive. */
{
int ii, jj, fftlen, offset, numtosearch = 0, lobin, hibin, numspread = 0;
float powargr, powargi, *fullpows = NULL, *sumpows;
double twobypi, minpow, minsig, dr, numindep;
fcomplex *spread;
/* Override the value of numbetween if interbinning */
if (interptype == INTERBIN)
numbetween = 2;
/* Prep some other values we will need */
dr = 1.0 / (double) numbetween;
twobypi = 2.0 / PI;
fftlen = numminifft * numbetween;
for (ii = 0; ii < numcands; ii++) {
cands[ii].mini_sigma = 0.0;
cands[ii].mini_power = 0.0;
}
lobin = ceil(2 * numminifft * min_orb_p / timefullfft);
if (lobin <= 0)
lobin = 1;
hibin = floor(2 * numminifft * max_orb_p / timefullfft);
if (hibin >= 2 * numminifft)
hibin = 2 * numminifft - 1;
lobin *= numbetween;
hibin *= numbetween;
/* Spread and interpolate the fft */
numtosearch = (checkaliased == CHECK_ALIASED) ? 2 * fftlen : fftlen;
numspread = numminifft * numbetween + 1;
if (interptype == INTERPOLATE) { /* INTERPOLATE */
spread = minifft;
} else { /* INTERBIN */
spread = gen_cvect(numspread);
spread_with_pad(minifft, numminifft, spread, numspread, numbetween, 0);
for (ii = 1; ii < fftlen; ii += 2) {
spread[ii].r = twobypi * (spread[ii - 1].r - spread[ii + 1].r);
spread[ii].i = twobypi * (spread[ii - 1].i - spread[ii + 1].i);
}
}
spread[0].r = spread[fftlen].r = 1.0;
spread[0].i = spread[fftlen].i = 0.0;
fullpows = gen_fvect(numtosearch);
fullpows[0] = 1.0;
if (checkaliased == CHECK_ALIASED)
fullpows[fftlen] = 1.0; /* used to be nyquist^2 */
/* The following wraps the data around the Nyquist freq such that */
/* we consider aliased frequencies as well (If CHECK_ALIASED). */
if (checkaliased == CHECK_ALIASED)
for (ii = 1, jj = numtosearch - 1; ii < fftlen; ii++, jj--)
fullpows[ii] = fullpows[jj] = POWER(spread[ii].r, spread[ii].i);
else
for (ii = 1; ii < numtosearch; ii++)
fullpows[ii] = POWER(spread[ii].r, spread[ii].i);
if (interptype == INTERBIN)
vect_free(spread);
/* Search the raw powers */
numindep = hibin - lobin + 1.0;
minpow = power_for_sigma(MINRETURNSIG, 1, numindep);
for (ii = lobin; ii < hibin; ii++) {
if (fullpows[ii] > minpow) {
cands[numcands - 1].mini_r = dr * (double) ii;
cands[numcands - 1].mini_power = fullpows[ii];
cands[numcands - 1].mini_numsum = 1.0;
cands[numcands - 1].mini_sigma = candidate_sigma(fullpows[ii], 1, numindep);
minsig = percolate_rawbincands(cands, numcands);
if (cands[numcands - 1].mini_power > minpow)
minpow = cands[numcands - 1].mini_power;
}
}
/* If needed, sum and search the harmonics */
if (numharmsum > 1) {
sumpows = gen_fvect(numtosearch);
memcpy(sumpows, fullpows, sizeof(float) * numtosearch);
for (ii = 2; ii <= numharmsum; ii++) {
offset = ii / 2;
numindep = (hibin - lobin + 1.0) / (double) ii;
if (cands[numcands - 1].mini_sigma < MINRETURNSIG)
minsig = MINRETURNSIG;
else
minsig = cands[numcands - 1].mini_sigma;
minpow = power_for_sigma(minsig, ii, numindep);
for (jj = lobin * ii; jj < hibin; jj++) {
sumpows[jj] += fullpows[(jj + offset) / ii];
if (sumpows[jj] > minpow) {
cands[numcands - 1].mini_r = (dr * (double) jj) / ii;
cands[numcands - 1].mini_power = sumpows[jj];
cands[numcands - 1].mini_numsum = (double) ii;
cands[numcands - 1].mini_sigma =
candidate_sigma(sumpows[jj], ii, numindep);
minsig = percolate_rawbincands(cands, numcands);
if (minsig > MINRETURNSIG)
minpow = power_for_sigma(minsig, ii, numindep);
}
}
}
vect_free(sumpows);
}
vect_free(fullpows);
/* Add the rest of the rawbincand data to the candidate array */
for (ii = 0; ii < numcands; ii++) {
cands[ii].full_N = numfullfft;
cands[ii].full_T = timefullfft;
cands[ii].full_lo_r = lorfullfft;
cands[ii].mini_N = 2 * numminifft; /* # of real points */
cands[ii].psr_p = timefullfft / (lorfullfft + numminifft);
cands[ii].orb_p = timefullfft * cands[ii].mini_r / cands[ii].mini_N;
}
}
void rfifind_plot(int numchan, int numint, int ptsperint,
float timesigma, float freqsigma,
float inttrigfrac, float chantrigfrac,
float **dataavg, float **datastd, float **datapow,
int *userchan, int numuserchan,
int *userints, int numuserints,
infodata * idata, unsigned char **bytemask,
mask * oldmask, mask * newmask,
rfi * rfivect, int numrfi, int rfixwin, int rfips, int xwin)
/* Make the beautiful multi-page rfifind plots */
{
int ii, jj, ct, loops = 1;
float *freqs, *chans, *times, *ints;
float *avg_chan_avg, *std_chan_avg, *pow_chan_avg;
float *avg_chan_med, *std_chan_med, *pow_chan_med;
float *avg_chan_std, *std_chan_std, *pow_chan_std;
float *avg_int_avg, *std_int_avg, *pow_int_avg;
float *avg_int_med, *std_int_med, *pow_int_med;
float *avg_int_std, *std_int_std, *pow_int_std;
float dataavg_avg, datastd_avg, datapow_avg;
float dataavg_med, datastd_med, datapow_med;
float dataavg_std, datastd_std, datapow_std;
float avg_reject, std_reject, pow_reject;
double inttim, T, lof, hif;
inttim = ptsperint * idata->dt;
T = inttim * numint;
lof = idata->freq - 0.5 * idata->chan_wid;
hif = lof + idata->freqband;
avg_chan_avg = gen_fvect(numchan);
std_chan_avg = gen_fvect(numchan);
pow_chan_avg = gen_fvect(numchan);
avg_int_avg = gen_fvect(numint);
std_int_avg = gen_fvect(numint);
pow_int_avg = gen_fvect(numint);
avg_chan_med = gen_fvect(numchan);
std_chan_med = gen_fvect(numchan);
pow_chan_med = gen_fvect(numchan);
avg_int_med = gen_fvect(numint);
std_int_med = gen_fvect(numint);
pow_int_med = gen_fvect(numint);
avg_chan_std = gen_fvect(numchan);
std_chan_std = gen_fvect(numchan);
pow_chan_std = gen_fvect(numchan);
avg_int_std = gen_fvect(numint);
std_int_std = gen_fvect(numint);
pow_int_std = gen_fvect(numint);
chans = gen_fvect(numchan);
freqs = gen_fvect(numchan);
for (ii = 0; ii < numchan; ii++) {
chans[ii] = ii;
freqs[ii] = idata->freq + ii * idata->chan_wid;
}
ints = gen_fvect(numint);
times = gen_fvect(numint);
for (ii = 0; ii < numint; ii++) {
ints[ii] = ii;
times[ii] = 0.0 + ii * inttim;
}
/* Calculate the statistics of the full set */
ct = numchan * numint;
calc_avgmedstd(dataavg[0], ct, 0.8, 1, &dataavg_avg, &dataavg_med, &dataavg_std);
calc_avgmedstd(datastd[0], ct, 0.8, 1, &datastd_avg, &datastd_med, &datastd_std);
calc_avgmedstd(datapow[0], ct, 0.5, 1, &datapow_avg, &datapow_med, &datapow_std);
avg_reject = timesigma * dataavg_std;
std_reject = timesigma * datastd_std;
pow_reject = power_for_sigma(freqsigma, 1, ptsperint / 2);
/* Calculate the channel/integration statistics vectors */
for (ii = 0; ii < numint; ii++) {
calc_avgmedstd(dataavg[0] + ii * numchan, numchan, 0.8, 1,
avg_int_avg + ii, avg_int_med + ii, avg_int_std + ii);
calc_avgmedstd(datastd[0] + ii * numchan, numchan, 0.8, 1,
std_int_avg + ii, std_int_med + ii, std_int_std + ii);
calc_avgmedstd(datapow[0] + ii * numchan, numchan, 0.5, 1,
pow_int_avg + ii, pow_int_med + ii, pow_int_std + ii);
}
for (ii = 0; ii < numchan; ii++) {
calc_avgmedstd(dataavg[0] + ii, numint, 0.8, numchan,
avg_chan_avg + ii, avg_chan_med + ii, avg_chan_std + ii);
calc_avgmedstd(datastd[0] + ii, numint, 0.8, numchan,
std_chan_avg + ii, std_chan_med + ii, std_chan_std + ii);
calc_avgmedstd(datapow[0] + ii, numint, 0.5, numchan,
pow_chan_avg + ii, pow_chan_med + ii, pow_chan_std + ii);
/*
fprintf(stderr, "%12.7g %12.7g %12.7g %12.7g %12.7g %12.7g %12.7g %12.7g %12.7g \n",
avg_chan_avg[ii], avg_chan_med[ii], avg_chan_std[ii],
std_chan_avg[ii], std_chan_med[ii], std_chan_std[ii],
pow_chan_avg[ii], pow_chan_med[ii], pow_chan_std[ii]);
*/
}
/* Generate the byte mask */
/* Set the channels/intervals picked by the user */
if (numuserints)
for (ii = 0; ii < numuserints; ii++)
if (userints[ii] >= 0 && userints[ii] < numint)
for (jj = 0; jj < numchan; jj++)
bytemask[userints[ii]][jj] |= USERINTS;
if (numuserchan)
for (ii = 0; ii < numuserchan; ii++)
if (userchan[ii] >= 0 && userchan[ii] < numchan)
for (jj = 0; jj < numint; jj++)
bytemask[jj][userchan[ii]] |= USERCHAN;
/* Compare each point in an interval (or channel) with */
/* the interval's (or channel's) median and the overall */
/* standard deviation. If the channel/integration */
/* medians are more than sigma different than the global */
/* value, set them to the global. */
{
float int_med, chan_med;
for (ii = 0; ii < numint; ii++) {
for (jj = 0; jj < numchan; jj++) {
{ /* Powers */
if (datapow[ii][jj] > pow_reject)
if (!(bytemask[ii][jj] & PADDING))
bytemask[ii][jj] |= BAD_POW;
}
{ /* Averages */
if (fabs(avg_int_med[ii] - dataavg_med) > timesigma * dataavg_std)
int_med = dataavg_med;
else
int_med = avg_int_med[ii];
if (fabs(avg_chan_med[jj] - dataavg_med) > timesigma * dataavg_std)
chan_med = dataavg_med;
else
chan_med = avg_chan_med[jj];
if (fabs(dataavg[ii][jj] - int_med) > avg_reject ||
fabs(dataavg[ii][jj] - chan_med) > avg_reject)
if (!(bytemask[ii][jj] & PADDING))
bytemask[ii][jj] |= BAD_AVG;
}
{ /* Standard Deviations */
if (fabs(std_int_med[ii] - datastd_med) > timesigma * datastd_std)
int_med = datastd_med;
else
int_med = std_int_med[ii];
if (fabs(std_chan_med[jj] - datastd_med) > timesigma * datastd_std)
chan_med = datastd_med;
else
chan_med = std_chan_med[jj];
if (fabs(datastd[ii][jj] - int_med) > std_reject ||
fabs(datastd[ii][jj] - chan_med) > std_reject)
if (!(bytemask[ii][jj] & PADDING))
bytemask[ii][jj] |= BAD_STD;
}
}
}
}
/* Step over the intervals and channels and count how many are set "bad". */
/* For a given interval, if the number of bad channels is greater than */
/* chantrigfrac*numchan then reject the whole interval. */
/* For a given channel, if the number of bad intervals is greater than */
/* inttrigfrac*numint then reject the whole channel. */
{
int badnum, trignum;
/* Loop over the intervals */
trignum = (int) (numchan * chantrigfrac);
for (ii = 0; ii < numint; ii++) {
if (!(bytemask[ii][0] & USERINTS)) {
badnum = 0;
for (jj = 0; jj < numchan; jj++)
if (bytemask[ii][jj] & BADDATA)
badnum++;
if (badnum > trignum) {
userints[numuserints++] = ii;
for (jj = 0; jj < numchan; jj++)
bytemask[ii][jj] |= USERINTS;
}
}
}
/* Loop over the channels */
trignum = (int) (numint * inttrigfrac);
for (ii = 0; ii < numchan; ii++) {
if (!(bytemask[0][ii] & USERCHAN)) {
badnum = 0;
for (jj = 0; jj < numint; jj++)
if (bytemask[jj][ii] & BADDATA)
badnum++;
if (badnum > trignum) {
userchan[numuserchan++] = ii;
for (jj = 0; jj < numint; jj++)
bytemask[jj][ii] |= USERCHAN;
}
}
}
}
/* Generate the New Mask */
fill_mask(timesigma, freqsigma, idata->mjd_i + idata->mjd_f,
ptsperint * idata->dt, idata->freq, idata->chan_wid,
numchan, numint, ptsperint, numuserchan, userchan,
numuserints, userints, bytemask, newmask);
/* Place the oldmask over the newmask for plotting purposes */
if (oldmask->numchan)
set_oldmask_bits(oldmask, bytemask);
/*
* Now plot the results
*/
if (xwin)
loops = 2;
for (ct = 0; ct < loops; ct++) { /* PS/XWIN Plot Loop */
float min, max, tr[6], locut, hicut;
float left, right, top, bottom;
float xl, xh, yl, yh;
float tt, ft, th, fh; /* thin and fat thicknesses and heights */
float lm, rm, tm, bm; /* LRTB margins */
float xarr[2], yarr[2];
char outdev[100];
int ii, mincol, maxcol, numcol;
/*Set the PGPLOT device to an X-Window */
if (ct == 1)
strcpy(outdev, "/XWIN");
else
sprintf(outdev, "%s.ps/CPS", idata->name);
/* Open and prep our device */
cpgopen(outdev);
cpgpap(10.25, 8.5 / 11.0);
cpgpage();
cpgiden();
cpgsch(0.7);
cpgqcir(&mincol, &maxcol);
numcol = maxcol - mincol + 1;
for (ii = mincol; ii <= maxcol; ii++) {
float color;
color = (float) (maxcol - ii) / (float) numcol;
cpgscr(ii, color, color, color);
}
/* Set thicknesses and margins */
lm = 0.04;
rm = 0.04;
bm = 0.08;
tm = 0.05;
ft = 3.0; /* This sets fat thickness = 3 x thin thickness */
tt = 0.92 / (6.0 + 4.0 * ft);
ft *= tt;
fh = 0.55;
th = tt * 11.0 / 8.5;
{ /* Powers Histogram */
float *theo, *hist, *hpows, *tpows, maxhist = 0.0, maxtheo = 0.0;
int numhist = 40, numtheo = 200, bin, numpows;
double dtheo, dhist, spacing;
/* Calculate the predicted distribution of max powers */
numpows = numint * numchan;
find_min_max_arr(numpows, datapow[0], &min, &max);
min = (min < 5.0) ? log10(5.0 * 0.95) : log10(min * 0.95);
max = log10(max * 1.05);
dhist = (max - min) / numhist;
theo = gen_fvect(numtheo);
tpows = gen_fvect(numtheo);
hist = gen_fvect(numhist);
hpows = gen_fvect(numhist);
for (ii = 0; ii < numhist; ii++) {
hist[ii] = 0.0;
hpows[ii] = min + ii * dhist;
}
for (ii = 0; ii < numpows; ii++) {
bin = (*(datapow[0] + ii) == 0.0) ? 0 :
(log10(*(datapow[0] + ii)) - min) / dhist;
if (bin < 0)
bin = 0;
if (bin >= numhist)
bin = numhist;
hist[bin] += 1.0;
}
for (ii = 0; ii < numhist; ii++)
if (hist[ii] > maxhist)
maxhist = hist[ii];
maxhist *= 1.1;
dtheo = (max - min) / (double) (numtheo - 1);
for (ii = 0; ii < numtheo; ii++) {
tpows[ii] = min + ii * dtheo;
theo[ii] = single_power_pdf(pow(10.0, tpows[ii]),
ptsperint / 2) * numpows;
spacing = (pow(10.0, tpows[ii] + dhist) - pow(10.0, tpows[ii]));
theo[ii] *= spacing;
if (theo[ii] > maxtheo)
maxtheo = theo[ii];
}
maxtheo *= 1.1;
if (maxtheo > maxhist)
maxhist = maxtheo;
left = lm;
right = lm + ft + tt;
bottom = 0.80;
top = 0.96;
cpgsvp(left, right, bottom, top);
xl = min;
xh = max;
yl = 0.0;
yh = maxhist;
cpgswin(xl, xh, yl, yh);
cpgmtxt("L", 1.1, 0.5, 0.5, "Number");
cpgmtxt("B", 2.1, 0.5, 0.5, "Max Power");
cpgbin(numhist, hpows, hist, 0);
cpgscr(maxcol, 0.5, 0.5, 0.5);
cpgsci(maxcol); /* Grey */
cpgline(numtheo, tpows, theo);
xarr[0] = log10(power_for_sigma(freqsigma, 1, ptsperint / 2));
xarr[1] = xarr[0];
yarr[0] = yl;
yarr[1] = yh;
cpgsls(4); /* Dotted line */
cpgscr(maxcol, 1.0, 0.0, 0.0);
cpgsci(maxcol); /* Red */
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgbox("BCLNST", 0.0, 0, "BC", 0.0, 0);
vect_free(hist);
vect_free(theo);
vect_free(tpows);
vect_free(hpows);
}
/* Maximum Powers */
left = lm;
right = lm + ft;
bottom = bm;
top = bm + fh;
xl = 0.0;
xh = numchan;
yl = 0.0;
yh = T;
cpgsvp(left, right, bottom, top);
cpgswin(xl, xh, yl, yh);
cpgscr(maxcol, 1.0, 0.0, 0.0); /* Red */
locut = 0.0;
hicut = pow_reject;
tr[2] = tr[4] = 0.0;
tr[1] = (xh - xl) / numchan;
tr[0] = xl - (tr[1] / 2);
tr[5] = (yh - yl) / numint;
tr[3] = yl - (tr[5] / 2);
cpgimag(datapow[0], numchan, numint, 1, numchan, 1, numint, locut, hicut, tr);
cpgswin(xl, xh, yl, yh);
cpgbox("BNST", 0.0, 0, "BNST", 0.0, 0);
cpgmtxt("B", 2.6, 0.5, 0.5, "Channel");
cpgmtxt("L", 2.1, 0.5, 0.5, "Time (s)");
xl = lof;
xh = hif;
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("CST", 0.0, 0, "CST", 0.0, 0);
/* Max Power Label */
left = lm + ft;
right = lm + ft + tt;
bottom = bm + fh;
top = bm + fh + th;
cpgsvp(left, right, bottom, top);
cpgswin(0.0, 1.0, 0.0, 1.0);
cpgscr(maxcol, 1.0, 0.0, 0.0);
cpgsci(maxcol); /* Red */
cpgptxt(0.5, 0.7, 0.0, 0.5, "Max");
cpgptxt(0.5, 0.3, 0.0, 0.5, "Power");
cpgsci(1); /* Default color */
/* Max Power versus Time */
left = lm + ft;
right = lm + ft + tt;
bottom = bm;
top = bm + fh;
cpgsvp(left, right, bottom, top);
find_min_max_arr(numint, pow_int_med, &min, &max);
xl = 0.0;
xh = 1.5 * pow_reject;
yl = 0.0;
yh = T;
cpgswin(xl, xh, yl, yh);
cpgbox("BCST", 0.0, 0, "BST", 0.0, 0);
cpgscr(maxcol, 1.0, 0.0, 0.0);
cpgsci(maxcol); /* Red */
yarr[0] = yl;
yarr[1] = yh;
xarr[0] = xarr[1] = datapow_med;
cpgline(2, xarr, yarr);
cpgsls(4); /* Dotted line */
xarr[0] = xarr[1] = pow_reject;
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgline(numint, pow_int_med, times);
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("", 0.0, 0, "CMST", 0.0, 0);
/* cpgmtxt("R", 2.3, 0.5, 0.5, "Interval Number"); */
/* Max Power versus Channel */
left = lm;
right = lm + ft;
bottom = bm + fh;
top = bm + fh + th;
cpgsvp(left, right, bottom, top);
find_min_max_arr(numchan, pow_chan_med, &min, &max);
xl = 0.0;
xh = numchan;
yl = 0.0;
yh = 1.5 * pow_reject;
cpgswin(xl, xh, yl, yh);
cpgbox("BST", 0.0, 0, "BCST", 0.0, 0);
cpgscr(maxcol, 1.0, 0.0, 0.0);
cpgsci(maxcol); /* Red */
xarr[0] = xl;
xarr[1] = xh;
yarr[0] = yarr[1] = datapow_med;
cpgline(2, xarr, yarr);
cpgsls(4); /* Dotted line */
yarr[0] = yarr[1] = pow_reject;
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgline(numchan, chans, pow_chan_med);
xl = lof;
xh = hif;
cpgswin(xl, xh, yl, yh);
cpgbox("CMST", 0.0, 0, "", 0.0, 0);
cpgmtxt("T", 1.8, 0.5, 0.5, "Frequency (MHz)");
/* Standard Deviations */
left = lm + ft + 2.0 * tt;
right = lm + 2.0 * ft + 2.0 * tt;
bottom = bm;
top = bm + fh;
xl = 0.0;
xh = numchan;
yl = 0.0;
yh = T;
cpgsvp(left, right, bottom, top);
cpgswin(xl, xh, yl, yh);
cpgscr(mincol, 0.7, 1.0, 0.7); /* Light Green */
cpgscr(maxcol, 0.3, 1.0, 0.3); /* Dark Green */
locut = datastd_med - timesigma * datastd_std;
hicut = datastd_med + timesigma * datastd_std;
tr[2] = tr[4] = 0.0;
tr[1] = (xh - xl) / numchan;
tr[0] = xl - (tr[1] / 2);
tr[5] = (yh - yl) / numint;
tr[3] = yl - (tr[5] / 2);
cpgimag(datastd[0], numchan, numint, 1, numchan, 1, numint, locut, hicut, tr);
cpgswin(xl, xh, yl, yh);
cpgbox("BNST", 0.0, 0, "BNST", 0.0, 0);
cpgmtxt("B", 2.6, 0.5, 0.5, "Channel");
xl = lof;
xh = hif;
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("CST", 0.0, 0, "CST", 0.0, 0);
/* Data Sigma Label */
left = lm + 2.0 * ft + 2.0 * tt;
right = lm + 2.0 * ft + 3.0 * tt;
bottom = bm + fh;
top = bm + fh + th;
cpgsvp(left, right, bottom, top);
cpgswin(0.0, 1.0, 0.0, 1.0);
cpgscr(maxcol, 0.0, 1.0, 0.0);
cpgsci(maxcol); /* Green */
cpgptxt(0.5, 0.7, 0.0, 0.5, "Data");
cpgptxt(0.5, 0.3, 0.0, 0.5, "Sigma");
cpgsci(1); /* Default color */
/* Data Sigma versus Time */
left = lm + 2.0 * ft + 2.0 * tt;
right = lm + 2.0 * ft + 3.0 * tt;
bottom = bm;
top = bm + fh;
cpgsvp(left, right, bottom, top);
xl = datastd_med - 2.0 * std_reject;
xh = datastd_med + 2.0 * std_reject;
yl = 0.0;
yh = T;
cpgswin(xl, xh, yl, yh);
cpgbox("BCST", 0.0, 0, "BST", 0.0, 0);
cpgscr(maxcol, 0.0, 1.0, 0.0);
cpgsci(maxcol); /* Green */
yarr[0] = yl;
yarr[1] = yh;
xarr[0] = xarr[1] = datastd_med;
cpgline(2, xarr, yarr);
cpgsls(4); /* Dotted line */
xarr[0] = xarr[1] = datastd_med + std_reject;
cpgline(2, xarr, yarr);
xarr[0] = xarr[1] = datastd_med - std_reject;
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgline(numint, std_int_med, times);
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("", 0.0, 0, "CMST", 0.0, 0);
/* cpgmtxt("R", 2.3, 0.5, 0.5, "Interval Number"); */
/* Data Sigma versus Channel */
left = lm + ft + 2.0 * tt;
right = lm + 2.0 * ft + 2.0 * tt;
bottom = bm + fh;
top = bm + fh + th;
cpgsvp(left, right, bottom, top);
xl = 0.0;
xh = numchan;
yl = datastd_med - 2.0 * std_reject;
yh = datastd_med + 2.0 * std_reject;
cpgswin(xl, xh, yl, yh);
cpgbox("BST", 0.0, 0, "BCST", 0.0, 0);
cpgscr(maxcol, 0.0, 1.0, 0.0);
cpgsci(maxcol); /* Green */
xarr[0] = xl;
xarr[1] = xh;
yarr[0] = yarr[1] = datastd_med;
cpgline(2, xarr, yarr);
cpgsls(4); /* Dotted line */
yarr[0] = yarr[1] = datastd_med + std_reject;
cpgline(2, xarr, yarr);
yarr[0] = yarr[1] = datastd_med - std_reject;
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgline(numchan, chans, std_chan_med);
xl = lof;
xh = hif;
cpgswin(xl, xh, yl, yh);
cpgbox("CMST", 0.0, 0, "", 0.0, 0);
cpgmtxt("T", 1.8, 0.5, 0.5, "Frequency (MHz)");
/* Data Mean */
left = lm + 2.0 * ft + 4.0 * tt;
right = lm + 3.0 * ft + 4.0 * tt;
bottom = bm;
top = bm + fh;
xl = 0.0;
xh = numchan;
yl = 0.0;
yh = T;
cpgsvp(left, right, bottom, top);
cpgswin(xl, xh, yl, yh);
cpgscr(mincol, 0.7, 0.7, 1.0); /* Light Blue */
cpgscr(maxcol, 0.3, 0.3, 1.0); /* Dark Blue */
locut = dataavg_med - timesigma * dataavg_std;
hicut = dataavg_med + timesigma * dataavg_std;
tr[2] = tr[4] = 0.0;
tr[1] = (xh - xl) / numchan;
tr[0] = xl - (tr[1] / 2);
tr[5] = (yh - yl) / numint;
tr[3] = yl - (tr[5] / 2);
cpgimag(dataavg[0], numchan, numint, 1, numchan, 1, numint, locut, hicut, tr);
cpgswin(xl, xh, yl, yh);
cpgbox("BNST", 0.0, 0, "BNST", 0.0, 0);
cpgmtxt("B", 2.6, 0.5, 0.5, "Channel");
xl = lof;
xh = hif;
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("CST", 0.0, 0, "CST", 0.0, 0);
/* Data Mean Label */
left = lm + 3.0 * ft + 4.0 * tt;
right = lm + 3.0 * ft + 5.0 * tt;
bottom = bm + fh;
top = bm + fh + th;
cpgsvp(left, right, bottom, top);
cpgswin(0.0, 1.0, 0.0, 1.0);
cpgscr(maxcol, 0.0, 0.0, 1.0);
cpgsci(maxcol); /* Blue */
cpgptxt(0.5, 0.7, 0.0, 0.5, "Data");
cpgptxt(0.5, 0.3, 0.0, 0.5, "Mean");
cpgsci(1); /* Default color */
/* Data Mean versus Time */
left = lm + 3.0 * ft + 4.0 * tt;
right = lm + 3.0 * ft + 5.0 * tt;
bottom = bm;
top = bm + fh;
cpgsvp(left, right, bottom, top);
xl = dataavg_med - 2.0 * avg_reject;
xh = dataavg_med + 2.0 * avg_reject;
yl = 0.0;
yh = T;
cpgswin(xl, xh, yl, yh);
cpgbox("BCST", 0.0, 0, "BST", 0.0, 0);
cpgscr(maxcol, 0.0, 0.0, 1.0);
cpgsci(maxcol); /* Blue */
yarr[0] = yl;
yarr[1] = yh;
xarr[0] = xarr[1] = dataavg_med;
cpgline(2, xarr, yarr);
cpgsls(4); /* Dotted line */
xarr[0] = xarr[1] = dataavg_med + avg_reject;
cpgline(2, xarr, yarr);
xarr[0] = xarr[1] = dataavg_med - avg_reject;
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgline(numint, avg_int_med, times);
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("", 0.0, 0, "CMST", 0.0, 0);
/* Data Mean versus Channel */
left = lm + 2.0 * ft + 4.0 * tt;
right = lm + 3.0 * ft + 4.0 * tt;
bottom = bm + fh;
top = bm + fh + th;
cpgsvp(left, right, bottom, top);
xl = 0.0;
xh = numchan;
yl = dataavg_med - 2.0 * avg_reject;
yh = dataavg_med + 2.0 * avg_reject;
cpgswin(xl, xh, yl, yh);
cpgbox("BST", 0.0, 0, "BCST", 0.0, 0);
cpgscr(maxcol, 0.0, 0.0, 1.0);
cpgsci(maxcol); /* Blue */
xarr[0] = xl;
xarr[1] = xh;
yarr[0] = yarr[1] = dataavg_med;
cpgline(2, xarr, yarr);
cpgsls(4); /* Dotted line */
yarr[0] = yarr[1] = dataavg_med + avg_reject;
cpgline(2, xarr, yarr);
yarr[0] = yarr[1] = dataavg_med - avg_reject;
cpgline(2, xarr, yarr);
cpgsls(1); /* Solid line */
cpgsci(1); /* Default color */
cpgline(numchan, chans, avg_chan_med);
xl = lof;
xh = hif;
cpgswin(xl, xh, yl, yh);
cpgbox("CMST", 0.0, 0, "", 0.0, 0);
cpgmtxt("T", 1.8, 0.5, 0.5, "Frequency (MHz)");
{ /* Add the Data Info area */
char out[200], out2[100];
float dy = 0.025;
cpgsvp(0.0, 1.0, 0.0, 1.0);
cpgswin(0.0, 1.0, 0.0, 1.0);
left = lm + ft + 1.5 * tt;
top = 1.0 - tm;
cpgsch(1.0);
sprintf(out, "%-s", idata->name);
cpgptxt(0.5, 1.0 - 0.5 * tm, 0.0, 0.5, out);
cpgsch(0.8);
sprintf(out, "Object:");
cpgtext(left + 0.0, top - 0 * dy, out);
sprintf(out, "%-s", idata->object);
cpgtext(left + 0.1, top - 0 * dy, out);
sprintf(out, "Telescope:");
cpgtext(left + 0.0, top - 1 * dy, out);
sprintf(out, "%-s", idata->telescope);
cpgtext(left + 0.1, top - 1 * dy, out);
sprintf(out, "Instrument:");
cpgtext(left + 0.0, top - 2 * dy, out);
sprintf(out, "%-s", idata->instrument);
cpgtext(left + 0.1, top - 2 * dy, out);
ra_dec_to_string(out2, idata->ra_h, idata->ra_m, idata->ra_s);
sprintf(out, "RA\\dJ2000\\u");
cpgtext(left + 0.0, top - 3 * dy, out);
sprintf(out, "= %-s", out2);
cpgtext(left + 0.08, top - 3 * dy, out);
ra_dec_to_string(out2, idata->dec_d, idata->dec_m, idata->dec_s);
sprintf(out, "DEC\\dJ2000\\u");
cpgtext(left + 0.0, top - 4 * dy, out);
sprintf(out, "= %-s", out2);
cpgtext(left + 0.08, top - 4 * dy, out);
sprintf(out, "Epoch\\dtopo\\u");
cpgtext(left + 0.0, top - 5 * dy, out);
sprintf(out, "= %-.11f", idata->mjd_i + idata->mjd_f);
cpgtext(left + 0.08, top - 5 * dy, out);
sprintf(out, "T\\dsample\\u (s)");
cpgtext(left + 0.0, top - 6 * dy, out);
sprintf(out, "= %g", idata->dt);
cpgtext(left + 0.08, top - 6 * dy, out);
sprintf(out, "T\\dtotal\\u (s)");
cpgtext(left + 0.0, top - 7 * dy, out);
sprintf(out, "= %g", T);
cpgtext(left + 0.08, top - 7 * dy, out);
left = lm + ft + 7.8 * tt;
sprintf(out, "Num channels");
cpgtext(left + 0.0, top - 0 * dy, out);
sprintf(out, "= %-d", numchan);
cpgtext(left + 0.12, top - 0 * dy, out);
sprintf(out, "Pts per int");
cpgtext(left + 0.19, top - 0 * dy, out);
sprintf(out, "= %-d", ptsperint);
cpgtext(left + 0.29, top - 0 * dy, out);
sprintf(out, "Num intervals");
cpgtext(left + 0.0, top - 1 * dy, out);
sprintf(out, "= %-d", numint);
cpgtext(left + 0.12, top - 1 * dy, out);
sprintf(out, "Time per int");
cpgtext(left + 0.19, top - 1 * dy, out);
sprintf(out, "= %-g", inttim);
cpgtext(left + 0.29, top - 1 * dy, out);
sprintf(out, "Power:");
cpgtext(left + 0.0, top - 2 * dy, out);
sprintf(out, "median");
cpgtext(left + 0.06, top - 2 * dy, out);
sprintf(out, "= %-.3f", datapow_med);
cpgtext(left + 0.12, top - 2 * dy, out);
sprintf(out, "\\gs");
cpgtext(left + 0.21, top - 2 * dy, out);
sprintf(out, "= %-.3g", datapow_std);
cpgtext(left + 0.245, top - 2 * dy, out);
find_min_max_arr(numint * numchan, datapow[0], &min, &max);
sprintf(out, "min");
cpgtext(left + 0.06, top - 3 * dy, out);
sprintf(out, "= %-.3f", min);
cpgtext(left + 0.12, top - 3 * dy, out);
sprintf(out, "max");
cpgtext(left + 0.21, top - 3 * dy, out);
sprintf(out, "= %-.3f", max);
cpgtext(left + 0.245, top - 3 * dy, out);
sprintf(out, "Sigma:");
cpgtext(left + 0.0, top - 4 * dy, out);
sprintf(out, "median");
cpgtext(left + 0.06, top - 4 * dy, out);
sprintf(out, "= %-.3f", datastd_med);
cpgtext(left + 0.12, top - 4 * dy, out);
sprintf(out, "\\gs");
cpgtext(left + 0.21, top - 4 * dy, out);
sprintf(out, "= %-.3g", datastd_std);
cpgtext(left + 0.245, top - 4 * dy, out);
find_min_max_arr(numint * numchan, datastd[0], &min, &max);
sprintf(out, "min");
cpgtext(left + 0.06, top - 5 * dy, out);
sprintf(out, "= %-.3f", min);
cpgtext(left + 0.12, top - 5 * dy, out);
sprintf(out, "max");
cpgtext(left + 0.21, top - 5 * dy, out);
sprintf(out, "= %-.3f", max);
cpgtext(left + 0.245, top - 5 * dy, out);
sprintf(out, "Mean:");
cpgtext(left + 0.0, top - 6 * dy, out);
sprintf(out, "median");
cpgtext(left + 0.06, top - 6 * dy, out);
sprintf(out, "= %-.3f", dataavg_med);
cpgtext(left + 0.12, top - 6 * dy, out);
sprintf(out, "\\gs");
cpgtext(left + 0.21, top - 6 * dy, out);
sprintf(out, "= %-.3g", dataavg_std);
cpgtext(left + 0.245, top - 6 * dy, out);
find_min_max_arr(numint * numchan, dataavg[0], &min, &max);
sprintf(out, "min");
cpgtext(left + 0.06, top - 7 * dy, out);
sprintf(out, "= %-.3f", min);
cpgtext(left + 0.12, top - 7 * dy, out);
sprintf(out, "max");
cpgtext(left + 0.21, top - 7 * dy, out);
sprintf(out, "= %-.3f", max);
cpgtext(left + 0.245, top - 7 * dy, out);
}
{ /* Plot the Mask */
unsigned char byte;
char temp[200];
float **plotmask, rr, gg, bb, page;
plotmask = gen_fmatrix(numint, numchan);
for (ii = 0; ii < numint; ii++) {
for (jj = 0; jj < numchan; jj++) {
byte = bytemask[ii][jj];
plotmask[ii][jj] = 0.0;
if (byte & PADDING)
plotmask[ii][jj] = 1.0;
if (byte & OLDMASK)
plotmask[ii][jj] = 2.0;
if (byte & USERZAP)
plotmask[ii][jj] = 3.0;
if (byte & BAD_POW)
plotmask[ii][jj] = 4.0;
else if (byte & BAD_AVG)
plotmask[ii][jj] = 5.0;
else if (byte & BAD_STD)
plotmask[ii][jj] = 6.0;
}
}
/* Set the colors */
numcol = 7;
maxcol = mincol + numcol - 1;
cpgscir(mincol, maxcol);
cpgqcr(0, &rr, &gg, &bb);
cpgscr(mincol + 0, rr, gg, bb); /* GOODDATA = background */
cpgscr(mincol + 1, 0.7, 0.7, 0.7); /* PADDING = light grey */
cpgscr(mincol + 2, 0.3, 0.3, 0.3); /* OLDMASK = dark grey */
cpgqcr(1, &rr, &gg, &bb);
cpgscr(mincol + 3, rr, gg, bb); /* USERZAP = foreground */
cpgscr(mincol + 4, 1.0, 0.0, 0.0); /* BAD+POW = red */
cpgscr(mincol + 5, 0.0, 0.0, 1.0); /* BAD+AVG = blue */
cpgscr(mincol + 6, 0.0, 1.0, 0.0); /* BAD+STD = green */
/* Prep the image */
for (page = 0; page <= 1; page++) {
xl = 0.0;
xh = numchan;
yl = 0.0;
yh = T;
locut = 0.0;
hicut = 6.0;
tr[2] = tr[4] = 0.0;
tr[1] = (xh - xl) / numchan;
tr[0] = xl - (tr[1] / 2);
tr[5] = (yh - yl) / numint;
tr[3] = yl - (tr[5] / 2);
if (page == 0) {
left = lm + 3.0 * ft + 6.0 * tt;
right = lm + 4.0 * ft + 6.0 * tt;
bottom = bm;
top = bm + fh;
} else {
cpgpage();
cpgiden();
left = 0.06;
right = 0.94;
bottom = 0.06;
top = 0.88;
}
cpgsvp(left, right, bottom, top);
cpgswin(xl, xh, yl, yh);
cpgimag(plotmask[0], numchan, numint, 1,
numchan, 1, numint, locut, hicut, tr);
cpgswin(xl, xh, yl, yh);
cpgbox("BNST", 0.0, 0, "BNST", 0.0, 0);
cpgmtxt("B", 2.6, 0.5, 0.5, "Channel");
if (page)
cpgmtxt("L", 2.1, 0.5, 0.5, "Time (s)");
xl = lof;
xh = hif;
yl = 0.0;
yh = numint;
cpgswin(xl, xh, yl, yh);
cpgbox("CMST", 0.0, 0, "CMST", 0.0, 0);
cpgmtxt("T", 1.8, 0.5, 0.5, "Frequency (MHz)");
cpgmtxt("R", 2.3, 0.5, 0.5, "Interval Number");
/* Add the Labels */
cpgsvp(0.0, 1.0, 0.0, 1.0);
cpgswin(0.0, 1.0, 0.0, 1.0);
cpgsch(0.8);
if (page == 0) {
cpgsci(mincol + 1);
cpgptxt(left, top + 0.1, 0.0, 0.0, "Padding");
cpgsci(mincol + 2);
cpgptxt(left, top + 0.08, 0.0, 0.0, "Old Mask");
cpgsci(mincol + 3);
cpgptxt(left, top + 0.06, 0.0, 0.0, "User Zap");
cpgsci(mincol + 4);
cpgptxt(right, top + 0.1, 0.0, 1.0, "Power");
cpgsci(mincol + 6);
cpgptxt(right, top + 0.08, 0.0, 1.0, "Sigma");
cpgsci(mincol + 5);
cpgptxt(right, top + 0.06, 0.0, 1.0, "Mean");
cpgsci(1);
} else {
cpgsci(mincol + 1);
cpgptxt(1.0 / 12.0, 0.955, 0.0, 0.5, "Padding");
cpgsci(mincol + 2);
cpgptxt(3.0 / 12.0, 0.955, 0.0, 0.5, "Old Mask");
cpgsci(mincol + 3);
cpgptxt(5.0 / 12.0, 0.955, 0.0, 0.5, "User Zap");
cpgsci(mincol + 4);
cpgptxt(7.0 / 12.0, 0.955, 0.0, 0.5, "Max Power");
cpgsci(mincol + 6);
cpgptxt(9.0 / 12.0, 0.955, 0.0, 0.5, "Data Sigma");
cpgsci(mincol + 5);
cpgptxt(11.0 / 12.0, 0.955, 0.0, 0.5, "Data Mean");
cpgsci(1);
cpgsch(0.9);
sprintf(temp, "Recommended Mask for '%-s'", idata->name);
cpgptxt(0.5, 0.985, 0.0, 0.5, temp);
}
}
vect_free(plotmask[0]);
vect_free(plotmask);
}
if (ct == 0)
printf("There are %d RFI instances.\n\n", numrfi);
if ((ct == 0 && rfips) || (ct == 1 && rfixwin)) { /* Plot the RFI instances */
int maxcol, mincol, numperpage = 25, numtoplot;
float dy = 0.035, top = 0.95, rr, gg, bb;
char temp[200];
qsort(rfivect, numrfi, sizeof(rfi), compare_rfi_freq);
/* qsort(rfivect, numrfi, sizeof(rfi), compare_rfi_sigma); */
for (ii = 0; ii <= (numrfi - 1) / numperpage; ii++) {
cpgpage();
cpgiden();
cpgsvp(0.0, 1.0, 0.0, 1.0);
cpgswin(0.0, 1.0, 0.0, 1.0);
cpgsch(0.8);
sprintf(temp, "%-s", idata->name);
cpgtext(0.05, 0.985, temp);
cpgsch(0.6);
sprintf(temp, "Freq (Hz)");
cpgptxt(0.03, 0.96, 0.0, 0.0, temp);
sprintf(temp, "Period (ms)");
cpgptxt(0.12, 0.96, 0.0, 0.0, temp);
sprintf(temp, "Sigma");
cpgptxt(0.21, 0.96, 0.0, 0.0, temp);
sprintf(temp, "Number");
cpgptxt(0.27, 0.96, 0.0, 0.0, temp);
cpgsvp(0.33, 0.64, top - dy, top);
cpgswin(lof, hif, 0.0, 1.0);
cpgbox("CIMST", 0.0, 0, "", 0.0, 0);
cpgmtxt("T", 2.5, 0.5, 0.5, "Frequency (MHz)");
cpgsvp(0.65, 0.96, top - dy, top);
cpgswin(0.0, T, 0.0, 1.0);
cpgbox("CIMST", 0.0, 0, "", 0.0, 0);
cpgmtxt("T", 2.5, 0.5, 0.5, "Time (s)");
cpgqcir(&mincol, &maxcol);
maxcol = mincol + 1;
cpgscir(mincol, maxcol);
cpgqcr(0, &rr, &gg, &bb);
cpgscr(mincol, rr, gg, bb); /* background */
cpgqcr(1, &rr, &gg, &bb);
/* cpgscr(maxcol, rr, gg, bb); foreground */
cpgscr(maxcol, 0.5, 0.5, 0.5); /* grey */
if (ii == (numrfi - 1) / numperpage)
numtoplot = numrfi % numperpage;
else
numtoplot = numperpage;
for (jj = 0; jj < numtoplot; jj++)
plot_rfi(rfivect + ii * numperpage + jj,
top - jj * dy, numint, numchan, T, lof, hif);
cpgsvp(0.33, 0.64, top - jj * dy, top - (jj - 1) * dy);
cpgswin(0.0, numchan, 0.0, 1.0);
cpgbox("BINST", 0.0, 0, "", 0.0, 0);
cpgmtxt("B", 2.5, 0.5, 0.5, "Channel");
cpgsvp(0.65, 0.96, top - jj * dy, top - (jj - 1) * dy);
cpgswin(0.0, numint, 0.0, 1.0);
cpgbox("BINST", 0.0, 0, "", 0.0, 0);
cpgmtxt("B", 2.5, 0.5, 0.5, "Interval");
}
}
cpgclos();
} /* Plot for loop */
/* Free our arrays */
vect_free(freqs);
vect_free(chans);
vect_free(times);
vect_free(ints);
vect_free(avg_chan_avg);
vect_free(std_chan_avg);
vect_free(pow_chan_avg);
vect_free(avg_int_avg);
vect_free(std_int_avg);
vect_free(pow_int_avg);
vect_free(avg_chan_med);
vect_free(std_chan_med);
vect_free(pow_chan_med);
vect_free(avg_int_med);
vect_free(std_int_med);
vect_free(pow_int_med);
vect_free(avg_chan_std);
vect_free(std_chan_std);
vect_free(pow_chan_std);
vect_free(avg_int_std);
vect_free(std_int_std);
vect_free(pow_int_std);
}
fftcand *search_fft(fcomplex * fft, int numfft, int lobin, int hibin,
int numharmsum, int numbetween,
presto_interptype interptype,
float norm, float sigmacutoff, int *numcands,
float *powavg, float *powvar, float *powmax)
/* This routine searches a short FFT of 'numfft' complex freqs */
/* and returns a candidate vector of fftcand structures containing */
/* information about the best candidates found. */
/* The routine uses either interbinning or interpolation as well */
/* as harmonic summing during the search. */
/* The number of candidates returned is either 'numcands' if != 0, */
/* or is determined automatically by 'sigmacutoff' -- which */
/* takes into account the number of bins searched. */
/* The returned vector is sorted in order of decreasing power. */
/* Arguments: */
/* 'fft' is the FFT to search (complex valued) */
/* 'numfft' is the number of complex points in 'fft' */
/* 'lobin' is the lowest Fourier freq to search */
/* 'hibin' is the highest Fourier freq to search */
/* 'numharmsum' the number of harmonics to sum during the search */
/* 'numbetween' the points to interpolate per bin */
/* 'interptype' is either INTERBIN or INTERPOLATE. */
/* INTERBIN = (interbinning) is fast but less sensitive. */
/* NOTE: INTERBINNING is conducted by this routine! */
/* INTERPOLATE = (Fourier interpolation) is slower but more */
/* sensitive. */
/* NOTE: The interpolation is assumed to ALREADY have been */
/* completed by the calling function! The easiest */
/* way is by zero-padding to 2*numfft and FFTing. */
/* If you use this method, make sure numfft is the */
/* original length rather than the interpolated */
/* length and also make sure numbetween is correct. */
/* 'norm' is the normalization constant to multiply each power by */
/* 'sigmacutoff' if the number of candidates will be determined */
/* automatically, is the minimum Gaussian significance of */
/* candidates to keep -- taking into account the number of */
/* bins searched */
/* 'numcands' if !0, is the number of candates to return. */
/* if 0, is a return value giving the number of candidates. */
/* 'powavg' is a return value giving the average power level */
/* 'powvar' is a return value giving the power level variance */
/* 'powmax' is a return value giving the maximum power */
{
int ii, jj, offset, numtosearch, dynamic = 0;
int numspread = 0, nc = 0, startnc = 10;
float powargr, powargi, *fullpows = NULL, *sumpows, ftmp;
double twobypi, minpow = 0.0, tmpminsig = 0.0, dr, davg, dvar;
fftcand *cands, newcand;
fcomplex *spread;
/* Override the value of numbetween if interbinning */
if (interptype == INTERBIN)
numbetween = 2;
norm = 1.0 / norm;
*powmax = 0.0;
/* Decide if we will manage the number of candidates */
if (*numcands > 0)
startnc = *numcands;
else {
dynamic = 1;
minpow = power_for_sigma(sigmacutoff, 1, hibin - lobin);
}
cands = (fftcand *) malloc(startnc * sizeof(fftcand));
for (ii = 0; ii < startnc; ii++)
cands[ii].sig = 0.0;
/* Prep some other values we will need */
dr = 1.0 / (double) numbetween;
twobypi = 2.0 / PI;
numtosearch = numfft * numbetween;
/* Spread and interpolate the fft */
numspread = numfft * numbetween + 1;
if (interptype == INTERPOLATE) { /* INTERPOLATE */
spread = fft;
} else { /* INTERBIN */
spread = gen_cvect(numspread);
spread_with_pad(fft, numfft, spread, numspread, numbetween, 0);
for (ii = 1; ii < numtosearch; ii += 2) {
spread[ii].r = twobypi * (spread[ii - 1].r - spread[ii + 1].r);
spread[ii].i = twobypi * (spread[ii - 1].i - spread[ii + 1].i);
}
}
spread[0].r = spread[numtosearch].r = 1.0;
spread[0].i = spread[numtosearch].i = 0.0;
/* First generate the original powers in order to */
/* calculate the statistics. Yes, this is inefficient... */
fullpows = gen_fvect(numtosearch);
for (ii = lobin, jj = 0; ii < hibin; ii++, jj++) {
ftmp = POWER(fft[ii].r, fft[ii].i) * norm;
fullpows[jj] = ftmp;
if (ftmp > *powmax)
*powmax = ftmp;
}
avg_var(fullpows, hibin - lobin, &davg, &dvar);
*powavg = davg;
*powvar = dvar;
fullpows[0] = 1.0;
for (ii = 1; ii < numtosearch; ii++)
fullpows[ii] = POWER(spread[ii].r, spread[ii].i) * norm;
if (interptype == INTERBIN)
vect_free(spread);
/* Search the raw powers */
for (ii = lobin * numbetween; ii < hibin * numbetween; ii++) {
if (fullpows[ii] > minpow) {
newcand.r = dr * (double) ii;
newcand.p = fullpows[ii];
newcand.sig = candidate_sigma(fullpows[ii], 1, hibin - lobin);
newcand.nsum = 1;
cands[startnc - 1] = newcand;
tmpminsig = percolate_fftcands(cands, startnc);
if (dynamic) {
nc++;
if (nc == startnc) {
startnc *= 2;
cands = (fftcand *) realloc(cands, startnc * sizeof(fftcand));
for (jj = nc; jj < startnc; jj++)
cands[jj].sig = 0.0;
}
} else {
minpow = cands[startnc - 1].p;
if (nc < startnc)
nc++;
}
}
}
/* If needed, sum and search the harmonics */
if (numharmsum > 1) {
sumpows = gen_fvect(numtosearch);
memcpy(sumpows, fullpows, sizeof(float) * numtosearch);
for (ii = 2; ii <= numharmsum; ii++) {
offset = ii / 2;
if (dynamic)
minpow = power_for_sigma(sigmacutoff, ii, hibin - lobin);
else
minpow = power_for_sigma(tmpminsig, ii, hibin - lobin);
for (jj = lobin * numbetween; jj < numtosearch; jj++) {
sumpows[jj] += fullpows[(jj + offset) / ii];
if (sumpows[jj] > minpow) {
newcand.r = dr * (double) jj;
newcand.p = sumpows[jj];
newcand.sig = candidate_sigma(sumpows[jj], ii, hibin - lobin);
newcand.nsum = ii;
cands[startnc - 1] = newcand;
tmpminsig = percolate_fftcands(cands, startnc);
if (dynamic) {
nc++;
if (nc == startnc) {
startnc *= 2;
cands = (fftcand *) realloc(cands, startnc * sizeof(fftcand));
for (jj = nc; jj < startnc; jj++)
cands[jj].sig = 0.0;
}
} else {
minpow = power_for_sigma(tmpminsig, ii, hibin - lobin);
if (nc < startnc)
nc++;
}
}
}
}
vect_free(sumpows);
}
vect_free(fullpows);
/* Chop off the unused parts of the dynamic array */
if (dynamic)
cands = (fftcand *) realloc(cands, nc * sizeof(fftcand));
*numcands = nc;
return cands;
}
