rework generate_plot script

This commit is contained in:
2025-03-10 08:26:16 +01:00
parent df79c8bdcc
commit 01ae4f9084
2 changed files with 692 additions and 667 deletions
+692
View File
@@ -0,0 +1,692 @@
import numpy as np
import pandas as pd
import itertools
from main.astrodatagui.db.StarsDB import StarDB
import matplotlib.ticker as tck
from matplotlib.pyplot import MaxNLocator
import matplotlib.pyplot as plt
from datetime import datetime
from errno import EEXIST
from os import makedirs, path
import shutil
def normalizePhase(phase, phaseMin = None, phaseMax = None):
if(phaseMin is None):
phaseMin = np.abs(np.min(phase))
if(phaseMax is None):
phaseMax = np.abs(np.max(phase))
return (phase + phaseMin) / (phaseMin + phaseMax) * (2)
def mkdir_p(mypath):
'''Creates a directory. equivalent to using mkdir -p on the command line'''
try:
makedirs(mypath)
except OSError as exc: # Python >2.5
if exc.errno == EEXIST and path.isdir(mypath):
pass
else: raise
fileName = "datav4.cff"
data = pd.read_pickle(fileName)
binList = [10, 20, 30]
spType = ["M", "K", "G", "F"]
useKepler = True
useK2 = True
useTESS = True
showSourceFilter = np.full(len(data), False)
if(useKepler):
showKepler = data["Source"] == "Kepler"
showSourceFilter |= showKepler
if(useK2):
showK2 = data["Source"] == "K2"
showSourceFilter |= showK2
if(useTESS):
showTESS = data["Source"] == "TESS"
showSourceFilter |= showTESS
current = datetime.now()
date = f"{current.year}-{current.month}-{current.day}"
time = f"{current.hour}-{current.minute}-{current.second}"
folderPath = f"../{date}/"
#folderPath = f"G:/Meine Ablage/Masterthesis/{date}/"
mkdir_p(folderPath)
# remove any data that has no period
data = data[(data["FitType"] == "sine") | (data["FitType"] == "poly")]
# remove data with multiple minima/maxima present
filterArray = []
for ind, row in data.reset_index().iterrows():
if(len(row["pdcsapPeriodMinima"]) == len(row["pdcsapPeriodMaxima"]) and
len(row["pdcsapPeriodMinima"]) == 1):
filterArray.append(True)
else:
filterArray.append(False)
filterArray = np.array(filterArray)
data = data[filterArray]
validStarPeriodMap = []
starList = set(list(data["StarName"]))
def allValuesWithin3Std(values: list):
if(not values or len(values) == 1):
return True
return max(values) - min(values) <= 3*np.std(values)
def getMeanPeriod(values: list):
return np.mean(values)
for starName in starList:
periods = list(data[data["StarName"] == starName]["pdcsapPeriod"])
validStarPeriodMap.append({"StarName": starName,
"MeanPeriod": getMeanPeriod(periods),
"PeriodWithinStd": allValuesWithin3Std(periods)})
validStarPeriodMap = pd.DataFrame(validStarPeriodMap)
periodsCutList = [0.5, 1, 1.5, 2, 5, 10, 15, 20]
data = pd.merge(data, validStarPeriodMap, on="StarName")
starDB: StarDB = StarDB.getInstance("stars.db")
def plotBinsHistogram(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, bins, title, filename):
figHisto, ((axHisto)) = plt.subplots(nrows=1, ncols=1)
y, binEdges, _ = axHisto.hist(PDCSAPdataList, bins,
label=PDCSAPlabelList,
color=PDCSAPcolorList,
stacked=True,
range=[0, 2])
bincenters = 0.5*(binEdges[1:]+binEdges[:-1])
if(isinstance(y[0], np.ndarray)):
y = y[-1]
n_i = y
m_i = bincenters * np.pi
N = np.sum(n_i)
mean = np.sum(n_i * m_i)/N
stdDev = np.sqrt(np.sum(((n_i - mean)**2)) / (N-1))
menStd = np.sqrt(y)
if(np.isinf(stdDev) or np.isnan(stdDev)):
stdDev = np.mean(menStd)
if(np.isinf(stdDev) or np.isnan(stdDev)):
stdDev = 0
axHisto.bar(bincenters[y > 0], y[y > 0], width=0, color='r', yerr=stdDev)
try:
axHisto.set_ylim(0, max(y) + stdDev)
except Exception as e:
print(e)
print(max(y), stdDev)
quit()
axHisto.set_ylabel("Num. flares")
axHisto.set_xlabel("Phase")
axHisto.set_title(title)
axHisto.xaxis.set_major_formatter(tck.FormatStrFormatter('%g $\pi$'))
axHisto.xaxis.set_major_locator(MaxNLocator(5))
axHisto.legend()
axHistoPhase = axHisto.twinx()
secAxisXdata = np.linspace(0, 2, num=10000)
secAxisYdata = np.cos(secAxisXdata*np.pi) + 1
axHistoPhase.plot(secAxisXdata, secAxisYdata)
axHistoPhase.set_ylim(0, 7)
plt.savefig(filename)
plt.close()
def plotFlarePhasePeakHistogram(xData, yData, bins, maxY, title, filename):
figFlarepeakHist, ((axFlarepeakHist)) = plt.subplots(nrows=1, ncols=1)
axFlarepeakHist.set_ylabel("Flare peak")
axFlarepeakHist.set_xlabel("Phase")
H, xedges, yedges = np.histogram2d(xData, yData, bins=bins, range=[[0, 2], [0.99, maxY]])
cmax = 11
H_clipped = np.clip(H, None, cmax)
im = axFlarepeakHist.imshow(H_clipped.T, origin='lower', interpolation='nearest',
extent=[xedges[0], xedges[-1], yedges[0], yedges[-1]],
aspect='auto', cmap='viridis')
figFlarepeakHist.colorbar(im, label='Counts', ax=axFlarepeakHist)
axFlarepeakHist.xaxis.set_major_formatter(tck.FormatStrFormatter('%g $\pi$'))
axFlarepeakHist.xaxis.set_major_locator(MaxNLocator(5))
axFlarepeakHist.set_title(title)
plt.savefig(filename)
plt.close()
def plotFlarePeaks(plotdata, filters, labels, colors, maxY, title, filename):
if(isinstance(labels, list) and isinstance(colors, list)):
_, ((axFlarePeaks)) = plt.subplots(nrows=1, ncols=1)
for f, l, c in zip(filters, labels, colors):
axFlarePeaks.scatter(plotdata[f]["PDCSAPNormPhase"],
plotdata[f]["Peak"],
label=l, color=c)
elif(isinstance(labels, str) and isinstance(colors, str)):
_, ((axFlarePeaks)) = plt.subplots(nrows=1, ncols=1)
if(filters is None):
axFlarePeaks.scatter(plotdata[:]["PDCSAPNormPhase"],
plotdata[:]["Peak"],
label=labels, color=colors)
else:
axFlarePeaks.scatter(plotdata[filters]["PDCSAPNormPhase"],
plotdata[filters]["Peak"],
label=labels, color=colors)
else:
return
axFlarePeaks.set_xlim(0, 2)
axFlarePeaks.set_ylim(0.99, maxY)
axFlarePeaks.set_ylabel("Flare peak")
axFlarePeaks.set_xlabel("Phase")
axFlarePeaks.set_title(title)
axFlarePeaks.xaxis.set_major_formatter(tck.FormatStrFormatter('%g $\pi$'))
axFlarePeaks.xaxis.set_major_locator(MaxNLocator(5))
axFlarePeaks.legend()
plt.savefig(filename)
plt.close()
def setupxyDataAndDataList(PDCSAPdataList2dhistPhase, PDCSAPdataList2dhistPeak,
PDCSAPdataList=None, dataFilter=None):
xData = pd.DataFrame()
yData = pd.DataFrame()
for aX, aY in zip(PDCSAPdataList2dhistPhase, PDCSAPdataList2dhistPeak):
xData = pd.concat([xData, aX], ignore_index=True)
yData = pd.concat([yData, aY], ignore_index=True)
xData = np.asarray(xData.values)[:,0]
yData = np.asarray(yData.values)[:,0]
if(PDCSAPdataList is not None):
if(dataFilter is not None):
PDCSAPdataList.append(pdcsapbinningData[dataFilter]["PDCSAPNormPhase"])
else:
PDCSAPdataList.append(pdcsapbinningData[:]["PDCSAPNormPhase"])
return xData, yData, PDCSAPdataList
def generatePlots(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, histogramTitleArg, histogramFilenameArg,
xData, yData, peak2DHistogramTitleArg, peak2DfilenameArg,
plotdata, filters, flarePlotLabels, flarePlotColors, flarePlotTitleArg, flarePlotFilenameArg):
for bins in binList:
histogramTitle = histogramTitleArg.replace("@bins", str(bins))
histogramFilename = histogramFilenameArg.replace("@bins", str(bins))
plotBinsHistogram(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, bins, histogramTitle, histogramFilename)
for maxY in [1.05, 1.1, 1.2, 1.5, 2, 2.5, 3, 5, max(yData)]:
peak2DHistogramTitle = peak2DHistogramTitleArg.replace("@bins", str(bins))
peak2Dfilename = peak2DfilenameArg.replace("@bins", str(bins)).replace("@maxY", str(maxY))
plotFlarePhasePeakHistogram(xData, yData, bins, maxY, peak2DHistogramTitle, peak2Dfilename)
for maxY in [1.05, 1.1, 1.2, 1.5, 2, 2.5, 3, 5, max(yData)]:
flarePlotTitle = flarePlotTitleArg
flarePlotFilename = flarePlotFilenameArg.replace("@maxY", str(maxY))
plotFlarePeaks(plotdata, filters, flarePlotLabels, flarePlotColors, maxY, flarePlotTitle, flarePlotFilename)
# All stars
for starName in starDB.getAllStars():
finalData = pd.DataFrame()
starNameR = starName.replace('*', '_star_')
nameFilter = data["StarName"] == starName
nameFilter &= showSourceFilter
finalData = pd.concat([finalData, data[nameFilter]], ignore_index=True)
pdcsapbinningData = []
locFolder = f"{folderPath}/stars/{starNameR}/"
mkdir_p(f"{locFolder}/")
csvFile = open(f"{locFolder}/{starNameR}.csv", "a")
csvFile.write("Star Name,Spectral Type,Source,File,Flare Time,Flare Peak,Period,Normalized Phase of Peak,Peak in Period")
csvFile.write("\n")
for ind, row in finalData.reset_index().iterrows():
PDCSAPminOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][0]
PDCSAPmaxOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][1]
if(len(row["pdcsapFoldedPeaksPhasePair"]) > 0):
pdcsapVals = pd.DataFrame(row["pdcsapFoldedPeaksPhasePair"])
for td, peak, pv in zip(pdcsapVals["Phase"], pdcsapVals["Peak"], row["pdcsapPeaks"]):
normPhase = normalizePhase(td.value, np.abs(PDCSAPminOrigPhase), np.abs(PDCSAPmaxOrigPhase))
csvFile.write(f"{row['StarName']},{row['SpType']},{row['Source']},{row['FilePath']},{pv['FlarePeakTime']},{pv['FlarePeak']},{row['pdcsapPeriod']},{normPhase},{peak['FlarePeak']}")
csvFile.write("\n")
pdcsapbinningData.append({"SpType": f'{row["SpType"][0:2] if len(row["SpType"]) > 1 else row["SpType"][0]}',
"PDCSAPNormPhase": normPhase,
"Peak": peak["FlarePeak"]})
if(peak["FlarePeak"] > 100):
print(row["StarName"], "has over 100 peak")
csvFile.close()
pdcsapbinningData = pd.DataFrame(pdcsapbinningData)
PDCSAPdataList = []
PDCSAPlabelList = []
PDCSAPcolorList = []
PDCSAPdataList2dhistPhase = []
PDCSAPdataList2dhistPeak = []
if(len(pdcsapbinningData) < 1):
continue
else:
if(pdcsapbinningData["SpType"][0][0] == "M"):
color = "red"
elif(pdcsapbinningData["SpType"][0][0] == "K"):
color = "orange"
elif(pdcsapbinningData["SpType"][0][0] == "G"):
color = "yellow"
elif(pdcsapbinningData["SpType"][0][0] == "F"):
color = "greenyellow"
else:
color = "gray"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[:]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[:]["Peak"])
xData, yData, PDCSAPdataList = setupxyDataAndDataList(PDCSAPdataList2dhistPhase, PDCSAPdataList2dhistPeak, PDCSAPdataList)
PDCSAPlabelList.append(f"{starName}")
PDCSAPcolorList.append(color)
plotdata = pdcsapbinningData
filters = None
flarePlotLabels = f"{starName}"
flarePlotColors = color
generatePlots(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, f"Flare count in phase of {starName} with @bins bins", f"{locFolder}/{starNameR}-Flarecount-@bins_Bins.png",
xData, yData, f"Flare peak per phase histogram of {starName} with @bins bins", f"{locFolder}/{starNameR}-Flarepeaks-@bins_Bins_maxY-@maxY.png",
plotdata, filters, flarePlotLabels, flarePlotColors, f"Flare peaks per phase of {starName}", f"{locFolder}/{starNameR}-Flarepeaks_maxY-@maxY.png")
for comboLength in range(1, len(spType) + 1):
for combo in itertools.combinations(spType, comboLength):
for maxFlarePeak in [1.01, 1.05, 1.1, 1.25, 1.5]:
# max Flare Peak cut
finalDataMaxFlarePeak = pd.DataFrame()
if("M" in combo):
Mfilter = data["SpType"].str.startswith("M")
Mfilter &= showSourceFilter
finalDataMaxFlarePeak = pd.concat([finalDataMaxFlarePeak, data[Mfilter]], ignore_index=True)
if("K" in combo):
Kfilter = data["SpType"].str.startswith("K")
Kfilter &= showSourceFilter
finalDataMaxFlarePeak = pd.concat([finalDataMaxFlarePeak, data[Kfilter]], ignore_index=True)
if("G" in combo):
Gfilter = data["SpType"].str.startswith("G")
Gfilter &= showSourceFilter
finalDataMaxFlarePeak = pd.concat([finalDataMaxFlarePeak, data[Gfilter]], ignore_index=True)
if("F" in combo):
Ffilter = data["SpType"].str.startswith("F")
Ffilter &= showSourceFilter
finalDataMaxFlarePeak = pd.concat([finalDataMaxFlarePeak, data[Ffilter]], ignore_index=True)
numStars = len(set(finalDataMaxFlarePeak["StarName"]))
pdcsapbinningData = []
locFolder = f"{folderPath}/{''.join(combo)}/maxFlarePeaks/{maxFlarePeak}/"
mkdir_p(f"{locFolder}/")
csvFile = open(f"{locFolder}/{''.join(combo)}_maxFlarePeak_{maxFlarePeak}.csv", "a")
csvFile.write("Star Name,Spectral Type,Source,File,Flare Time,Flare Peak,Period,Normalized Phase of Peak,Peak in Period")
csvFile.write("\n")
for ind, row in finalDataMaxFlarePeak.reset_index().iterrows():
PDCSAPminOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][0]
PDCSAPmaxOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][1]
if(len(row["pdcsapFoldedPeaksPhasePair"]) > 0):
pdcsapVals = pd.DataFrame(row["pdcsapFoldedPeaksPhasePair"])
for td, peak, pv in zip(pdcsapVals["Phase"], pdcsapVals["Peak"], row["pdcsapPeaks"]):
if(peak["FlarePeak"] <= maxFlarePeak):
normPhase = normalizePhase(td.value, np.abs(PDCSAPminOrigPhase), np.abs(PDCSAPmaxOrigPhase))
csvFile.write(f"{row['StarName']},{row['SpType']},{row['Source']},{row['FilePath']},{pv['FlarePeakTime']},{pv['FlarePeak']},{row['pdcsapPeriod']},{normPhase},{peak['FlarePeak']}")
csvFile.write("\n")
pdcsapbinningData.append({"SpType": row["SpType"][0],
"PDCSAPNormPhase": normPhase,
"Peak": peak["FlarePeak"]})
if(peak["FlarePeak"] > 100):
print(row["StarName"], "has over 100 peak")
csvFile.close()
if(len(pdcsapbinningData) < 1):
continue
pdcsapbinningData = pd.DataFrame(pdcsapbinningData)
PDCSAPdataList = []
PDCSAPlabelList = []
PDCSAPcolorList = []
PDCSAPdataList2dhistPhase = []
PDCSAPdataList2dhistPeak = []
PDCSAPlabelList2dhist = []
PDCSAPcolorList2dhist = []
plotFilters = []
if("M" in combo):
Mfilter = pdcsapbinningData["SpType"] == "M"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Mfilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Mfilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Mfilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("M Stars")
PDCSAPcolorList.append("red")
plotFilters.append(Mfilter)
if("K" in combo):
Kfilter = pdcsapbinningData["SpType"] == "K"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Kfilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Kfilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Kfilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("K Stars")
PDCSAPcolorList.append("orange")
plotFilters.append(Kfilter)
if("G" in combo):
Gfilter = pdcsapbinningData["SpType"] == "G"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Gfilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Gfilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Gfilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("G Stars")
PDCSAPcolorList.append("yellow")
plotFilters.append(Gfilter)
if("F" in combo):
Ffilter = pdcsapbinningData["SpType"] == "F"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Ffilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Ffilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Ffilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("F Stars")
PDCSAPcolorList.append("greenyellow")
plotFilters.append(Ffilter)
xData, yData, _ = setupxyDataAndDataList(PDCSAPdataList2dhistPhase, PDCSAPdataList2dhistPeak)
plotdata = pdcsapbinningData
filters = plotFilters
generatePlots(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, f"Flare count per phase of {', '.join(combo)} type stars with @bins bins ({numStars} stars)", f"{locFolder}/{''.join(combo)}_maxFlarePeak_{maxFlarePeak}-Flarecount-@bins_Bins.png",
xData, yData, f"Flare peak per phase histogram of {', '.join(combo)} type stars with @bins bins ({numStars} stars)", f"{locFolder}/{''.join(combo)}_maxFlarePeak_{maxFlarePeak}-Flarepeaks-@bins_Bins_maxY-@maxY.png",
plotdata, filters, PDCSAPlabelList, PDCSAPcolorList, f"Flare peaks per phase of {', '.join(combo)} type stars ({numStars} stars)", f"{locFolder}/{''.join(combo)}_maxFlarePeak_{maxFlarePeak}-Flarepeaks_maxY-@maxY.png")
# all flare peaks
finalDataAllFlarePeaks = pd.DataFrame()
if("M" in combo):
Mfilter = data["SpType"].str.startswith("M")
Mfilter &= showSourceFilter
finalDataAllFlarePeaks = pd.concat([finalDataAllFlarePeaks, data[Mfilter]], ignore_index=True)
if("K" in combo):
Kfilter = data["SpType"].str.startswith("K")
Kfilter &= showSourceFilter
finalDataAllFlarePeaks = pd.concat([finalDataAllFlarePeaks, data[Kfilter]], ignore_index=True)
if("G" in combo):
Gfilter = data["SpType"].str.startswith("G")
Gfilter &= showSourceFilter
finalDataAllFlarePeaks = pd.concat([finalDataAllFlarePeaks, data[Gfilter]], ignore_index=True)
if("F" in combo):
Ffilter = data["SpType"].str.startswith("F")
Ffilter &= showSourceFilter
finalDataAllFlarePeaks = pd.concat([finalDataAllFlarePeaks, data[Ffilter]], ignore_index=True)
numStars = len(set(finalDataAllFlarePeaks["StarName"]))
pdcsapbinningData = []
locFolder = f"{folderPath}/{''.join(combo)}/"
mkdir_p(f"{locFolder}/")
csvFile = open(f"{locFolder}/{''.join(combo)}.csv", "a")
csvFile.write("Star Name,Spectral Type,Source,File,Flare Time,Flare Peak,Period,Normalized Phase of Peak,Peak in Period")
csvFile.write("\n")
for ind, row in finalDataAllFlarePeaks.reset_index().iterrows():
PDCSAPminOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][0]
PDCSAPmaxOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][1]
if(len(row["pdcsapFoldedPeaksPhasePair"]) > 0):
pdcsapVals = pd.DataFrame(row["pdcsapFoldedPeaksPhasePair"])
for td, peak, pv in zip(pdcsapVals["Phase"], pdcsapVals["Peak"], row["pdcsapPeaks"]):
normPhase = normalizePhase(td.value, np.abs(PDCSAPminOrigPhase), np.abs(PDCSAPmaxOrigPhase))
csvFile.write(f"{row['StarName']},{row['SpType']},{row['Source']},{row['FilePath']},{pv['FlarePeakTime']},{pv['FlarePeak']},{row['pdcsapPeriod']},{normPhase},{peak['FlarePeak']}")
csvFile.write("\n")
pdcsapbinningData.append({"SpType": row["SpType"][0],
"PDCSAPNormPhase": normPhase,
"Peak": peak["FlarePeak"]})
if(peak["FlarePeak"] > 100):
print(row["StarName"], "has over 100 peak")
csvFile.close()
pdcsapbinningData = pd.DataFrame(pdcsapbinningData)
PDCSAPdataList = []
PDCSAPlabelList = []
PDCSAPcolorList = []
PDCSAPdataList2dhistPhase = []
PDCSAPdataList2dhistPeak = []
PDCSAPlabelList2dhist = []
PDCSAPcolorList2dhist = []
plotFilters = []
if("M" in combo):
Mfilter = pdcsapbinningData["SpType"] == "M"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Mfilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Mfilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Mfilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("M Stars")
PDCSAPcolorList.append("red")
plotFilters.append(Mfilter)
if("K" in combo):
Kfilter = pdcsapbinningData["SpType"] == "K"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Kfilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Kfilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Kfilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("K Stars")
PDCSAPcolorList.append("orange")
plotFilters.append(Kfilter)
if("G" in combo):
Gfilter = pdcsapbinningData["SpType"] == "G"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Gfilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Gfilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Gfilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("G Stars")
PDCSAPcolorList.append("yellow")
plotFilters.append(Gfilter)
if("F" in combo):
Ffilter = pdcsapbinningData["SpType"] == "F"
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[Ffilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[Ffilter]["Peak"])
PDCSAPdataList.append(pdcsapbinningData[Ffilter]["PDCSAPNormPhase"])
PDCSAPlabelList.append("F Stars")
PDCSAPcolorList.append("greenyellow")
plotFilters.append(Ffilter)
xData, yData, _ = setupxyDataAndDataList(PDCSAPdataList2dhistPhase, PDCSAPdataList2dhistPeak)
plotdata = pdcsapbinningData
filters = plotFilters
generatePlots(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, f"Flare count per phase of {', '.join(combo)} type stars with @bins bins ({numStars} stars)", f"{locFolder}/{''.join(combo)}-Flarecount-@bins_Bins.png",
xData, yData, f"Flare peak per phase histogram of {', '.join(combo)} type stars with @bins bins ({numStars} stars)", f"{locFolder}/{''.join(combo)}-Flarepeaks-@bins_Bins_maxY-@maxY.png",
plotdata, filters, PDCSAPlabelList, PDCSAPcolorList, f"Flare peaks per phase of {', '.join(combo)} type stars ({numStars} stars)", f"{locFolder}/{''.join(combo)}-Flarepeaks_maxY-@maxY.png")
if(comboLength == 1):
spTypes = [f"{combo}0", f"{combo}1", f"{combo}2", f"{combo}3", f"{combo}4", f"{combo}5", f"{combo}6", f"{combo}7", f"{combo}8", f"{combo}9"]
match combo:
case "M":
color = "red"
case "K":
color = "orange"
case "G":
color = "yellow"
case "F":
color = "greenyellow"
for spTyp in spTypes:
finalDataAccSpTypes = pd.DataFrame()
typeFilter = data["SpType"].str.startswith(spTyp)
numStars = len(set(data[typeFilter]["StarName"]))
typeFilter &= showSourceFilter
finalDataAccSpTypes = pd.concat([finalDataAccSpTypes, data[typeFilter]], ignore_index=True)
pdcsapbinningData = []
locFolder = f"{folderPath}/{spTyp}/"
mkdir_p(f"{locFolder}/")
csvFile = open(f"{locFolder}/{spTyp}.csv", "a")
csvFile.write("Star Name,Spectral Type,Source,File,Flare Time,Flare Peak,Period,Normalized Phase of Peak,Peak in Period")
csvFile.write("\n")
for ind, row in finalDataAccSpTypes.reset_index().iterrows():
PDCSAPminOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][0]
PDCSAPmaxOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][1]
if(len(row["pdcsapFoldedPeaksPhasePair"]) > 0):
pdcsapVals = pd.DataFrame(row["pdcsapFoldedPeaksPhasePair"])
for td, peak, pv in zip(pdcsapVals["Phase"], pdcsapVals["Peak"], row["pdcsapPeaks"]):
normPhase = normalizePhase(td.value, np.abs(PDCSAPminOrigPhase), np.abs(PDCSAPmaxOrigPhase))
csvFile.write(f"{row['StarName']},{row['SpType']},{row['Source']},{row['FilePath']},{pv['FlarePeakTime']},{pv['FlarePeak']},{row['pdcsapPeriod']},{normPhase},{peak['FlarePeak']}")
csvFile.write("\n")
pdcsapbinningData.append({"SpType": f'{row["SpType"][0]}{row["SpType"][1]}',
"PDCSAPNormPhase": normPhase,
"Peak": peak["FlarePeak"]})
if(peak["FlarePeak"] > 100):
print(row["StarName"], "has over 100 peak")
csvFile.close()
pdcsapbinningData = pd.DataFrame(pdcsapbinningData)
PDCSAPdataList = []
PDCSAPlabelList = []
PDCSAPcolorList = []
PDCSAPdataList2dhistPhase = []
PDCSAPdataList2dhistPeak = []
PDCSAPlabelList2dhist = []
PDCSAPcolorList2dhist = []
try:
SpTypefilter = pdcsapbinningData["SpType"] == spTyp
except:
shutil.rmtree(locFolder)
continue
PDCSAPdataList2dhistPhase.append(pdcsapbinningData[SpTypefilter]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeak.append(pdcsapbinningData[SpTypefilter]["Peak"])
PDCSAPlabelList.append(f"{spTyp} Stars")
PDCSAPcolorList.append(color)
xData, yData, PDCSAPdataList = setupxyDataAndDataList(PDCSAPdataList2dhistPhase, PDCSAPdataList2dhistPeak, PDCSAPdataList, SpTypefilter)
plotdata = pdcsapbinningData
filters = SpTypefilter
generatePlots(PDCSAPdataList, PDCSAPlabelList, PDCSAPcolorList, f"Flare count in phase of {spTyp} type stars with @bins bins ({numStars} stars)", f"{locFolder}/{''.join(spTyp)}-Flarecount-@bins_Bins.png",
xData, yData, f"Flare peak per phase histogram of {spTyp} type stars with @bins bins ({numStars} stars)", f"{locFolder}/{''.join(spTyp)}-Flarepeaks-@bins_Bins_maxY-@maxY.png",
plotdata, filters, PDCSAPlabelList, PDCSAPcolorList, f"Flare peaks per phase of {spTyp} type stars ({numStars} stars)", f"{locFolder}/{''.join(spTyp)}-Flarepeaks_maxY-@maxY.png")
# per Period
for periodCut in periodsCutList:
finalDataPeriod = pd.DataFrame()
if("M" in combo):
Mfilter = data["SpType"].str.startswith("M")
Mfilter &= showSourceFilter
Mfilter &= data["PeriodWithinStd"] == True
finalDataPeriod = pd.concat([finalDataPeriod, data[Mfilter]], ignore_index=True)
if("K" in combo):
Kfilter = data["SpType"].str.startswith("K")
Kfilter &= showSourceFilter
Kfilter &= data["PeriodWithinStd"] == True
finalDataPeriod = pd.concat([finalDataPeriod, data[Kfilter]], ignore_index=True)
if("G" in combo):
Gfilter = data["SpType"].str.startswith("G")
Gfilter &= showSourceFilter
Gfilter &= data["PeriodWithinStd"] == True
finalDataPeriod = pd.concat([finalDataPeriod, data[Gfilter]], ignore_index=True)
if("F" in combo):
Ffilter = data["SpType"].str.startswith("F")
Ffilter &= showSourceFilter
Ffilter &= data["PeriodWithinStd"] == True
finalDataPeriod = pd.concat([finalDataPeriod, data[Ffilter]], ignore_index=True)
pdcsapbinningDataU = []
pdcsapbinningDataO = []
locFolder = f"{folderPath}/{''.join(combo)}/periodCuts/{periodCut}/"
mkdir_p(f"{locFolder}/")
csvFileU = open(f"{locFolder}/under_{periodCut}.csv", "a")
csvFileO = open(f"{locFolder}/over_{periodCut}.csv", "a")
csvFileU.write("Star Name,Spectral Type,Source,File,Flare Time,Flare Peak,Period,Mean Period,Normalized Phase of Peak,Peak in Period")
csvFileU.write("\n")
csvFileO.write("Star Name,Spectral Type,Source,File,Flare Time,Flare Peak,Period,Mean Period,Normalized Phase of Peak,Peak in Period")
csvFileO.write("\n")
for ind, row in finalDataPeriod.reset_index().iterrows():
PDCSAPminOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][0]
PDCSAPmaxOrigPhase = row["pdcsapFoldedFitPhaseStarEnd"][1]
if(len(row["pdcsapFoldedPeaksPhasePair"]) > 0):
pdcsapVals = pd.DataFrame(row["pdcsapFoldedPeaksPhasePair"])
for td, peak, pv in zip(pdcsapVals["Phase"], pdcsapVals["Peak"], row["pdcsapPeaks"]):
normPhase = normalizePhase(td.value, np.abs(PDCSAPminOrigPhase), np.abs(PDCSAPmaxOrigPhase))
if(row["MeanPeriod"] <= periodCut):
csvFileU.write(f"{row['StarName']},{row['SpType']},{row['Source']},{row['FilePath']},{pv['FlarePeakTime']},{pv['FlarePeak']},{row['pdcsapPeriod']},{row["MeanPeriod"]},{normPhase},{peak['FlarePeak']}")
csvFileU.write("\n")
pdcsapbinningDataU.append({"SpType": f'{row["SpType"][0]}',
"PDCSAPNormPhase": normPhase,
"Peak": peak["FlarePeak"]})
else:
csvFileO.write(f"{row['StarName']},{row['SpType']},{row['Source']},{row['FilePath']},{pv['FlarePeakTime']},{pv['FlarePeak']},{row['pdcsapPeriod']},{row["MeanPeriod"]},{normPhase},{peak['FlarePeak']}")
csvFileO.write("\n")
pdcsapbinningDataO.append({"SpType": f'{row["SpType"][0]}',
"PDCSAPNormPhase": normPhase,
"Peak": peak["FlarePeak"]})
if(peak["FlarePeak"] > 100):
print(row["StarName"], "has over 100 peak")
csvFileU.close()
csvFileO.close()
pdcsapbinningDataU = pd.DataFrame(pdcsapbinningDataU)
pdcsapbinningDataO = pd.DataFrame(pdcsapbinningDataO)
PDCSAPdataListU = []; PDCSAPdataListO = []
PDCSAPlabelList = []
PDCSAPcolorList = []
PDCSAPdataList2dhistPhaseU = []; PDCSAPdataList2dhistPeakU = []
PDCSAPdataList2dhistPhaseO = []; PDCSAPdataList2dhistPeakO = []
plotFiltersU = []; plotFiltersO = [];
if("M" in combo):
PDCSAPlabelList.append("M Stars")
PDCSAPcolorList.append("red")
if("K" in combo):
PDCSAPlabelList.append("K Stars")
PDCSAPcolorList.append("orange")
if("G" in combo):
PDCSAPlabelList.append("G Stars")
PDCSAPcolorList.append("yellow")
if("F" in combo):
PDCSAPlabelList.append("F Stars")
PDCSAPcolorList.append("greenyellow")
if(len(pdcsapbinningDataU) > 0):
if("M" in combo):
MfilterU = pdcsapbinningDataU["SpType"] == "M"
PDCSAPdataList2dhistPhaseU.append(pdcsapbinningDataU[MfilterU]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakU.append(pdcsapbinningDataU[MfilterU]["Peak"])
PDCSAPdataListU.append(pdcsapbinningDataU[MfilterU]["PDCSAPNormPhase"])
plotFiltersU.append(MfilterU)
if("K" in combo):
KfilterU = pdcsapbinningDataU["SpType"] == "K"
PDCSAPdataList2dhistPhaseU.append(pdcsapbinningDataU[KfilterU]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakU.append(pdcsapbinningDataU[KfilterU]["Peak"])
PDCSAPdataListU.append(pdcsapbinningDataU[KfilterU]["PDCSAPNormPhase"])
plotFiltersU.append(KfilterU)
if("G" in combo):
GfilterU = pdcsapbinningDataU["SpType"] == "G"
PDCSAPdataList2dhistPhaseU.append(pdcsapbinningDataU[GfilterU]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakU.append(pdcsapbinningDataU[GfilterU]["Peak"])
PDCSAPdataListU.append(pdcsapbinningDataU[GfilterU]["PDCSAPNormPhase"])
plotFiltersU.append(GfilterU)
if("F" in combo):
FfilterU = pdcsapbinningDataU["SpType"] == "F"
PDCSAPdataList2dhistPhaseU.append(pdcsapbinningDataU[FfilterU]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakU.append(pdcsapbinningDataU[FfilterU]["Peak"])
PDCSAPdataListU.append(pdcsapbinningDataU[FfilterU]["PDCSAPNormPhase"])
plotFiltersU.append(FfilterU)
if(len(pdcsapbinningDataO) > 0):
if("M" in combo):
MfilterO = pdcsapbinningDataO["SpType"] == "M"
PDCSAPdataList2dhistPhaseO.append(pdcsapbinningDataO[MfilterO]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakO.append(pdcsapbinningDataO[MfilterO]["Peak"])
PDCSAPdataListO.append(pdcsapbinningDataO[MfilterO]["PDCSAPNormPhase"])
plotFiltersO.append(MfilterO)
if("K" in combo):
KfilterO = pdcsapbinningDataO["SpType"] == "K"
PDCSAPdataList2dhistPhaseO.append(pdcsapbinningDataO[KfilterO]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakO.append(pdcsapbinningDataO[KfilterO]["Peak"])
PDCSAPdataListO.append(pdcsapbinningDataO[KfilterO]["PDCSAPNormPhase"])
plotFiltersO.append(KfilterO)
if("G" in combo):
GfilterO = pdcsapbinningDataO["SpType"] == "G"
PDCSAPdataList2dhistPhaseO.append(pdcsapbinningDataO[GfilterO]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakO.append(pdcsapbinningDataO[GfilterO]["Peak"])
PDCSAPdataListO.append(pdcsapbinningDataO[GfilterO]["PDCSAPNormPhase"])
plotFiltersO.append(GfilterO)
if("F" in combo):
FfilterO = pdcsapbinningDataO["SpType"] == "F"
PDCSAPdataList2dhistPhaseO.append(pdcsapbinningDataO[FfilterO]["PDCSAPNormPhase"])
PDCSAPdataList2dhistPeakO.append(pdcsapbinningDataO[FfilterO]["Peak"])
PDCSAPdataListO.append(pdcsapbinningDataO[FfilterO]["PDCSAPNormPhase"])
plotFiltersO.append(FfilterO)
if(len(PDCSAPdataList2dhistPhaseU) > 0 and len(PDCSAPdataList2dhistPeakU) > 0):
xDataU, yDataU, _ = setupxyDataAndDataList(PDCSAPdataList2dhistPhaseU, PDCSAPdataList2dhistPeakU)
if(len(PDCSAPdataListU) > 0 and len(xDataU) > 0 and len(yDataU) > 0):
generatePlots(PDCSAPdataListU, PDCSAPlabelList, PDCSAPcolorList, f"Flare count per phase of {', '.join(combo)} type stars with @bins bins (Rot. Period under {periodCut} days)", f"{locFolder}/{''.join(combo)}-Flarecount-Period_u_{periodCut}-@bins_Bins.png",
xDataU, yDataU, f"Flare peak per phase histogram of {', '.join(combo)} type stars with @bins bins (Rot. Period under {periodCut} days)", f"{locFolder}/{''.join(combo)}-Flarepeaks-Period_u_{periodCut}-@bins_Bins_maxY-@maxY.png",
pdcsapbinningDataU, plotFiltersU, PDCSAPlabelList, PDCSAPcolorList, f"Flare peaks per phase of {', '.join(combo)} type stars (Rot. Period under {periodCut} days)", f"{locFolder}/{''.join(combo)}-Flarepeaks-Period_u_{periodCut}-maxY_@maxY.png")
if(len(PDCSAPdataList2dhistPhaseO) > 0 and len(PDCSAPdataList2dhistPeakO) > 0):
xDataO, yDataO, _ = setupxyDataAndDataList(PDCSAPdataList2dhistPhaseO, PDCSAPdataList2dhistPeakO)
if(len(PDCSAPdataListO) > 0 and len(xDataO) > 0 and len(yDataO) > 0):
generatePlots(PDCSAPdataListO, PDCSAPlabelList, PDCSAPcolorList, f"Flare count per phase of {', '.join(combo)} type stars with @bins bins (Rot. Period over {periodCut} days)", f"{locFolder}/{''.join(combo)}-Flarecount-Period_o_{periodCut}-@bins_Bins.png",
xDataO, yDataO, f"Flare peak per phase histogram of {', '.join(combo)} type stars with @bins bins (Rot. Period over {periodCut} days)", f"{locFolder}/{''.join(combo)}-Flarepeaks-Period_o_{periodCut}-@bins_Bins_maxY-@maxY.png",
pdcsapbinningDataO, plotFiltersO, PDCSAPlabelList, PDCSAPcolorList, f"Flare peaks per phase of {', '.join(combo)} type stars (Rot. Period over {periodCut} days)", f"{locFolder}/{''.join(combo)}-Flarepeaks-Period_o_{periodCut}-maxY_@maxY.png")