aboutsummaryrefslogtreecommitdiffstats
path: root/astro
diff options
context:
space:
mode:
authorAaron LI <aly@aaronly.me>2017-11-25 17:12:48 +0800
committerAaron LI <aly@aaronly.me>2017-11-25 17:12:48 +0800
commit71c183da0193040bd89ea6be1f52666ddbda5425 (patch)
treea6e65eba32b6a8c98dadb8f12cf4fc1885ec9064 /astro
parent8a351c1a0bd413dd055b7529e97fe4e7ca2654c3 (diff)
downloadatoolbox-71c183da0193040bd89ea6be1f52666ddbda5425.tar.bz2
Add astro/eor_window.py: Calc total power within the defined EoR window
Diffstat (limited to 'astro')
-rwxr-xr-xastro/eor_window.py287
-rwxr-xr-xastro/ps2d.py5
2 files changed, 289 insertions, 3 deletions
diff --git a/astro/eor_window.py b/astro/eor_window.py
new file mode 100755
index 0000000..39f645a
--- /dev/null
+++ b/astro/eor_window.py
@@ -0,0 +1,287 @@
+#!/usr/bin/env python3
+#
+# Copyright (c) Weitian LI <weitian@aaronly.me>
+# MIT license
+#
+
+"""
+Calculate the total power within the EoR window on the 2D power spectrum.
+
+The adopted EoR window definition is from [thyagarajan2013],Eq.(26),Fig.(11).
+
+.. [thyagarajan2013]
+ Thyagarajan et al. 2013, ApJ, 776, 6
+"""
+
+import argparse
+from functools import lru_cache
+
+import numpy as np
+from astropy.io import fits
+from astropy.cosmology import FlatLambdaCDM
+import astropy.constants as ac
+
+import matplotlib.pyplot as plt
+from matplotlib.backends.backend_agg import FigureCanvasAgg as FigureCanvas
+from matplotlib.figure import Figure
+
+
+plt.style.use("ggplot")
+
+# HI line frequency
+freq21cm = 1420.405751 # [MHz]
+# Adopted cosmology
+H0 = 71.0 # [km/s/Mpc]
+OmegaM0 = 0.27
+cosmo = FlatLambdaCDM(H0=H0, Om0=OmegaM0)
+
+
+@lru_cache()
+def freq2z(freq):
+ z = freq21cm / freq - 1.0
+ return z
+
+
+class PS2D:
+ """
+ 2D cylindrically averaged power spectrum; calculated by ``ps2d.py``.
+
+ Attributes
+ ----------
+ ps2d : 2D `~numpy.ndarray`
+ Shape: [n_k_los, n_k_perp]
+ """
+ def __init__(self, infile):
+ self.infile = infile
+ with fits.open(infile) as f:
+ self.header = f[0].header
+ self.ps2d = f[0].data[0, :, :] # errors ignored
+ self.freqc = self.header["Freq_C"]
+ self.freqmin = self.header["Freq_Min"]
+ self.freqmax = self.header["Freq_Max"]
+ self.bandwidth = self.freqmax - self.freqmin # [MHz]
+ self.zc = self.header["Z_C"]
+ self.pixelsize = self.header["PixSize"]
+ self.unit = self.header["BUNIT"]
+
+ @property
+ def k_perp(self):
+ dk = self.header["CDELT1P"]
+ nk = self.header["NAXIS1"]
+ return np.arange(nk) * dk
+
+ @property
+ def k_los(self):
+ dk = self.header["CDELT2P"]
+ nk = self.header["NAXIS2"]
+ return np.arange(nk) * dk
+
+ @property
+ def k_perp_min(self):
+ return self.k_perp[1] # ignore the first 0
+
+ @property
+ def k_perp_max(self):
+ return self.k_perp[-1]
+
+ @property
+ def k_los_min(self):
+ return self.k_los[1] # ignore the first 0
+
+ @property
+ def k_los_max(self):
+ return self.k_los[-1]
+
+ def sum_power(self, window):
+ """
+ Sum the power within the defined window.
+
+ NOTE: The cylindrical average should be accounted for.
+ """
+ k_perp = self.k_perp
+ k_los = self.k_los
+ dk_perp = k_perp[1] - k_perp[0]
+ dk_los = k_los[1] - k_los[0]
+ volume = np.zeros_like(self.ps2d)
+ volume[0, :] = 2*np.pi * k_perp * dk_perp * dk_los
+ for i in range(1, len(k_los)):
+ # The extra "2" to account for the average on +k_los and -k_los
+ volume[i, :] = 2*np.pi * k_perp * dk_perp * dk_los * 2
+
+ power = np.sum(self.ps2d * window * volume)
+ return power
+
+ def eor_window(self, fov, e,
+ k_perp_min=None, k_perp_max=None,
+ k_los_min=None, k_los_max=None):
+ """
+ Determine the EoR window region.
+
+ Parameters
+ ----------
+ fov : float
+ instrumental field of view (FoV)
+ Unit: [deg]
+ e : float
+ Thyagarajan proposed characteristic convolution width factor,
+ generally 0-3
+
+ Returns
+ -------
+ window : 2D bool `~numpy.ndarray`
+ 2D array mask of the same size of the power spectrum indicating
+ the defined EoR window region.
+ header : fits.Header
+ FITS header with the keywords recording the EoR window variables
+ """
+ if k_perp_min is None:
+ k_perp_min = self.k_perp_min
+ if k_perp_max is None:
+ k_perp_max = self.k_perp_max
+ if k_los_min is None:
+ k_los_min = self.k_los_min
+ if k_los_max is None:
+ k_los_max = self.k_los_max
+
+ window = np.ones_like(self.ps2d, dtype=bool)
+ k_perp = self.k_perp
+ k_los = self.k_los
+ k_wedge = self.wedge_edge(k_perp, fov=fov, e=e)
+ window[k_los < k_los_min, :] = False
+ window[k_los > k_los_max, :] = False
+ window[:, k_perp < k_perp_min] = False
+ window[:, k_perp > k_perp_max] = False
+ for i, k in enumerate(k_wedge):
+ window[k_los < k, i] = False
+
+ header = self.eor_window_header(fov=fov, e=e,
+ k_perp_min=k_perp_min,
+ k_perp_max=k_perp_max,
+ k_los_min=k_los_min,
+ k_los_max=k_los_max)
+ return (window, header)
+
+ def eor_window_header(self, fov, e, k_perp_min, k_perp_max,
+ k_los_min, k_los_max):
+ header = self.header.copy(strip=True)
+ header["FoV"] = (fov, "[deg] Field of view to determine EoR window")
+ header["e_ConvW"] = (e, "characteristic convolution width")
+ header["kper_min"] = (k_perp_min, "[Mpc^-1] minimum k_perp")
+ header["kper_max"] = (k_perp_max, "[Mpc^-1] maximum k_perp")
+ header["klos_min"] = (k_los_min, "[Mpc^-1] minimum k_los")
+ header["klos_max"] = (k_los_max, "[Mpc^-1] maximum k_los")
+ return header
+
+ def wedge_edge(self, k_perp, fov, e):
+ """
+ The boundary/edge between the EoR window (top-left) and the
+ foreground wedge (bottom-right).
+ """
+ Hz = cosmo.H(self.zc).value # [km/s/Mpc]
+ Dc = cosmo.comoving_distance(self.zc).value # [Mpc]
+ c = ac.c.to("km/s").value # [km/s]
+ coef = Hz * Dc / (c * (1+self.zc))
+ term1 = np.sin(np.deg2rad(fov)) * k_perp # [Mpc^-1]
+ term2 = ((2*np.pi * e * freq21cm / self.bandwidth) /
+ ((1 + self.zc) * Dc)) # [Mpc^-1]
+ k_los = coef * (term1 + term2)
+ return k_los
+
+ def plot(self, ax, fov, e,
+ k_perp_min=None, k_perp_max=None,
+ k_los_min=None, k_los_max=None,
+ colormap="jet"):
+ """
+ Plot the 2D power spectrum with EoR window marked on.
+ """
+ if k_perp_min is None:
+ k_perp_min = self.k_perp_min
+ if k_perp_max is None:
+ k_perp_max = self.k_perp_max
+ if k_los_min is None:
+ k_los_min = self.k_los_min
+ if k_los_max is None:
+ k_los_max = self.k_los_max
+
+ x = self.k_perp
+ y = self.k_los
+ y_wedge = self.wedge_edge(x, fov=fov, e=e)
+ title = "EoR Window (fov=%.1f[deg], e=%.1f)" % (fov, e)
+
+ # data
+ mappable = ax.pcolormesh(x[1:], y[1:],
+ np.log10(self.ps2d[1:, 1:]),
+ cmap=colormap)
+ # EoR window
+ ax.axvline(x=k_perp_min, color="black", linewidth=2, linestyle="--")
+ ax.axvline(x=k_perp_max, color="black", linewidth=2, linestyle="--")
+ ax.axhline(y=k_los_min, color="black", linewidth=2, linestyle="--")
+ ax.axhline(y=k_los_max, color="black", linewidth=2, linestyle="--")
+ ax.plot(x, y_wedge, color="black", linewidth=2, linestyle="--")
+ #
+ ax.set(xscale="log", yscale="log",
+ xlim=(x[1], x[-1]), ylim=(y[1], y[-1]),
+ xlabel=r"k$_{\perp}$ [Mpc$^{-1}$]",
+ ylabel=r"k$_{||}$ [Mpc$^{-1}$]",
+ title=title)
+ cb = ax.figure.colorbar(mappable, ax=ax, pad=0.01, aspect=30)
+ cb.ax.set_xlabel(r"[%s$^2$ Mpc$^3$]" % self.unit)
+ return ax
+
+
+def main():
+ parser = argparse.ArgumentParser(
+ description="Determine EoR window region and calculate total power")
+ parser.add_argument("-F", "--fov", dest="fov",
+ type=float, required=True,
+ help="instrumental FoV to determine the EoR window")
+ parser.add_argument("-e", "--conv-width", dest="conv_width",
+ type=float, default=3.0,
+ help="characteristic convolution width (default: 3.0)")
+ parser.add_argument("-p", "--k-perp-min", dest="k_perp_min", type=float,
+ help="minimum k wavenumber perpendicular to LoS; " +
+ "unit: [Mpc^-1]")
+ parser.add_argument("-P", "--k-perp-max", dest="k_perp_max", type=float,
+ help="maximum k wavenumber perpendicular to LoS")
+ parser.add_argument("-l", "--k-los-min", dest="k_los_min", type=float,
+ help="minimum k wavenumber along LoS")
+ parser.add_argument("-L", "--k-los-max", dest="k_los_max", type=float,
+ help="maximum k wavenumber along LoS")
+ parser.add_argument("--save-window", dest="save_window",
+ help="save the determined EoR window into FITS " +
+ "file with the provided filename")
+ parser.add_argument("--plot", dest="plot",
+ help="plot the 2D power spectrum with the " +
+ "determined EoR window marked, and save into " +
+ "the specified file")
+ parser.add_argument("infile", help="2D power spectrum FITS file")
+ args = parser.parse_args()
+
+ ps2d = PS2D(args.infile)
+ window, header = ps2d.eor_window(fov=args.fov, e=args.conv_width,
+ k_perp_min=args.k_perp_min,
+ k_perp_max=args.k_perp_max,
+ k_los_min=args.k_los_min,
+ k_los_max=args.k_los_max)
+ power = ps2d.sum_power(window)
+ print("Total power within EoR window: %g [%s]" % (power, ps2d.unit))
+
+ if args.save_window:
+ hdu = fits.PrimaryHDU(data=window.astype(np.int16), header=header)
+ hdu.writeto(args.save_window)
+ print("Saved EoR window into file: %s" % args.save_window)
+
+ if args.plot:
+ fig = Figure(figsize=(8, 8), dpi=150)
+ FigureCanvas(fig)
+ ax = fig.add_subplot(1, 1, 1)
+ ps2d.plot(ax=ax, fov=args.fov, e=args.conv_width,
+ k_perp_min=args.k_perp_min, k_perp_max=args.k_perp_max,
+ k_los_min=args.k_los_min, k_los_max=args.k_los_max)
+ fig.tight_layout()
+ fig.savefig(args.plot)
+ print("Plotted 2D PSD with EoR window and saved to: %s" % args.plot)
+
+
+if __name__ == "__main__":
+ main()
diff --git a/astro/ps2d.py b/astro/ps2d.py
index 718b59a..0bd0ac4 100755
--- a/astro/ps2d.py
+++ b/astro/ps2d.py
@@ -339,9 +339,8 @@ class PS2D:
Reference: Ref.[liu2014].Eq.(A9)
"""
dfreq = self.dfreq # [MHz]
- c = ac.c.si.value # [m/s]
- Ez = cosmo.efunc(self.zc)
- Hz = Ez * H0 * 1000.0 # [m/s/Mpc]
+ c = ac.c.to("km/s").value # [km/s]
+ Hz = cosmo.H(self.zc).value # [km/s/Mpc]
d_z = c * (1+self.zc)**2 * dfreq / Hz / freq21cm
return d_z