Fix ell range (#60)

* fix ell scales and comments regarding bandpass construction * update dataset * setting test back to low accuracy * setting low accuracy camb params in the Fisher section of the notebook * fixing notebook kernel * update README
simonsobs · Jul 10, 2023 · 6527b41 · 6527b41
1 parent 4fdfb2e
commit 6527b41
Show file tree

Hide file tree

Showing 7 changed files with 180 additions and 139 deletions.
diff --git a/README.rst b/README.rst
@@ -76,6 +76,8 @@ directory.
 Simulation releases
 -------------------
 
+- ``v0.8`` release has spectra produced with high accuracy camb parameters up to ``ell = 9000``, the foreground components are integrated in frequency using top-hat bandpasses, stored in the ``sacc`` files
+
 - ``v0.7.1`` is a bugfix release of ``v0.7`` release where the mono-frequency and dirac bandpasses
   have been correctly set with the ``sacc`` files
 

diff --git a/examples/mflike_example.yaml b/examples/mflike_example.yaml
@@ -4,7 +4,7 @@ debug: True
 
 likelihood:
   mflike.MFLike:
-    input_file: data_sacc_00044.fits
+    input_file: LAT_simu_sacc_00044.fits
     cov_Bbl_file: data_sacc_w_covar_and_Bbl.fits
 
 params:
@@ -70,7 +70,16 @@ theory:
   camb:
     stop_at_error: False
     extra_args:
-      lens_potential_accuracy: 1
+      kmax: 10
+      k_per_logint: 130
+      nonlinear: True
+      lens_potential_accuracy: 8
+      lens_margin: 2050
+      AccuracyBoost: 2.0
+      lSampleBoost: 2.0
+      lAccuracyBoost: 2.0
+      DoLateRadTruncation: False
+
 
 sampler:
   mcmc:

diff --git a/mflike/MFLike.yaml b/mflike/MFLike.yaml
@@ -1,6 +1,6 @@
 # A simple cobaya likelihood for SO/LAT
 
-data_folder: MFLike/v0.7.1
+data_folder: MFLike/v0.8
 # Path to the input SACC file, containing, minimally,
 # information about the different tracers (i.e. frequency
 # bands) and the set of power spectra.
@@ -23,10 +23,10 @@ defaults:
   polarizations: [TT, TE, ET, EE]
   # Scale cuts (in ell) for each spectrum
   scales:
-    TT: [50, 5000]
-    TE: [50, 5000]
-    ET: [50, 5000]
-    EE: [50, 5000]
+    TT: [30, 9000]
+    TE: [30, 9000]
+    ET: [30, 9000]
+    EE: [30, 9000]
   # If True, TE' = (TE + ET) / 2 will be used
   # instead of TE and ET separately.
   symmetrize: false
@@ -71,7 +71,8 @@ data:
 # (if bandpass in sacc is a single freq, no band integration)
 # the default is to read bandpasses from file, to build top-hat uncomment the
 # parameters of the block!
-# Bandpass has to be in RJ units
+# Bandpass has to be divided by nu^2 if measured with respect to a RJ source
+# the nu^2 factor is set back again in the _bandpass_construction function
 top_hat_band:
 #  nsteps: 1
 #  bandwidth: 0

diff --git a/mflike/mflike.py b/mflike/mflike.py
@@ -20,7 +20,7 @@
 
 class MFLike(InstallableLikelihood):
     _url = "https://portal.nersc.gov/cfs/sobs/users/MFLike_data"
-    _release = "v0.7.1"
+    _release = "v0.8"
     install_options = {"download_url": f"{_url}/{_release}.tar.gz"}
 
     # attributes set from .yaml

diff --git a/mflike/tests/test_mflike.py b/mflike/tests/test_mflike.py
@@ -7,7 +7,7 @@
 )
 
 cosmo_params = {
-    "cosmomc_theta": 0.0104085,
+    "cosmomc_theta": 0.0104092,
     "As": 2.0989031673191437e-09,
     "ombh2": 0.02237,
     "omch2": 0.1200,
@@ -20,17 +20,17 @@
     "a_tSZ": 3.30,
     "a_kSZ": 1.60,
     "a_p": 6.90,
-    "beta_p": 2.08,
+    "beta_p": 2.20,
     "a_c": 4.90,
     "beta_c": 2.20,
     "a_s": 3.10,
     "T_d": 9.60,
-    "a_gtt": 2.81,
+    "a_gtt": 2.80,
     "a_gte": 0.10,
     "a_gee": 0.10,
     "a_psee": 0,
     "a_pste": 0,
-    "xi": 0.20,
+    "xi": 0.10,
     "bandint_shift_LAT_93": 0,
     "bandint_shift_LAT_145": 0,
     "bandint_shift_LAT_225": 0,
@@ -50,12 +50,12 @@
 }
 
 chi2s = {
-    "tt": 625.889,
-    "te-et": 925.511,
-    "ee": 722.076,
-    "tt-te-et-ee": 2265.168,
+    "tt": 920.2630056646808,
+    "te-et": 1375.0253752060808,
+    "ee": 850.9616194876838,
+    "tt-te-et-ee": 3137.874002938912,
 }
-pre = "data_sacc_"
+pre = "LAT_simu_sacc_"
 
 
 class MFLikeTest(unittest.TestCase):
@@ -68,7 +68,8 @@ def test_mflike(self):
         import camb
 
         camb_cosmo = cosmo_params.copy()
-        camb_cosmo.update({"lmax": 9000, "lens_potential_accuracy": 1})
+        #using camb low accuracy parameters for the test
+        camb_cosmo.update({"lmax": 9001, "lens_potential_accuracy": 1})
         pars = camb.set_params(**camb_cosmo)
         results = camb.get_results(pars)
         powers = results.get_cmb_power_spectra(pars, CMB_unit="muK")
@@ -80,14 +81,14 @@ def test_mflike(self):
                 {
                     "packages_path": packages_path,
                     "input_file": pre + "00000.fits",
-                    "cov_Bbl_file": pre + "w_covar_and_Bbl.fits",
+                    "cov_Bbl_file":  "data_sacc_w_covar_and_Bbl.fits",
                     "defaults": {
                         "polarizations": select.upper().split("-"),
                         "scales": {
-                            "TT": [50, 5000],
-                            "TE": [50, 5000],
-                            "ET": [50, 5000],
-                            "EE": [50, 5000],
+                            "TT": [30, 9000],
+                            "TE": [30, 9000],
+                            "ET": [30, 9000],
+                            "EE": [30, 9000],
                         },
                         "symmetrize": False,
                     },
@@ -101,8 +102,8 @@ def test_cobaya(self):
             "likelihood": {
                 "mflike.MFLike": {
                     "input_file": pre + "00000.fits",
-                    "cov_Bbl_file": pre + "w_covar_and_Bbl.fits",
-                }
+                    "cov_Bbl_file": "data_sacc_w_covar_and_Bbl.fits",
+                    },
             },
             "theory": {"camb": {"extra_args": {"lens_potential_accuracy": 1}}},
             "params": cosmo_params,
@@ -133,7 +134,7 @@ def get_model(nsteps, bandwidth):
                 "likelihood": {
                     "mflike.MFLike": {
                         "input_file": pre + "00000.fits",
-                        "cov_Bbl_file": pre + "w_covar_and_Bbl.fits",
+                        "cov_Bbl_file": "data_sacc_w_covar_and_Bbl.fits",
                         "top_hat_band": {
                             "nsteps": nsteps,
                             "bandwidth": bandwidth,
@@ -158,8 +159,8 @@ def get_model(nsteps, bandwidth):
 
         # chi2 reference results for the different models and different bandshifts
         chi2s = {
-            "model1": [2265.168, 3742.528, 43753.809],
-            "model2": [2345.508, 4458.148, 47533.477],
+            "model1": [4614.10014353, 4929.98648522, 92894.01666485],
+            "model2": [3734.34320513, 6034.08286905, 103597.03300165],
         }
 
         for model, chi2 in chi2s.items():

diff --git a/mflike/theoryforge.py b/mflike/theoryforge.py
@@ -5,9 +5,11 @@
 from cobaya.log import LoggedError
 
 
-# Converts from cmb units to brightness.
-# There is an additional nu**2 factor to convert the bandpasses
-# from RJ to brightness units.
+# Converts from cmb temperature to differential source intensity
+# (see eq. 8 of https://arxiv.org/abs/1303.5070).
+# The bandpass transmission needs to be divided by
+# nu^2 if measured with respect to a RJ source.
+# This factor is already included here.
 # Numerical factors not included,
 # it needs proper normalization when used.
 def _cmb2bb(nu):
@@ -81,7 +83,9 @@ def __init__(self, mflike=None):
 
     # Takes care of the bandpass construction. It returns a list of nu-transmittance
     # for each frequency or an array with the effective freqs.
-    # bandpasses saved in the sacc file have to be in RJ units
+    # bandpasses saved in the sacc file have to be divided by nu^2
+    # if measured with respect to a RJ source.
+    # This factor is already included in the _cmb2bb function
     def _bandpass_construction(self, **params):
         data_are_monofreq = False
         self.bandint_freqs = []