Merge branch 'main' into issue/616/scipy_simps_rename

Author: hsinfan1996

INSTALL.md

@@ -13,13 +13,13 @@
 ## Basic procedure <a name="basic_install"></a>
 
 Here we provide a quick guide for a basic instalation, this will install all the packages in your current environment.
-To create a specific conda environment for CLMM, we recommend you to check the begining of section 
+To create a specific conda environment for CLMM, we recommend you to check the begining of section
 [Access to the proper environment on cori.nersc.gov](#access_to_the_proper_environment_on_cori).
 
 ### Theory backend installation
 First, choose and install a theory backend for CLMM.
 This can be CCL (versions between 2.7.1.dev10+gf81b59a4 and 3),
-NumCosmo (v0.15 or later),
+NumCosmo (v0.19 or later),
 or cluster_toolkit and they are installable as follows.
 
 To install CCL as the theory/cosmology backend, run
@@ -60,7 +60,7 @@ Now, you can install CLMM and its dependencies as
     pip install pytest sphinx sphinx_rtd_theme
     pip install jupyter  # need to have jupyter notebook tied to this environment, you can then see the environment in jupyter.nersc.gov
     git clone https://github.com/LSSTDESC/CLMM.git  # If you'd like to contribute but don't have edit permissions to the CLMM repo, see below how to fork the repo instead.
-    cd CLMM   
+    cd CLMM
     python setup.py install     # build from source
 ```
 
@@ -72,9 +72,9 @@ If you have access to NERSC, this will likely be the easiest to make sure you ha
     module load python  # Also loads anaconda
     conda create --name clmmenv  # Create an anaconda environment for clmm
     source activate clmmenv  # switch to your newly created environment
-    conda install pip  # need pip to install everything else necessary for clmm 
+    conda install pip  # need pip to install everything else necessary for clmm
     conda install ipython # need to have the ipython tied to this environment
-    conda install -c conda-forge firefox  # Need a browser to view jupyter notebooks  
+    conda install -c conda-forge firefox  # Need a browser to view jupyter notebooks
 ```
 
 Note, for regular contributions and use, we recommend adding `module load python` to your `~/.bashrc` so you have anaconda installed every time you log in.  You will subseqeuntly also want to be in the correct environment whenever working with `clmm`, which means running `source activate clmmenv` at the start of each session.

README.md

@@ -24,7 +24,7 @@ CLMM requires Python version 3.8 or later.  CLMM has the following dependencies:
 
 - [NumPy](https://www.numpy.org/) (v1.17 or later)
 - [SciPy](https://scipy.org/) (v1.6 or later)
-- [Astropy](https://www.astropy.org/) (v4.0 or later for units and cosmology dependence)  
+- [Astropy](https://www.astropy.org/) (v4.0 or later for units and cosmology dependence)
 (Please avoid Astropy v5.0 since there is bug breaking CCL backend. It has been fixed in Astropy v5.0.1.)
 - [Matplotlib](https://matplotlib.org/) (for plotting and going through tutorials)
 
@@ -36,7 +36,7 @@ For the theoretical predictions of the signal, CLMM relies on existing libraries
 
 - [cluster-toolkit](https://cluster-toolkit.readthedocs.io/en/latest/)
 - [CCL](https://ccl.readthedocs.io/en/latest/) (versions between 2.7.1.dev10+gf81b59a4 and 3)
-- [NumCosmo](https://numcosmo.github.io/) (v0.15 or later)
+- [NumCosmo](https://numcosmo.github.io/) (v0.19 or later)
 
 
 (See the [INSTALL documentation](INSTALL.md) for more detailed installation instructions.)
@@ -91,8 +91,8 @@ non-DESC project concept and progress to be presented to the working group,
 so working group members can help co-identify tools and/or ongoing development
 that might mutually benefit your non-DESC project and ongoing DESC projects.
 
-**External Projects by Non-DESC members**: If you are not from the DESC 
-community, you are also welcome to contact CLMM Topical Team leads to introduce 
+**External Projects by Non-DESC members**: If you are not from the DESC
+community, you are also welcome to contact CLMM Topical Team leads to introduce
 your project and share feedback.
 
 

clmm/__init__.py

@@ -26,4 +26,4 @@
 )
 from . import support
 
-__version__ = "1.10.0"
+__version__ = "1.11.0"

clmm/cosmology/numcosmo.py

@@ -33,6 +33,7 @@ class NumCosmoCosmology(CLMMCosmology):
 
     def __init__(self, dist=None, dist_zmax=15.0, **kwargs):
         self.dist = None
+        self.smd = None
 
         super().__init__(**kwargs)
 
@@ -54,6 +55,7 @@ def __init__(self, dist=None, dist_zmax=15.0, **kwargs):
         self.smd = Nc.WLSurfaceMassDensity.new(self.dist)
         self.smd.prepare_if_needed(self.be_cosmo)
 
+
     def _init_from_cosmo(self, be_cosmo):
         assert isinstance(be_cosmo, Nc.HICosmo)
         assert isinstance(be_cosmo, Nc.HICosmoDECpl)
@@ -63,7 +65,7 @@ def _init_from_cosmo(self, be_cosmo):
 
     def _init_from_params(self, H0, Omega_b0, Omega_dm0, Omega_k0):
         # pylint: disable=arguments-differ
-        self.be_cosmo = Nc.HICosmo.new_from_name(Nc.HICosmo, "NcHICosmoDECpl{'massnu-length':<1>}")
+        self.be_cosmo = Nc.HICosmoDECpl(massnu_length=1)
         self.be_cosmo.omega_x2omega_k()
         self.be_cosmo.param_set_by_name("w0", -1.0)
         self.be_cosmo.param_set_by_name("w1", 0.0)
@@ -136,6 +138,10 @@ def _update_vec_funcs(self):
         self._eval_sigma_crit_core = np.vectorize(
             lambda z_len, z_src: (self.smd.sigma_critical(self.be_cosmo, z_src, z_len, z_len))
         )
+        if self.dist is not None:
+            self.dist.prepare_if_needed(self.be_cosmo)
+        if self.smd is not None:
+            self.smd.prepare_if_needed(self.be_cosmo)
 
     def _get_rho_c(self, z):
         return (

clmm/theory/numcosmo.py

@@ -193,6 +193,7 @@ def set_mset(self, mset):
 
         self.hdpm = mset.get(Nc.HaloDensityProfile.id())
         self.cosmo.smd = mset.get(Nc.WLSurfaceMassDensity.id())
+        self.cosmo.smd.prepare_if_needed(self.cosmo.be_cosmo)
 
 
 Cosmology = NumCosmoCosmology

examples/NumCosmo/Example1_Fit_Halo_Mass_to_Shear_Catalog_NC.ipynb

@@ -585,20 +585,20 @@
     "        self.props.z_src = Ncm.Vector.new_array(z_src)\n",
     "        self.props.r_source = Ncm.Vector.new_array(r_source)\n",
     "\n",
-    "        self.y.set_array(gt_profile)\n",
+    "        self.peek_mean().set_array(gt_profile)\n",
     "\n",
-    "        self.sigma.set_all(\n",
+    "        self.peek_std().set_all(\n",
     "            1.0e-2\n",
     "        )  # Diagonal covariance matrix: all points have the same standard deviation value\n",
     "\n",
     "        self.set_init(True)\n",
     "\n",
     "    # Once the NcmDataGaussDiag is initialized, its parent class variable np is set with the n_points value.\n",
     "    def do_get_length(self):\n",
-    "        return self.np\n",
+    "        return self.get_size()\n",
     "\n",
     "    def do_get_dof(self):\n",
-    "        return self.np\n",
+    "        return self.get_size()\n",
     "\n",
     "    def do_begin(self):\n",
     "        pass\n",
@@ -686,8 +686,8 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "fit1 = Ncm.Fit.new(Ncm.FitType.NLOPT, \"ln-neldermead\", lh1, mset1, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
-    "fit2 = Ncm.Fit.new(Ncm.FitType.NLOPT, \"ln-neldermead\", lh2, mset2, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
+    "fit1 = Ncm.Fit.factory(Ncm.FitType.NLOPT, \"ln-neldermead\", lh1, mset1, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
+    "fit2 = Ncm.Fit.factory(Ncm.FitType.NLOPT, \"ln-neldermead\", lh2, mset2, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
     "\n",
     "fit1.run(Ncm.FitRunMsgs.SIMPLE)\n",
     "fit1.fisher()\n",

examples/NumCosmo/Example2_Fit_Halo_Mass_to_Shear_Catalog_NC.ipynb

@@ -468,20 +468,20 @@
     "        if z_err:\n",
     "            self.props.r_source = Ncm.Vector.new_array(z_err)\n",
     "\n",
-    "        self.y.set_array(gt_profile)\n",
+    "        self.peek_mean().set_array(gt_profile)\n",
     "\n",
-    "        self.sigma.set_array(\n",
+    "        self.peek_std().set_array(\n",
     "            gt_err\n",
     "        )  # Diagonal covariance matrix: standard deviation values in gt_err.\n",
     "\n",
     "        self.set_init(True)\n",
     "\n",
     "    # Once the NcmDataGaussDiag is initialized, its parent class variable np is set with the n_points value.\n",
     "    def do_get_length(self):\n",
-    "        return self.np\n",
+    "        return self.get_size()\n",
     "\n",
     "    def do_get_dof(self):\n",
-    "        return self.np\n",
+    "        return self.get_size()\n",
     "\n",
     "    def do_begin(self):\n",
     "        pass\n",
@@ -605,9 +605,9 @@
    "metadata": {},
    "outputs": [],
    "source": [
-    "fit1 = Ncm.Fit.new(Ncm.FitType.NLOPT, \"ln-neldermead\", lh1, mset1, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
-    "fit2 = Ncm.Fit.new(Ncm.FitType.NLOPT, \"ln-neldermead\", lh2, mset2, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
-    "fit3 = Ncm.Fit.new(Ncm.FitType.NLOPT, \"ln-neldermead\", lh3, mset3, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
+    "fit1 = Ncm.Fit.factory(Ncm.FitType.NLOPT, \"ln-neldermead\", lh1, mset1, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
+    "fit2 = Ncm.Fit.factory(Ncm.FitType.NLOPT, \"ln-neldermead\", lh2, mset2, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
+    "fit3 = Ncm.Fit.factory(Ncm.FitType.NLOPT, \"ln-neldermead\", lh3, mset3, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
     "\n",
     "fit1.run(Ncm.FitRunMsgs.SIMPLE)\n",
     "fit1.fisher()\n",
@@ -768,7 +768,7 @@
     "init_sampler.set_cov_from_rescale(1.0e-1)\n",
     "\n",
     "nwalkers = 100  # Number of walkers\n",
-    "walker = Ncm.FitESMCMCWalkerAPS.new(nwalkers, mset3.fparams_len())\n",
+    "walker = Ncm.FitESMCMCWalkerAPES.new(nwalkers, mset3.fparams_len())\n",
     "\n",
     "# Ensemble Sampler MCMC\n",
     "esmcmc = Ncm.FitESMCMC.new(fit3, nwalkers, init_sampler, walker, Ncm.FitRunMsgs.SIMPLE)\n",
@@ -792,7 +792,7 @@
     "\n",
     "Here, instead of building an object directly on top of NcmDataGauss*, we use NumCosmo's framework to build non-binned likelihood for weak-lensing cluster analysis.\n",
     "\n",
-    "For that we need two objects: a NcGalaxyWLReducedShearGauss that model a Gaussian distributed reduced shear likelihood, here the observables matrix is simply $(r, \\gamma_t, \\sigma_{\\gamma_t})$ for each galaxy. If the data has spectroscopic redshifts then we use NcGalaxyRedshiftSpec with an array of real redshifts. When photometric errors are included we use the NcGalaxyRedshiftGauss object that receives $(z, \\sigma_z)$ for each galaxy. \n",
+    "For that we need two objects: a NcGalaxyWLEllipticityGauss that model a Gaussian distributed reduced shear likelihood, here the observables matrix is simply $(r, \\gamma_t, \\sigma_{\\gamma_t})$ for each galaxy. If the data has spectroscopic redshifts then we use NcGalaxyRedshiftSpec with an array of real redshifts. When photometric errors are included we use the NcGalaxyRedshiftGauss object that receives $(z, \\sigma_z)$ for each galaxy. \n",
     "\n",
     "Once we have the data objects ready we can proceed as in the previous examples.\n"
    ]
@@ -812,7 +812,7 @@
     "    sigma_g = 1.0e-4 if not sigma_g else sigma_g\n",
     "    m_obs = np.column_stack((r, g_t, np.repeat(sigma_g, len(r))))\n",
     "\n",
-    "    grsg = Nc.GalaxyWLReducedShearGauss(pos=Nc.GalaxyWLReducedShearGaussPos.R)\n",
+    "    grsg = Nc.GalaxyWLEllipticityGauss(pos=Nc.GalaxyWLEllipticityGaussPos.R)\n",
     "    grsg.set_obs(Ncm.Matrix.new_array(m_obs.flatten(), 3))\n",
     "\n",
     "    if sigma_z:\n",
@@ -837,7 +837,7 @@
     "    for data in data_array:\n",
     "        dset.append_data(data)\n",
     "    lh = Ncm.Likelihood.new(dset)\n",
-    "    fit = Ncm.Fit.new(Ncm.FitType.NLOPT, \"ln-neldermead\", lh, mset, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
+    "    fit = Ncm.Fit.factory(Ncm.FitType.NLOPT, \"ln-neldermead\", lh, mset, Ncm.FitGradType.NUMDIFF_FORWARD)\n",
     "    # fit.set_params_reltol (1.0e-8)\n",
     "    # fit.set_m2lnL_reltol (1.0e-11)\n",
     "\n",
@@ -948,7 +948,7 @@
     "init_sampler.set_cov_from_rescale(1.0e-1)\n",
     "\n",
     "nwalkers = 100\n",
-    "stretch = Ncm.FitESMCMCWalkerAPS.new(nwalkers, mset3.fparams_len())\n",
+    "stretch = Ncm.FitESMCMCWalkerAPES.new(nwalkers, mset3.fparams_len())\n",
     "\n",
     "esmcmc = Ncm.FitESMCMC.new(fit3, nwalkers, init_sampler, stretch, Ncm.FitRunMsgs.SIMPLE)\n",
     "esmcmc.set_data_file(\"example2_fit3_esmcmc_out_aps.fits\")\n",
@@ -1026,7 +1026,7 @@
    "outputs": [],
    "source": [
     "ser = Ncm.Serialize.new(0)\n",
-    "data = fit3.lh.dset.get_data(0)\n",
+    "data = fit3.peek_likelihood().peek_dataset().get_data(0)\n",
     "ser.to_file(data, \"example2_fit3_data.obj\")"
    ]
   }