Quick Fisher for TT TE EE BB.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Making some quick calculations to get $F_{ij}$ working for $C_l^{TT},C_l^{TE},C_l^{EE}$ observables"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "IMPORT ERROR: /opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/quicklens/mpi.pyc (No module named pypar). Could not load pbs or pypar. MPI will not be used.\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'\\nThis notebook originally a copy of the TT TE EE version, April 13, 2018.\\n\\n'"
      ]
     },
     "execution_count": 1,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import crosspower as cp # branch quickCl\n",
    "import FisherCl as fcl\n",
    "import noiseCl as ncl\n",
    "import camb\n",
    "\n",
    "# set plotting to be in the notebook instead of its own qt window\n",
    "%matplotlib inline\n",
    "\n",
    "\"\"\"\n",
    "This notebook originally a copy of the TT TE EE version, April 13, 2018.\n",
    "\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "\"\\ncosParams = {\\n        'H0'    : None, #67.51, #setting H0=None allows cosmomc_theta to be used instead\\n        'cosmomc_theta'           : 1.04087e-2,\\n        'ombh2' : 0.02226,\\n        'omch2' : 0.1193,\\n        'omk'   : 0,\\n        'tau'   : 0.063,\\n\\n        'As'    : 2.130e-9,\\n        'ns'    : 0.9653,\\n        'r'     : 0,\\n        'kPivot': 0.05,\\n    \\n        'w'     : -1.0, # actually is w0 but CAMB calls it w\\n        'wa'    : 0.0,\\n\\n        # if fiducial mnu is changed, need to adjust omch2 as well\\n        'mnu'   : 0.06, # (eV)\\n        #'mnu'   : 0.058, # Lloyd suggested this value for fiducial; adjust omch2 if I do use it\\n        'nnu'   : 3.046,\\n        'standard_neutrino_neff'  : 3.046,\\n        'num_massive_neutrinos'   : 1,\\n        'neutrino_hierarchy'      : 'normal'}\\n\\nAccuracyBoost = 3\\nnonlinear = True\\nnz = 30000\\nprint 'nonlinear = ',nonlinear\\n\""
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# set cosmological parameters\n",
    "\n",
    "# maybe load a FisherMatrix object and get these from that instead?\n",
    "\"\"\"\n",
    "cosParams = {\n",
    "        'H0'    : None, #67.51, #setting H0=None allows cosmomc_theta to be used instead\n",
    "        'cosmomc_theta'           : 1.04087e-2,\n",
    "        'ombh2' : 0.02226,\n",
    "        'omch2' : 0.1193,\n",
    "        'omk'   : 0,\n",
    "        'tau'   : 0.063,\n",
    "\n",
    "        'As'    : 2.130e-9,\n",
    "        'ns'    : 0.9653,\n",
    "        'r'     : 0,\n",
    "        'kPivot': 0.05,\n",
    "    \n",
    "        'w'     : -1.0, # actually is w0 but CAMB calls it w\n",
    "        'wa'    : 0.0,\n",
    "\n",
    "        # if fiducial mnu is changed, need to adjust omch2 as well\n",
    "        'mnu'   : 0.06, # (eV)\n",
    "        #'mnu'   : 0.058, # Lloyd suggested this value for fiducial; adjust omch2 if I do use it\n",
    "        'nnu'   : 3.046,\n",
    "        'standard_neutrino_neff'  : 3.046,\n",
    "        'num_massive_neutrinos'   : 1,\n",
    "        'neutrino_hierarchy'      : 'normal'}\n",
    "\n",
    "AccuracyBoost = 3\n",
    "nonlinear = True\n",
    "nz = 30000\n",
    "print 'nonlinear = ',nonlinear\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Load a saved FisherMatrix object for parameters and structure"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "obsNames:  ['k,k', 'k,g1', 'k,g2', 'k,g3', 'k,g4', 'k,g5', 'k,g6', 'g1,g1', 'g1,g2', 'g1,g3', 'g1,g4', 'g1,g5', 'g1,g6', 'g2,g2', 'g2,g3', 'g2,g4', 'g2,g5', 'g2,g6', 'g3,g3', 'g3,g4', 'g3,g5', 'g3,g6', 'g4,g4', 'g4,g5', 'g4,g6', 'g5,g5', 'g5,g6', 'g6,g6']\n",
      "paramList:  ['ombh2', 'omch2', 'cosmomc_theta', 'As', 'ns', 'tau', 'mnu', 'w', 'wa', 'bin1', 'bin2', 'bin3', 'bin4', 'bin5', 'bin6']\n",
      "nonlinear:  True\n",
      "nz:  300000\n"
     ]
    }
   ],
   "source": [
    "# load saved Fobj\n",
    "# get lensing and galaxy Fisher matrix to add\n",
    "\n",
    "#\"\"\"\n",
    "import pickle\n",
    "\n",
    "filehandler = open('Fisher_7_6A.obj','r') # AccuracyBoost=3, lmax=3000, lmaxP=5000, nz=3e5; nonlinear=True,\n",
    "#neutrino_hierarchy = 'normal', usePrimaryCMB = True; from FisherCl branch quickCl\n",
    "# kludge for this one: told cp.getDNDZinterp to use modelDNDZ3; set cp.tophat beesBins=True\n",
    "# biasByBin = False (uses sigma(b)/b), dw = 0.05, useNoise = True\n",
    "\n",
    "\n",
    "# these two have an additional parameter wa but not crossClsPlus,Minus for it, etc.\n",
    "#filehandler = open('Fisher_7_6X2_w2_wan.obj','r') # AccuracyBoost=3, lmax=3000, lmaxP=5000, nz=3e5; nonlinear=True,\n",
    "#filehandler = open('Fisher_7_6X4_w2_wan.obj','r') # AccuracyBoost=3, lmax=3000, lmaxP=5000, nz=3e5; nonlinear=False,\n",
    "\n",
    "Fobj = pickle.load(filehandler) \n",
    "\n",
    "lmin = Fobj.lmin\n",
    "#lmin = 2\n",
    "lmax = Fobj.lmax\n",
    "#lmax = 3000\n",
    "\n",
    "nMaps = Fobj.nBins + 1\n",
    "nCosParams = Fobj.nParams - Fobj.nBins\n",
    "ells = Fobj.ells\n",
    "obsNames = Fobj.obsNames\n",
    "paramList = Fobj.paramList\n",
    "\n",
    "nz = Fobj.nz\n",
    "nonlinear = Fobj.nonlinear\n",
    "AccuracyBoost = Fobj.AccuracyBoost\n",
    "myParams = Fobj.cosParams\n",
    "\n",
    "\n",
    "print 'obsNames: ',obsNames\n",
    "print 'paramList: ',paramList\n",
    "print 'nonlinear: ',nonlinear\n",
    "print 'nz: ',nz\n",
    "#\"\"\"\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "neutrino_hierarchy =  normal\n",
      "starting makePkInterp.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "finishing makePkInterp.\n",
      "done.\n"
     ]
    }
   ],
   "source": [
    "# get a matterPower object\n",
    "#reload(cp)\n",
    "myPk = cp.MatterPower(nz=nz,AccuracyBoost=AccuracyBoost,nonlinear=nonlinear,**myParams)\n",
    "print 'done.'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "# verify that unlensed spectra have zero B component\n",
    "#doBcheck = True\n",
    "doBcheck = False\n",
    "if doBcheck:\n",
    "    # get pars\n",
    "    pars = myPk.getPars(lmax=lmax+1,lpa=5.0,AccuracyBoost=AccuracyBoost,**myParams)\n",
    "    #get the selected power spectra\n",
    "    results = camb.get_results(pars)\n",
    "    powers = results.get_cmb_power_spectra(pars)\n",
    "    totCL=powers['total']\n",
    "    unlensedCL=powers['unlensed_scalar']\n",
    "    print(totCL.shape,unlensedCL.shape)\n",
    "    \n",
    "    \n",
    "    # plot the B components\n",
    "    #Python CL arrays are all zero based (starting at L=0), Note L=0,1 entries will be zero by default.\n",
    "    #The different CL are always in the order TT, EE, BB, TE (with BB=0 for unlensed scalar results).    \n",
    "    ls = np.arange(totCL.shape[0])\n",
    "    fig, ax = plt.subplots(4,2, figsize = (24,12))\n",
    "    \n",
    "    ax[0,0].plot(ls,totCL[:,0], color='k')\n",
    "    ax[0,0].plot(ls,unlensedCL[:,0], color='r')\n",
    "    ax[0,0].set_title(r'$TT$')\n",
    "    ax[0,1].plot(ls[2:], 1-unlensedCL[2:,0]/totCL[2:,0])\n",
    "    ax[0,1].set_title(r'$\\Delta TT$')\n",
    "    \n",
    "    ax[1,0].plot(ls,totCL[:,1], color='k')\n",
    "    ax[1,0].plot(ls,unlensedCL[:,1], color='r')\n",
    "    ax[1,0].set_title(r'$EE$')\n",
    "    ax[1,1].plot(ls[2:], 1-unlensedCL[2:,1]/totCL[2:,1])\n",
    "    ax[1,1].set_title(r'$\\Delta EE$')\n",
    "    \n",
    "    ax[2,0].plot(ls,totCL[:,0], color='k')\n",
    "    ax[2,0].plot(ls,unlensedCL[:,3], color='r')\n",
    "    ax[2,0].set_title(r'$TE$')\n",
    "    ax[2,1].plot(ls[2:], 1-unlensedCL[2:,3]/totCL[2:,3])\n",
    "    ax[2,1].set_title(r'$\\Delta TE$')\n",
    "    \n",
    "    ax[3,0].plot(ls,totCL[:,1], color='k')\n",
    "    ax[3,0].plot(ls,unlensedCL[:,2], color='r')\n",
    "    ax[3,0].set_title(r'$BB$')\n",
    "    ax[3,1].plot(ls[2:], 1-unlensedCL[2:,2]/totCL[2:,2])\n",
    "    ax[3,1].set_title(r'$\\Delta BB$')\n",
    "    \n",
    "    \n",
    "    for ax in ax.reshape(-1): ax.set_xlim([2,2500])\n",
    "    plt.show()\n",
    "\n",
    "    "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# get sigmas for comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "ombh2 :  8.80988471849915e-06\n",
      "omch2 :  0.0003959759210277408\n",
      "cosmomc_theta :  8.680089639141158e-06\n",
      "As :  5.923696448050441e-10\n",
      "ns :  0.0014251121121470452\n",
      "tau :  0.12582349865118583\n",
      "mnu :  0.49521756886450985\n",
      "w :  52.29416127562633\n",
      "wa :  134.59611994264677\n"
     ]
    }
   ],
   "source": [
    "doComparison = True\n",
    "myLmin = 30\n",
    "getSigmas = lambda Fij: np.sqrt(np.diag(np.linalg.inv(Fij)))\n",
    "if doComparison:\n",
    "    Fij_TEB_original = Fobj.makeFisher(myLmin,TE=True)\n",
    "    sigmas_TEB_original = getSigmas(Fij_TEB_original)\n",
    "    for i, paramName in enumerate(paramList[:nCosParams]):\n",
    "        print paramName, ': ',sigmas_TEB_original[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Get all cross power spectra"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "# function for checking the settings as they are being used\n",
    "\n",
    "def checkPars(pars):\n",
    "    H0 = pars.H0\n",
    "    print 'ombh2: ',pars.omegab*(H0/100)**2\n",
    "    print 'omch2: ',pars.omegac*(H0/100)**2\n",
    "    print 'H_0: ',H0, '(proxy for theta?)'\n",
    "    #print 'initial condition vector: ',pars.InitialConditionVector\n",
    "    print 'As: ',pars.InitPower.ScalarPowerAmp[0]\n",
    "    print 'ns: ',pars.InitPower.an[0]\n",
    "    print 'tau: ',pars.Reion.optical_depth\n",
    "    #print 'omega_n: ',pars.omegan\n",
    "    print 'omega_n*94*h**2: ',pars.omegan*94*(H0/100)**2\n",
    "    # need to find w and wa\n",
    "    \n",
    "    #print '\\n'\n",
    "    #print pars\n",
    "    \n",
    "#checkPars(parsTest)\n",
    "if False:\n",
    "    for cParamNum in range(nCosParams):\n",
    "        print '\\nContents of pars objects for parameter set ',cParamNum+1,' for parameter ',paramList[cParamNum], \\\n",
    "              ' with delta ',deltaP[cParamNum]\n",
    "        checkPars(myParsUpper[cParamNum])\n",
    "        checkPars(myParsLower[cParamNum])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "creating modified parameter lists . . . \n",
      "done.\n"
     ]
    }
   ],
   "source": [
    "# copy some setup from FisherCl.py\n",
    "\n",
    "# parameters list:\n",
    "# step sizes for discrete derivatives: must correspond to paramList entries!\n",
    "#   from Allison et. al. (2015) Table III.\n",
    "nCosParams = 9 # 6 LCDM + Mnu + w0 + wa\n",
    "paramList = ['ombh2','omch2','cosmomc_theta',  'As', 'ns','tau','mnu', 'w', 'wa']\n",
    "deltaP =    [ 0.0008, 0.0030,      0.0050e-2,0.1e-9,0.010,0.020,0.020,0.05,0.025] #mnu one in eV\n",
    "\n",
    "# for testing\n",
    "#nCosParams = 2\n",
    "#paramList = [ 'w', 'wa']\n",
    "#deltaP =    [0.05,0.025]\n",
    "\n",
    "# cut deltaP in half to match Byeonghee's stepsizes\n",
    "deltaP = np.array(deltaP)/2.\n",
    "\n",
    "# double deltaP to see effect\n",
    "#deltaP = np.array(deltaP)*2.\n",
    "\n",
    "\"\"\"\n",
    "# get more parameter sets for numeric differentiation\n",
    "print 'creating parameter sets for making discrete derivatives...'\n",
    "myParamsUpper = []\n",
    "myParamsLower = []\n",
    "#myParsUpper = []\n",
    "#myParsLower = []\n",
    "for cParamNum in range(nCosParams):\n",
    "    #print 'creating matter power spectra for ',paramList[cParamNum],' derivative...'\n",
    "    print 'creating parameter set for ',paramList[cParamNum],' derivative...'\n",
    "    # add parameter dictionary to lists; HAVE TO BE COPIES!!!\n",
    "    myParamsUpper.append(myParams.copy())\n",
    "    myParamsLower.append(myParams.copy())\n",
    "    # modify parameter number cParamNum in dictionaries\n",
    "    myParamsUpper[cParamNum][paramList[cParamNum]] += deltaP[cParamNum]\n",
    "    myParamsLower[cParamNum][paramList[cParamNum]] -= deltaP[cParamNum]\n",
    "\n",
    "    #print 'cPramNum: ',cParamNum,', param name: ',paramList[cParamNum]\n",
    "    #print 'myParamsUpper[cParamNum][paramList[cParamNum]]: ',myParamsUpper[cParamNum][paramList[cParamNum]]\n",
    "    #print 'myParamsLower[cParamNum][paramList[cParamNum]]: ',myParamsLower[cParamNum][paramList[cParamNum]]\n",
    "    #print 'deltaP[cParamNum]: ',deltaP[cParamNum]\n",
    "\n",
    "    # create pars objects and add to lists. \n",
    "    #myParsUpper.append(cp.MatterPower.getPars(myPk,**myParamsUpper[cParamNum]))\n",
    "    #myParsLower.append(cp.MatterPower.getPars(myPk,**myParamsLower[cParamNum]))\n",
    "\n",
    "    ###### DOING THESE TWO GETPARS CALLS RIGHT AFTER EACH OTHER DESTROYS THE FIRST ONE'S DARK ENERGY SETTING! #####\n",
    "    ###### NEED TO RE-ORGANIZE SO NEED FOR MYPARSUPPER IS COMPLETELY DONE BEFORE CREATING MYPARSLOWER   #####\n",
    "\"\"\"    \n",
    "    \n",
    "# get modified parameter lists for numeric differentiation\n",
    "print 'creating modified parameter lists . . . '\n",
    "#myParams = self.cosParams\n",
    "myParamsUpper = []\n",
    "myParamsLower = []\n",
    "for cParamNum in range(nCosParams):\n",
    "    # add parameter dictionary to lists; HAVE TO BE COPIES!!!\n",
    "    myParamsUpper.append(myParams.copy())\n",
    "    myParamsLower.append(myParams.copy())\n",
    "    # modify parameter number cParamNum in ditionaries\n",
    "    myParamsUpper[cParamNum][paramList[cParamNum]] += deltaP[cParamNum]\n",
    "    myParamsLower[cParamNum][paramList[cParamNum]] -= deltaP[cParamNum]\n",
    "    \n",
    "print 'done.'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define function for getting cross power spectra for upper or lower \n",
    "\"\"\"\n",
    "# old version - delete this\n",
    "#def getCrossCls(paramList,myParams,myPk,nMaps=3,lmax=5000,lpa=5.0,\n",
    "#                AccuracyBoost=3,myClName='unlensed_scalar'):\n",
    "    #\"\n",
    "    Purpose:\n",
    "    \n",
    "    Inputs:\n",
    "        paramList: like FisherMatrix.paramList\n",
    "            Note: a list of length 1 will cause a differently shaped array to be returned\n",
    "        myParams: a list of lists of parameters like FisherMatrix.cosParams\n",
    "            This list must have the same length as paramList\n",
    "        myPk: a MatterPower object\n",
    "        \n",
    "        nMaps: number of maps.  Set to 3 for T,E,B\n",
    "        lmax: lmax for returned array\n",
    "        lpa: lensing_power_accuracy setting\n",
    "        AccuracyBoost: accuracy setting\n",
    "        myClName: which type of power spectrum to get from CAMB\n",
    "    Returns:\n",
    "        crossCls: numpy array of shape (nMaps,nMaps,nCosParams,lmax)\n",
    "          unless the lenth of paramList is 1, then shape will be (nMaps,nMaps,lmax)\n",
    "          Note Cl arrays are all zero based (starting at L=0).\n",
    "          Note L=0,1 entries will be zero by default.\n",
    "    #\"\n",
    "\n",
    "    nCosParams = paramList.__len__()\n",
    "    \n",
    "    crossCls = np.zeros((nMaps,nMaps,nCosParams,lmax+1))\n",
    "    for paramNum in range(nCosParams):\n",
    "        print 'getting Cl power spectra for perturbed parameter ',paramList[paramNum]\n",
    "        # get pars\n",
    "        pars = cp.MatterPower.getPars(myPk,lmax=lmax+1,lpa=lpa,AccuracyBoost=AccuracyBoost,\n",
    "                                      **myParams[paramNum])\n",
    "        #get the selected power spectra\n",
    "        results = camb.get_results(pars)\n",
    "        powers = results.get_cmb_power_spectra(pars)\n",
    "        myCl = powers[myClName]\n",
    "        \n",
    "        #store them\n",
    "        #The different CL are always in the order TT, EE, BB, TE (with BB=0 for unlensed scalar results).\n",
    "        crossCls[0,0,paramNum]  = myCl[:lmax+1,0] # TT\n",
    "        crossCls[0,1,paramNum]  = myCl[:lmax+1,3] # TE\n",
    "        crossCls[1,0,paramNum]  = myCl[:lmax+1,3] # ET\n",
    "        crossCls[1,1,paramNum]  = myCl[:lmax+1,1] # EE\n",
    "        if nMaps == 3:\n",
    "            crossCls[2,2,paramNum]  = myCl[:lmax+1,2] # BB\n",
    "        \n",
    "        # reshape for unperterbed version\n",
    "        if nCosParams == 1:\n",
    "            crossCls = np.reshape(crossCls,(nMaps,nMaps,lmax+1))\n",
    "    return crossCls\n",
    "\"\"\"    \n",
    "    \n",
    "def getCrossClsP(paramList,myParams,myPk, nMaps=3,lmax=5000,lpa=5.0,\n",
    "                   AccuracyBoost=3,myClName='unlensed_scalar'):\n",
    "    \"\"\"\n",
    "    Purpose:\n",
    "        get the crossCls for primary CMB: TT,TE,EE,BB\n",
    "    Inputs:\n",
    "        paramList: like FisherMatrix.paramList, but only contains parameter\n",
    "            names for those which are to be varied\n",
    "            Note: a list of length 1 will cause a differently shaped array \n",
    "            to be returned\n",
    "        myParams: a list of lists of parameters like FisherMatrix.cosParams\n",
    "            This list must have the same length as paramList\n",
    "        myPk: a MatterPower object\n",
    "        \n",
    "        nMaps: number of maps.  Set to 3 for T,E,B; 2 for T,E\n",
    "        lmax: lmax for returned array.  lmin will be zero\n",
    "        lpa: lensing_power_accuracy setting\n",
    "        AccuracyBoost: accuracy setting\n",
    "        myClName: which type of power spectrum to get from CAMB\n",
    "    Returns:\n",
    "        crossCls: numpy array of shape (nMaps,nMaps,nCosParams,lmax)\n",
    "          unless the lenth of paramList is 1, then shape will be \n",
    "            (nMaps,nMaps,lmax)\n",
    "          Note Cl arrays are all zero based (starting at L=0).\n",
    "          Note L=0,1 entries will be zero by default.\n",
    "          Units are microK^2\n",
    "    \"\"\"\n",
    "    nCosParams = paramList.__len__()\n",
    "    \n",
    "    crossCls = np.zeros((nMaps,nMaps,nCosParams,lmax+1))\n",
    "    for paramNum in range(nCosParams):\n",
    "        print 'getting Cl power spectra for perturbed parameter ', \\\n",
    "              paramList[paramNum]\n",
    "        # get pars\n",
    "        pars = cp.MatterPower.getPars(myPk,lmax=lmax+1,lpa=lpa,\n",
    "                                      AccuracyBoost=AccuracyBoost,\n",
    "                                      **myParams[paramNum])\n",
    "        #get the selected power spectra\n",
    "        results = camb.get_results(pars)\n",
    "        powers = results.get_cmb_power_spectra(pars, CMB_unit='muK')\n",
    "        myCl = powers[myClName]  #ummm... doing it this way probably creates all 6 spectra, then I throw away 5 of them.\n",
    "        \n",
    "        #store them\n",
    "        #The different CL are always in the order TT, EE, BB, TE \n",
    "        #  (with BB=0 for unlensed scalar results).\n",
    "        crossCls[0,0,paramNum]  = myCl[:lmax+1,0] # TT\n",
    "        crossCls[0,1,paramNum]  = myCl[:lmax+1,3] # TE\n",
    "        crossCls[1,0,paramNum]  = myCl[:lmax+1,3] # ET\n",
    "        crossCls[1,1,paramNum]  = myCl[:lmax+1,1] # EE\n",
    "        if nMaps == 3:\n",
    "            crossCls[2,2,paramNum]  = myCl[:lmax+1,2] # BB\n",
    "        \n",
    "        # reshape for unperterbed version\n",
    "        if nCosParams == 1:\n",
    "            crossCls = np.reshape(crossCls,(nMaps,nMaps,lmax+1))\n",
    "    return crossCls\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n",
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  (All Fiducial)\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "starting upper spectra for numeric derivatives . . . \n",
      "getting Cl power spectra for perturbed parameter  ombh2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  omch2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  cosmomc_theta\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  As\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  ns\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  tau\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  mnu\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Currently dark energy parameters are changed globally, not per parameter set\n",
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  w\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  wa\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "starting lower spectra for numeric derivatives . . . \n",
      "getting Cl power spectra for perturbed parameter  ombh2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  omch2\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  cosmomc_theta\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  As\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  ns\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  tau\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  mnu\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:Currently dark energy parameters are changed globally, not per parameter set\n",
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  w\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting Cl power spectra for perturbed parameter  wa\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "WARNING:root:accuracy parameters are changed globally, not yet per parameter set\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done\n"
     ]
    }
   ],
   "source": [
    "# get all cross power spectra\n",
    "# this cell modified to add BB and fix DE problem\n",
    "\n",
    "lmax = 5000 # may want to match lmax in other Fisher calculation\n",
    "lpa = 5.0\n",
    "pars = myPk.getPars(lmax=lmax,lpa=lpa,AccuracyBoost=AccuracyBoost)\n",
    "\n",
    "#get the total lensed CMB power spectra versus unlensed\n",
    "#myClName = 'total'\n",
    "myClName = 'unlensed_scalar'\n",
    "\n",
    "# select 2 for T,E, 3 for T,E,B\n",
    "nMaps = 2 # T,E\n",
    "#nMaps = 3 # T,E,B\n",
    "nCls = nMaps*(nMaps+1)/2\n",
    "\n",
    "#crossCls      = np.zeros((nMaps,nMaps,           lmax))\n",
    "#crossClsPlus  = np.zeros((nMaps,nMaps,nCosParams,lmax))\n",
    "#crossClsMinus = np.zeros((nMaps,nMaps,nCosParams,lmax))\n",
    "\n",
    "# get main set of power spectra\n",
    "crossCls = getCrossClsP(['(All Fiducial)'],[myParams],myPk,nMaps=nMaps,lmax=lmax,lpa=lpa,\n",
    "                            AccuracyBoost=AccuracyBoost,myClName=myClName)\n",
    "# get the perturbed versions\n",
    "print 'starting upper spectra for numeric derivatives . . . '\n",
    "crossClsPlus  = getCrossClsP(paramList,myParamsUpper,myPk,nMaps=nMaps,lmax=lmax,lpa=lpa,\n",
    "                            AccuracyBoost=AccuracyBoost,myClName=myClName)\n",
    "print 'starting lower spectra for numeric derivatives . . . '\n",
    "crossClsMinus = getCrossClsP(paramList,myParamsLower,myPk,nMaps=nMaps,lmax=lmax,lpa=lpa,\n",
    "                            AccuracyBoost=AccuracyBoost,myClName=myClName)\n",
    "\n",
    "# reset global (dark energy and AccuracyBoost) settings\n",
    "pars = myPk.getPars(lmax=lmax,lpa=lpa,AccuracyBoost=AccuracyBoost)\n",
    "\n",
    "ellsP = range(lmax+1)\n",
    "\n",
    "print 'done'\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(2, 2, 5001)\n",
      "(2, 2, 9, 5001)\n",
      "(2, 2, 9, 5001)\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYAAAAEICAYAAABWJCMKAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nc1QAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEICAYAAAC9E5gJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYKCUUgr4/wGk6Rf8TGAtfgAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAYMAAAEICAYAAAC9E5gJAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nsEPwZAAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXQAAAEICAYAAABPgw/pAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nD+BedcYTiRpFAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# plot a few\n",
    "print crossCls.shape\n",
    "print crossClsPlus.shape\n",
    "print crossClsMinus.shape\n",
    "for mapNum in range(nMaps):\n",
    "    for mapNum2 in range(nMaps):\n",
    "        plt.plot(ellsP,crossCls[mapNum,mapNum2])\n",
    "        plt.title('crossCls[{0},{1}]'.format(mapNum,mapNum2))\n",
    "        plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Calculate numerical derivatives"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "starting creation of C_l derivatives... \n",
      "starting derivative set  1  of  2 ... \n",
      "starting derivative set  2  of  2 ... \n",
      "done.\n"
     ]
    }
   ],
   "source": [
    "# get derivatives wrt parameters\n",
    "print 'starting creation of C_l derivatives... '\n",
    "\n",
    "dClVecs = np.empty((nCls, nCosParams, lmax+1))\n",
    "for map1 in range(nMaps):\n",
    "    print 'starting derivative set ',map1+1,' of ',nMaps,'... '\n",
    "    for map2 in range(map1,nMaps):\n",
    "        mapIdx  = map1*nMaps+map2 -map1*(map1+1)/2  \n",
    "                                   # mapIdx = map index\n",
    "        for pIdx in range(nCosParams):\n",
    "            dClPlus  = crossClsPlus[map1,map2,pIdx]\n",
    "            dClMinus = crossClsMinus[map1,map2,pIdx]\n",
    "            dClVecs[mapIdx, pIdx] = (dClPlus-dClMinus)/(2*deltaP[pIdx])\n",
    "\n",
    "print 'done.'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Add Noise"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "getting (primary CMB) detector noise...\n",
      "(2, 2, 5001)\n"
     ]
    }
   ],
   "source": [
    "print 'getting (primary CMB) detector noise...'\n",
    "import noiseCl as ncl\n",
    "# CMBS4 v1\n",
    "fwhm = 1; ST = 1; SP = ST*1.414\n",
    "\n",
    "def getNoiseClsP(fwhm,ST,SP,lmax,nMaps):\n",
    "    ellsP = range(lmax+1)\n",
    "    noiseCMBS4_TT1 = ncl.noisePower(ST,ST,fwhm,ellsP)\n",
    "    #noiseCMBS4_TP1 = ncl.noisePower(ST,SP,fwhm,ellsP)\n",
    "    noiseCMBS4_PP1 = ncl.noisePower(SP,SP,fwhm,ellsP)\n",
    "\n",
    "    # set TE noise to zero\n",
    "    zerosList = np.zeros(lmax+1)\n",
    "    noiseCMBS4_TP1 = zerosList\n",
    "\n",
    "    # shape like crossCls\n",
    "    if nMaps == 2:\n",
    "        noiseClsP = np.array([[noiseCMBS4_TT1,noiseCMBS4_TP1],[noiseCMBS4_TP1,noiseCMBS4_PP1]])\n",
    "    elif nMaps == 3:\n",
    "        noiseClsP = np.array([[noiseCMBS4_TT1, noiseCMBS4_TP1,      zerosList],\n",
    "                              [noiseCMBS4_TP1, noiseCMBS4_PP1,      zerosList],\n",
    "                              [     zerosList,      zerosList, noiseCMBS4_PP1]])\n",
    "    else:\n",
    "        'bad nMaps for noise maker'\n",
    "    return noiseClsP\n",
    "\n",
    "noiseClsP = getNoiseClsP(fwhm,ST,SP,lmax,nMaps)\n",
    "print noiseClsP.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Create Covariance Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "building covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "(3, 3, 5001)\n",
      "lmin =  0\n",
      "obsList:  [0. 1. 3.]\n"
     ]
    }
   ],
   "source": [
    "# make covariance matrices with noise\n",
    "print 'building covariance matrix... '\n",
    "covarP,invCovP,ellsP,obsListP = Fobj.makeCovar(crossCls,noiseClsP,0,lmax)\n",
    "\n",
    "print invCovP.shape\n",
    "print 'lmin = ',ellsP[0]\n",
    "print 'obsList: ',obsListP"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "# save covar\n",
    "# ell values go from 0 to lmax-1\n",
    "#import pickle \n",
    "#pickle.dump(covarP,open('covar.pkl','w'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Build the Fisher matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "Fij:\n",
      "[[ 1.54612418e+11  1.53857568e+08 -1.58274073e+12  7.69416058e+17\n",
      "   2.95888770e+09 -3.27391232e+09 -6.56868793e+06 -9.20514938e+05\n",
      "  -1.30546587e+06]\n",
      " [ 1.53857568e+08  3.06869876e+07  1.13177130e+10 -2.52517649e+15\n",
      "  -4.90817679e+05  1.07525175e+07  7.13834641e+04  2.84729582e+03\n",
      "   3.76702309e+03]\n",
      " [-1.58274073e+12  1.13177130e+10  3.81714811e+13 -8.93106668e+18\n",
      "  -3.35356443e+10  3.80029572e+10  1.97731220e+08  1.02794511e+07\n",
      "   1.37798899e+07]\n",
      " [ 7.69416058e+17 -2.52517649e+15 -8.93106668e+18  5.45869465e+24\n",
      "   1.80320977e+16 -2.32282219e+16 -3.70141418e+13 -6.56991055e+12\n",
      "  -9.40435098e+12]\n",
      " [ 2.95888770e+09 -4.90817679e+05 -3.35356443e+10  1.80320977e+16\n",
      "   6.51503757e+07 -7.67332747e+07 -1.39269264e+05 -2.16943797e+04\n",
      "  -3.08430253e+04]\n",
      " [-3.27391232e+09  1.07525175e+07  3.80029572e+10 -2.32282219e+16\n",
      "  -7.67332747e+07  9.88426272e+07  1.57502909e+05  2.79610448e+04\n",
      "   4.00200270e+04]\n",
      " [-6.56868793e+06  7.13834641e+04  1.97731220e+08 -3.70141418e+13\n",
      "  -1.39269264e+05  1.57502909e+05  1.06502859e+03  4.17973697e+01\n",
      "   5.44217145e+01]\n",
      " [-9.20514938e+05  2.84729582e+03  1.02794511e+07 -6.56991055e+12\n",
      "  -2.16943797e+04  2.79610448e+04  4.17973697e+01  8.07298291e+00\n",
      "   1.14434480e+01]\n",
      " [-1.30546587e+06  3.76702309e+03  1.37798899e+07 -9.40435098e+12\n",
      "  -3.08430253e+04  4.00200270e+04  5.44217145e+01  1.14434480e+01\n",
      "   1.63954858e+01]]\n",
      "eigenvalues of F_ij^-1:\n",
      "[-3.43672630e-16  4.02879647e-14  2.39780303e-11  5.19016926e-08\n",
      "  1.07532156e-06  4.18133856e-03  1.55688525e-01  7.24744759e+00\n",
      "  6.35236729e+01]\n"
     ]
    }
   ],
   "source": [
    "#Build Fisher matrix\n",
    "#myLmin = 50 #to match Allison et. al. and justify no ISW effect\n",
    "myLmin = 30 # match Byeonghee\n",
    "\n",
    "def getFisher(myLmin,invCovP,dClVecs,nCosParams,lmax):\n",
    "    #multply vector.T,invcov,vector and add up\n",
    "    print 'building Fisher matrix from components...'\n",
    "    print 'invCovP.shape: ',invCovP.shape,', dClVecs.shape: ',dClVecs.shape\n",
    "    Fij = np.zeros((nCosParams,nCosParams)) # indices match those in paramList\n",
    "    for i in range(nCosParams):\n",
    "        print 'starting bin set ',i+1,' of ',nCosParams\n",
    "        dClVec_i = dClVecs[:,i,:] # shape (nCls,nElls)\n",
    "        for j in range(nCosParams):\n",
    "            dClVec_j = dClVecs[:,j,:] # shape (nCls,nElls)\n",
    "            # ugh.  don't like nested loops in Python... but easier to program...\n",
    "            #for ell in range(myLmin-2,lmax+51-2):\n",
    "            for ell in range(myLmin-2,lmax-2):\n",
    "              myCov = invCovP[:,:,ell]  #note invCov has no ell=0,1 but dClVecs do\n",
    "              fij = np.dot(dClVec_i[:,ell+2],np.dot(myCov,dClVec_j[:,ell+2]))\n",
    "              Fij[i,j] += fij\n",
    "\n",
    "    print 'creation of Fisher Matrix complete!\\n'\n",
    "    return Fij\n",
    "\n",
    "Fij = getFisher(myLmin,invCovP,dClVecs,nCosParams,lmax)\n",
    "\n",
    "# Have a look at it\n",
    "print 'Fij:'\n",
    "print Fij\n",
    "# check eigenvalues\n",
    "w,v = np.linalg.eigh(np.linalg.inv(Fij))\n",
    "print 'eigenvalues of F_ij^-1:'\n",
    "print w"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Save the results in an augmented Fisher Matrix object file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Save these as part of Fobj.\n",
    "# Is Fobj loaded?\n",
    "Fobj.FijTE = Fij\n",
    "Fobj.dClVecsP = dClVecs[:,:,lmin:]\n",
    "Fobj.covarP = covarP[:,:,lmin:]\n",
    "Fobj.invCovP = invCovP[:,:,lmin:]\n",
    "\n",
    "Fobj.crossClsP      = crossCls[:,:,lmin:]\n",
    "Fobj.crossClsPPlus  = crossClsPlus[:,:,:,lmin:]\n",
    "Fobj.crossClsPMinus = crossClsMinus[:,:,:,lmin:]\n",
    "\n",
    "Fobj.noiseClsP = noiseClsP[:,:,lmin:]\n",
    "\n",
    "Fobj.ellsP = ellsP[lmin:]\n",
    "Fobj.lminP = lmin\n",
    "Fobj.lmaxP = lmax  \n",
    "Fobj.nMapsP = nMaps\n",
    "Fobj.obsListP = obsListP\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done\n"
     ]
    }
   ],
   "source": [
    "# save new Fobj\n",
    "doSave = True\n",
    "#doSave = False\n",
    "if doSave:\n",
    "    import pickle\n",
    "\n",
    "    filehandler = open('Fisher_7_6A2.obj','w')\n",
    "    #filehandler = open('Fisher_7_6X2_w2_wan_TEB.obj','w') # accuracyBoost=3,doNorm=True,lmax=2000, nz=3e5; nonlinear=True,\n",
    "    #neutrino_hierarchy = 'degenerate'; from FisherCl branch quickCl\n",
    "\n",
    "\n",
    "    pickle.dump(Fobj,filehandler)\n",
    "    filehandler.close()\n",
    "\n",
    "    print 'done'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXoAAAEMCAYAAADK231MAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "# just take a look at TT, TE, EE, BB\n",
    "plt.loglog(ellsP,crossCls[0,0]*ellsP*(ellsP+1),label='TT')\n",
    "#plt.loglog(ellsP,crossCls[0,1]*ellsP*(ellsP+1),label='ET')\n",
    "plt.loglog(ellsP,crossCls[1,0]*ellsP*(ellsP+1),label='TE')\n",
    "plt.loglog(ellsP,crossCls[1,1]*ellsP*(ellsP+1),label='EE')\n",
    "if nMaps == 3 and myClName == 'total': # BB is zero for unlensed spectra\n",
    "    plt.loglog(ellsP,crossCls[2,2]*ellsP*(ellsP+1),label='BB')\n",
    "plt.title('C_l TT, TE, EE (,BB)')\n",
    "plt.legend()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Check out these uncertainties!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [],
   "source": [
    "getSigmas = lambda Fij: np.sqrt(np.diag(np.linalg.inv(Fij)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "from TEB Fisher Matrix:\n",
      "ombh2 :  8.687822560083633e-06 , original:  8.80988471849915e-06\n",
      "omch2 :  0.0003929044726455512 , original:  0.0003959759210277408\n",
      "cosmomc_theta :  9.379899682808389e-06 , original:  8.680089639141158e-06\n",
      "As :  6.203960245671781e-10 , original:  5.923696448050441e-10\n",
      "ns :  0.0011818193930295436 , original:  0.0014251121121470452\n",
      "tau :  0.18207904393996394 , original:  0.12582349865118583\n",
      "mnu :  0.44965119118933 , original:  0.49521756886450985\n",
      "w :  57.827315831102126 , original:  52.29416127562633\n",
      "wa :  149.13581013376017 , original:  134.59611994264677\n"
     ]
    }
   ],
   "source": [
    "myLmin = 30\n",
    "FijTEB = Fobj.makeFisher(myLmin,TE=True)\n",
    "Fijkg = Fobj.makeFisher(myLmin)\n",
    "sigmasTEB = getSigmas(FijTEB)\n",
    "sigmaskg = getSigmas(Fijkg)\n",
    "\n",
    "print 'from TEB Fisher Matrix:'\n",
    "for i,paramName in enumerate(paramList[:nCosParams]):\n",
    "    print paramName, ': ',sigmasTEB[i],', original: ',sigmas_TEB_original[i]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# the difference comes from cutting deltaPs in half"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Loop over different noise parameters to create grid of $\\sigma(M_\\nu)$ values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define row,col deleting function\n",
    "def fixParam(Fij,pNum):\n",
    "    return np.delete(np.delete(Fij,pNum,0),pNum,1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [],
   "source": [
    "# define function to get f_ij^kk, or f_ij^gg\n",
    "def getFijkk(Fobj,myLmin=30,lmax=2000,verbose=False,gg=0):\n",
    "    \"\"\"\n",
    "        Inputs:\n",
    "            Fobj: a FisherMatrix object\n",
    "            myLmin, lmax:\n",
    "            verbose:\n",
    "            gg: index to extract from diagonal. \n",
    "              Must be 0 <= gg <= nBins\n",
    "              Default: 0 (for Kappa)\n",
    "        remake kk covar and fisher with noise\n",
    "        (copy some code from \"Rebuild covar with noise\")\n",
    "        Originally written for kk, but modified to do any gg; ZK, 2018.05.10\n",
    "    \"\"\"\n",
    "\n",
    "    \"\"\"\n",
    "    Clkk = Fobj.crossCls[0,0]\n",
    "    Nlkk = Fobj.noiseCls[0,0]\n",
    "    dClkk = Fobj.dClVecs[0]\n",
    "    nCosParams = Fobj.nCosParams\n",
    "    ells = Fobj.ells\n",
    "\n",
    "    # add noise to crossCls for making covar\n",
    "    Clkk2 = Clkk+Nlkk\n",
    "\n",
    "    print 'Clkk shape: ',Clkk2.shape\n",
    "    print 'dClkk shape: ',dClkk.shape\n",
    "    cov_kk2 = 2*Clkk2**2 / (2*ells+1)\n",
    "    \"\"\"\n",
    "    # check gg\n",
    "    if gg < 0 or gg > Fobj.nBins:\n",
    "        print 'covfefe!'\n",
    "        return 0\n",
    "    \n",
    "    if gg == 0:\n",
    "        nParams = Fobj.nCosParams\n",
    "    else:\n",
    "        nParams = Fobj.nParams\n",
    "    cov_kk2 = Fobj.covar[gg,gg]\n",
    "    dClkk = Fobj.dClVecs[gg]\n",
    "    Fij_kk2 = np.zeros((nParams,nParams))\n",
    "    lmin = Fobj.lmin\n",
    "    minEllInd = myLmin-lmin  # eg. 50-2\n",
    "    maxEllInd = lmax-lmin \n",
    "    for pNum1 in range(nParams):\n",
    "        for pNum2 in range(pNum1,nParams):\n",
    "            # the next line does the summation over ell\n",
    "            Fij_kk2[pNum1,pNum2] = np.sum( dClkk[pNum1,minEllInd:maxEllInd] * \\\n",
    "                                           dClkk[pNum2,minEllInd:maxEllInd] / cov_kk2[minEllInd:maxEllInd])\n",
    "            Fij_kk2[pNum2,pNum1] = Fij_kk2[pNum1,pNum2]\n",
    "    \n",
    "    if verbose:\n",
    "        print 'parameters: ',Fobj.paramList[:nParams]\n",
    "        print 'kappa-only, noise-included Fisher matrix (from ell = {0} to {1}):'.format(myLmin,lmax)\n",
    "        print Fij_kk2\n",
    "    return Fij_kk2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Load Planck Fisher matrices to add in\n",
    "Fij_plikHM_TTTEEE_lowTEB = np.loadtxt('Planck_TTTEEE_lowTEB_FromMCMC.txt')\n",
    "Fij_plikHM_lowTEB = np.loadtxt('Planck_lowTEB_FromMCMC.txt')\n",
    "\n",
    "# convert 100*theta_MC to theta_MC\n",
    "mcIndex = 2\n",
    "mcFac = 100\n",
    "Fij_plikHM_TTTEEE_lowTEB[:,mcIndex] *= mcFac\n",
    "Fij_plikHM_TTTEEE_lowTEB[mcIndex,:] *= mcFac\n",
    "Fij_plikHM_lowTEB[:,mcIndex] *= mcFac\n",
    "Fij_plikHM_lowTEB[mcIndex,:] *= mcFac\n",
    "\n",
    "# transform from {\\rm{ln}}(10^{10} A_s) to 10^9 As\n",
    "# only one partial derivative needed\n",
    "# d( ln(10^10*As)) / d(10^9*As) = 1/(10^9*As)*d(10*10^9*As)/d(10^9*As) = 1/(10^9*As)\n",
    "TenToTheNineAs_fid = 2.2\n",
    "AsIndex = 3\n",
    "Fij_plikHM_TTTEEE_lowTEB[:,AsIndex] *= 1/TenToTheNineAs_fid\n",
    "Fij_plikHM_TTTEEE_lowTEB[AsIndex,:] *= 1/TenToTheNineAs_fid\n",
    "Fij_plikHM_lowTEB[:,AsIndex] *= 1/TenToTheNineAs_fid\n",
    "Fij_plikHM_lowTEB[AsIndex,:] *= 1/TenToTheNineAs_fid\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6, 6)\n",
      "(6, 6)\n"
     ]
    }
   ],
   "source": [
    "print Fij_plikHM_TTTEEE_lowTEB.shape\n",
    "print Fij_plikHM_lowTEB.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [],
   "source": [
    "# those do not appear to include mnu.  Proceed anyway."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[ 9.36446387e+07  9.21901342e+06 -8.07056642e+08 -6.16081183e+05\n",
      "   6.17064306e+05  2.67991972e+06]\n",
      " [ 9.21901342e+06  4.26855526e+06  1.89304661e+08 -6.23192683e+05\n",
      "  -1.82030683e+05  2.74710377e+06]\n",
      " [-8.07056642e+08  1.89304661e+08  1.49584949e+11 -5.71563536e+07\n",
      "  -9.59196916e+07  2.50282581e+08]\n",
      " [-6.16081183e+05 -6.23192683e+05 -5.71563536e+07  1.66140356e+05\n",
      "   1.70076039e+05 -7.29190347e+05]\n",
      " [ 6.17064306e+05 -1.82030683e+05 -9.59196916e+07  1.70076039e+05\n",
      "   3.69396337e+05 -7.48672477e+05]\n",
      " [ 2.67991972e+06  2.74710377e+06  2.50282581e+08 -7.29190347e+05\n",
      "  -7.48672477e+05  3.20493551e+06]]\n",
      "[[ 2.74081311e+03 -5.51716128e+04 -1.86887985e+06  1.11630301e+03\n",
      "  -9.22019023e+03 -5.98536093e+03]\n",
      " [-5.51716128e+04 -3.79389870e+04  3.51786924e+05  2.12165464e+03\n",
      "  -4.31934392e+02 -1.13114252e+04]\n",
      " [-1.86887985e+06  3.51786924e+05  4.86980804e+08  1.62699110e+05\n",
      "   1.27844649e+05 -7.77298274e+05]\n",
      " [ 1.11630301e+03  2.12165464e+03  1.62699110e+05 -8.91652886e+01\n",
      "  -1.72737148e+02  5.58918536e+02]\n",
      " [-9.22019023e+03 -4.31934392e+02  1.27844649e+05 -1.72737148e+02\n",
      "   2.91003051e+03 -9.52598168e+01]\n",
      " [-5.98536093e+03 -1.13114252e+04 -7.77298274e+05  5.58918536e+02\n",
      "  -9.52598168e+01 -1.19040104e+03]]\n",
      "[[ 9.36418979e+07  9.27418504e+06 -8.05187762e+08 -6.17197486e+05\n",
      "   6.26284496e+05  2.68590508e+06]\n",
      " [ 9.27418504e+06  4.30649424e+06  1.88952874e+08 -6.25314338e+05\n",
      "  -1.81598749e+05  2.75841520e+06]\n",
      " [-8.05187762e+08  1.88952874e+08  1.49097968e+11 -5.73190527e+07\n",
      "  -9.60475363e+07  2.51059880e+08]\n",
      " [-6.17197486e+05 -6.25314338e+05 -5.73190527e+07  1.66229521e+05\n",
      "   1.70248776e+05 -7.29749265e+05]\n",
      " [ 6.26284496e+05 -1.81598749e+05 -9.60475363e+07  1.70248776e+05\n",
      "   3.66486307e+05 -7.48577217e+05]\n",
      " [ 2.68590508e+06  2.75841520e+06  2.51059880e+08 -7.29749265e+05\n",
      "  -7.48577217e+05  3.20612591e+06]]\n"
     ]
    }
   ],
   "source": [
    "print Fij_plikHM_TTTEEE_lowTEB\n",
    "print Fij_plikHM_lowTEB\n",
    "print Fij_plikHM_TTTEEE_lowTEB - Fij_plikHM_lowTEB"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "10000.0"
      ]
     },
     "execution_count": 40,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# compare to tau prior (5,5 component of Fij)\n",
    "(0.01**2)**(-1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Starting grid point ( 0 , 0 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 0 , 1 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 0 , 2 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 0 , 3 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 0 , 4 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 1 , 0 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 1 , 1 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 1 , 2 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 1 , 3 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 1 , 4 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 2 , 0 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 2 , 1 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 2 , 2 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 2 , 3 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 2 , 4 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 3 , 0 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 3 , 1 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 3 , 2 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 3 , 3 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 3 , 4 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 4 , 0 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 4 , 1 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 4 , 2 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 4 , 3 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "\n",
      "Starting grid point ( 4 , 4 )...\n",
      "building TE noise arrays...\n",
      "building TE covariance matrix... \n",
      "starting covariance set  1  of  2 ... \n",
      "starting covariance set  2  of  2 ... \n",
      "building Fisher matrix from components...\n",
      "invCovP.shape:  (3, 3, 5001) , dClVecs.shape:  (3, 9, 5001)\n",
      "starting bin set  1  of  9\n",
      "starting bin set  2  of  9\n",
      "starting bin set  3  of  9\n",
      "starting bin set  4  of  9\n",
      "starting bin set  5  of  9\n",
      "starting bin set  6  of  9\n",
      "starting bin set  7  of  9\n",
      "starting bin set  8  of  9\n",
      "starting bin set  9  of  9\n",
      "creation of Fisher Matrix complete!\n",
      "\n",
      "done.\n",
      "[[623.73708011 623.76322508 630.75762801 763.63954631 852.56601535]\n",
      " [623.75032356 623.98416055 649.35661989 800.09016043 871.37396427]\n",
      " [623.77678918 624.41835745 663.1839047  816.01139121 879.7008185 ]\n",
      " [623.81643453 625.05127388 675.66715508 824.61462656 886.49958658]\n",
      " [623.8691962  625.86295317 687.3271559  830.5203843  892.6758684 ]]\n"
     ]
    }
   ],
   "source": [
    "# this relies on most of what's defined above\n",
    "\n",
    "tempNoises = [1,3,5,7,9] # muK arcmin\n",
    "beamSizes  = [1,3,5,7,9] # arcmin\n",
    "mnuGrid = np.empty((5,5)) # to hold results (tempNoise,beamSize)\n",
    "\n",
    "# identify indices with parameters and define tau prior\n",
    "tauIndex = 5\n",
    "tauPrior = 1/0.01**2\n",
    "mnuIndex = 6\n",
    "w0Index  = 7\n",
    "waIndex  = 8\n",
    "\n",
    "# define fsky\n",
    "fsky_S4 = 0.4\n",
    "fsky_Pl = 0.2 # for additional Planck component\n",
    "#doPlanck = True\n",
    "doPlanck = False\n",
    "        \n",
    "for i,tempNoise in enumerate(tempNoises):\n",
    "    for j,beamSize in enumerate(beamSizes):\n",
    "        print '\\nStarting grid point (',i,',',j,')...'\n",
    "        \n",
    "        # get noise\n",
    "        print 'building TE noise arrays...'\n",
    "        fwhm = beamSize; ST = tempNoise; SP = ST*1.414\n",
    "        noiseClsP = getNoiseClsP(fwhm,ST,SP,lmax,nMaps)\n",
    "        \n",
    "        # make covariance matrices with noise\n",
    "        print 'building TE covariance matrix... '\n",
    "        covarP,invCovP,ellsP,obsListP = Fobj.makeCovar(crossCls,noiseClsP,0,lmax)\n",
    "\n",
    "        # get Fisher matrix\n",
    "        FijTEB = getFisher(myLmin,invCovP,dClVecs,nCosParams,lmax)\n",
    "        \n",
    "        # define and add kappa matrix\n",
    "        #print 'building reconstuction noise and kappa Fisher matrix...'\n",
    "        #ells,EB_noise = ncl.getRecNoise(Fobj.lmax,ST,SP,fwhm)\n",
    "        #Nlkk = EB_noise * ells*(ells+1)/4 # deflection to convergence\n",
    "        #Fobj.noiseCls[0,0] = Nlkk[Fobj.lmin:Fobj.lmax+1]\n",
    "        #Fijkk = getFijkk(Fobj,myLmin=30,lmax=2000,verbose=False,gg=0)\n",
    "        #FijTEB[:Fobj.nCosParams,:Fobj.nCosParams] += Fijkk\n",
    "        \n",
    "        # add sigma = 0.01 prior on tau\n",
    "        #FijTEB[tauIndex,tauIndex] += tauPrior\n",
    "        \n",
    "        # fix w0,wa\n",
    "        FijTEB = fixParam(FijTEB,waIndex)\n",
    "        FijTEB = fixParam(FijTEB,w0Index)\n",
    "        \n",
    "        # derive a TTTEEE only Fij\n",
    "        #Fij_plikHM_TTTEEE = Fij_plikHM_TTTEEE_lowTEB - Fij_plikHM_lowTEB\n",
    "        \n",
    "        # add in Planck and fsky\n",
    "        if doPlanck:\n",
    "            FijTEB[:6,:6] += Fij_plikHM_lowTEB  # for Planck l<30\n",
    "        FijTEB *= fsky_S4\n",
    "        if doPlanck:\n",
    "            FijTEB[:6,:6] += Fij_plikHM_TTTEEE_lowTEB*fsky_Pl # for additonal sky not covered by S4 and low ell\n",
    "            #FijTEB[:6,:6] += Fij_plikHM_TTTEEE*fsky_Pl # for additonal sky not covered by S4\n",
    "            \n",
    "        # get sigmas\n",
    "        sigmasTEB = getSigmas(FijTEB)\n",
    "        \n",
    "        # save sigma(mnu)\n",
    "        mnuGrid[i,j] = sigmasTEB[mnuIndex]*1000 # convert eV to meV\n",
    "        \n",
    "print 'done.'\n",
    "        \n",
    "print mnuGrid"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "done.\n"
     ]
    }
   ],
   "source": [
    "np.savetxt('mnuGrid2.txt',mnuGrid)\n",
    "print 'done.'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[241.66077999 241.66385384 242.4687353  258.86300829 265.04892358]\n",
      " [241.66240594 241.6899212  244.92298798 261.9105651  266.84716819]\n",
      " [241.66565586 241.74127492 247.0868522  262.97568599 267.97067094]\n",
      " [241.67052579 241.81643521 249.00601533 263.5430625  268.67925234]\n",
      " [241.67700981 241.91337132 250.67423556 263.93141347 269.15762199]]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "'\\nwithout Planck\\n[[146.02558465 146.02730834 146.4785892  155.58644436 159.00511842]\\n [146.02649587 146.04192532 147.85421888 157.27170006 160.00930829]\\n [146.02831722 146.07072121 149.06429058 157.8619916  160.63752371]\\n [146.03104647 146.11286602 150.13451008 158.17656595 161.03187293]\\n [146.03468029 146.16722078 151.06244736 158.39146605 161.2967478 ]]\\n \\nwith Planck\\n[[43.52908334 43.53212002 44.33913468 56.14975889 60.27794788]\\n [43.5306227  43.5578101  46.8418815  58.29214552 61.71640835]\\n [43.53369945 43.60844371 48.75420622 59.0357768  62.56418291]\\n [43.5383097  43.68260303 50.20945297 59.42766651 63.01726016]\\n [43.54444758 43.77833432 51.33965143 59.66949567 63.27737754]]\\n\\n'"
      ]
     },
     "execution_count": 30,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# apply fsky\n",
    "# no don't since I put it into the function above\n",
    "#fsky = 0.4\n",
    "#print mnuGrid*fsky**(-1/2.)\n",
    "\"\"\"\n",
    "Without Planck, kappa, tau prior\n",
    "[[623.73708011 623.76322508 630.75762801 763.63954631 852.56601535]\n",
    " [623.75032356 623.98416055 649.35661989 800.09016043 871.37396427]\n",
    " [623.77678918 624.41835745 663.1839047  816.01139121 879.7008185 ]\n",
    " [623.81643453 625.05127388 675.66715508 824.61462656 886.49958658]\n",
    " [623.8691962  625.86295317 687.3271559  830.5203843  892.6758684 ]]\n",
    "\n",
    "without Planck\n",
    "[[146.02558465 146.02730834 146.4785892  155.58644436 159.00511842]\n",
    " [146.02649587 146.04192532 147.85421888 157.27170006 160.00930829]\n",
    " [146.02831722 146.07072121 149.06429058 157.8619916  160.63752371]\n",
    " [146.03104647 146.11286602 150.13451008 158.17656595 161.03187293]\n",
    " [146.03468029 146.16722078 151.06244736 158.39146605 161.2967478 ]]\n",
    " \n",
    "with Planck lowEll\n",
    "[[152.83969718 152.84164125 153.35069298 163.71934163 167.63165798]\n",
    " [152.84072552 152.8581277  154.90289867 165.6467858  168.76896773]\n",
    " [152.84278095 152.89060664 156.27144657 166.32042739 169.47953326]\n",
    " [152.84586097 152.93814219 157.4852319  166.67926781 169.92767949]\n",
    " [152.84996182 152.99944996 158.54030702 166.92488253 170.23022702]]\n",
    " \n",
    "with Planck extra sky\n",
    "[[43.49546425 43.49849745 44.30453905 56.09930488 60.22180829]\n",
    " [43.49700204 43.52415835 46.80491626 58.23859274 61.6603379 ]\n",
    " [43.50007566 43.57473451 48.71527304 58.9808941  62.50704689]\n",
    " [43.5046812  43.64880989 50.1685217  59.37209887 62.95916629]\n",
    " [43.51081283 43.74443316 51.29687047 59.61348843 63.21860998]]\n",
    " \n",
    "with Planck lowEll and extra sky\n",
    "[[43.52908334 43.53212002 44.33913468 56.14975889 60.27794788]\n",
    " [43.5306227  43.5578101  46.8418815  58.29214552 61.71640835]\n",
    " [43.53369945 43.60844371 48.75420622 59.0357768  62.56418291]\n",
    " [43.5383097  43.68260303 50.20945297 59.42766651 63.01726016]\n",
    " [43.54444758 43.77833432 51.33965143 59.66949567 63.27737754]]\n",
    "\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Combine the Fisher matrices and see smaller sigmas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "comparison of sigma(T,E), sigma(kappa,g), and sigma(T,E,kappa,g):\n",
      "\n"
     ]
    },
    {
     "ename": "ValueError",
     "evalue": "operands could not be broadcast together with shapes (9,9) (7,7) (9,9) ",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mValueError\u001b[0m                                Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-31-6516f09c5bb6>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m()\u001b[0m\n\u001b[1;32m      3\u001b[0m \u001b[0;32mprint\u001b[0m \u001b[0;34m'comparison of sigma(T,E), sigma(kappa,g), and sigma(T,E,kappa,g):\\n'\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      4\u001b[0m \u001b[0mFijSum\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mFijkg\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcopy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m----> 5\u001b[0;31m \u001b[0mFijSum\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mnCosParams\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0mnCosParams\u001b[0m\u001b[0;34m]\u001b[0m \u001b[0;34m+=\u001b[0m \u001b[0mFijTEB\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m      6\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m      7\u001b[0m \u001b[0msigmasSum\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mgetSigmas\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mFijSum\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mValueError\u001b[0m: operands could not be broadcast together with shapes (9,9) (7,7) (9,9) "
     ]
    }
   ],
   "source": [
    "from prettytable import PrettyTable\n",
    "\n",
    "print 'comparison of sigma(T,E), sigma(kappa,g), and sigma(T,E,kappa,g):\\n'\n",
    "FijSum = Fijkg.copy()\n",
    "FijSum[:nCosParams,:nCosParams] += FijTEB\n",
    "\n",
    "sigmasSum = getSigmas(FijSum)\n",
    "#infSet = ['inf','inf','inf','inf','inf','inf'] # number of infs needs to be number of bins\n",
    "infSet = [0,0,0,0,0,0] # number of infs needs to be number of bins\n",
    "sigmasTEB = np.append(sigmasTEB,infSet)\n",
    "sigmasSum = np.append(sigmasSum,infSet)\n",
    "#print sigmasTEB\n",
    "    \n",
    "from prettytable import PrettyTable\n",
    "# for a set of fsky values\n",
    "for fsky in (1.0,0.75,0.4,0.25):\n",
    "    fskyFac = np.sqrt(1./fsky)\n",
    "    print fskyFac\n",
    "    sigmasA = sigmaskg*fskyFac\n",
    "    sigmasB = sigmasTEB*fskyFac\n",
    "    sigmasC = sigmasSum*fskyFac\n",
    "    t = PrettyTable(['Parameter', 'sigma(T,E)','sigma(k,g)','sigma(T,E,k,g)'])\n",
    "    for i,p in enumerate(Fobj.paramList):\n",
    "        t.add_row([p,\"%5.3g\" % sigmasB[i],\"%5.3g\" % sigmasA[i],\"%5.3g\" % sigmasC[i]])\n",
    "    print 'fsky = {0}:'.format(fsky)\n",
    "    print t\n",
    "\n",
    "print 'which Cl used for T,E,B: ',myClName"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Compare to Byeonghee's result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Byeonghee's result:\n",
    "# This used lmin 50, lmax 1000, halofit\n",
    "sigmasB1 = {\n",
    "    #sigma(b) (4 bins) =  [ 0.01220295  0.0092872   0.01058874  0.03529165]\n",
    "    'cosmomc_theta' : 0.00015113283201,\n",
    "    'ombh2' : 0.000504509008534,\n",
    "    'omch2' : 0.00151341881388,\n",
    "    'ns' : 0.00732594856064,\n",
    "    'As' : 0.0618279762973*1e-9,\n",
    "    'mnu' : 0.0564262366597,\n",
    "    'w' : 0.088058950571,\n",
    "    'tau' : 0.0595781322066,\n",
    "    'bin1' : 0.01220295,\n",
    "    'bin2' : 0.0092872,\n",
    "    'bin3' : 0.01058874,\n",
    "    'bin4' : 0.03529165\n",
    "}\n",
    "# This used lmin 50, lmax 1000, halofit, bias by bin\n",
    "sigmasB2 = {\n",
    "    'cosmomc_theta' : 7.31905124578e-05,\n",
    "    'ombh2' : 0.00117829505232,\n",
    "    'omch2' : 0.00253188668652,\n",
    "    'ns' : 0.0107531686379,\n",
    "    'As' : 0.0836156981509*1e-9,\n",
    "    'mnu' : 0.110362820478,\n",
    "    'w' : 0.0933472979446,\n",
    "    'tau' : 8.65882116522,\n",
    "    'bin1' : 0.01120403,\n",
    "    'bin2' : 0.0123868,\n",
    "    'bin3' : 0.01398429,\n",
    "    'bin4' : 0.0143556\n",
    "}\n",
    "# This one was after a late fix Byeonghee did:\n",
    "\"\"\"\n",
    "+---------------+--------------+--------------+\n",
    "|   Parameter   | sigma_Z(k,g) | sigma_B(k,g) | \n",
    "+---------------+--------------+--------------+\n",
    "|     ombh2     |   0.00182    |   0.00118    |    \n",
    "|     omch2     |   0.00509    |   0.00253    |     \n",
    "| cosmomc_theta |   6.51e-05   |   7.58e-05   |     \n",
    "|       As      |   9.26e-11   |   1.29e-10   |     \n",
    "|       ns      |    0.0116    |    0.0190    |    \n",
    "|      tau      |     0.33     |     8.26     |    \n",
    "|      mnu      |    0.0884    |     0.11     |      \n",
    "|       w       |    0.0448    |    0.0411    |       \n",
    "|      bin1     |    0.011     |    0.0063    |      \n",
    "|      bin2     |    0.0141    |    0.0056    |    \n",
    "|      bin3     |    0.0195    |    0.0054    |     \n",
    "|      bin4     |    0.0236    |    0.0047    |\n",
    "+---------------+--------------+--------------+\"\"\"\n",
    "sigmasB3 = {\n",
    "    'cosmomc_theta' : 7.58e-05,\n",
    "    'ombh2' : 0.00118,\n",
    "    'omch2' : 0.00253,\n",
    "    'ns' : 0.0190,\n",
    "    'As' : 1.29e-10,\n",
    "    'mnu' : 0.11,\n",
    "    'w' : 0.0411,\n",
    "    'tau' : 8.26,\n",
    "    'bin1' : 0.0063,\n",
    "    'bin2' : 0.0056,\n",
    "    'bin3' : 0.0054,\n",
    "    'bin4' : 0.0047\n",
    "}\n",
    "\n",
    "\n",
    "Bsigmas = sigmasB3\n",
    "\n",
    "print paramList\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "# remove tau from my Fisher matrix, since Byeonghee omitted it.\n",
    "# define row,col deleting function\n",
    "def fixParam(Fij,pNum):\n",
    "    return np.delete(np.delete(Fij,pNum,0),pNum,1)\n",
    "Fij3 = fixParam(Fij2,5) # 5 for neutrinos\n",
    "sigmas3 = np.sqrt(np.diag(np.linalg.inv(Fij3)))\n",
    "sigmas3 = np.insert(sigmas3,5,0)\n",
    "print sigmas3\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "# convert sigma_bi to sigma_bi/fid_bi\n",
    "fidBs = Fobj.binBs\n",
    "print 'fiducial B_i: ',fidBs\n",
    "sigmas3[-4:] = sigmas3[-4:]/fidBs\n",
    "print sigmas3\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "\"\"\"\n",
    "#paramList = ['ombh2','omch2','cosmomc_theta','As','ns','tau','mnu']\n",
    "mask = [True,True,False,True,True,False,True, True,True,True,True]\n",
    "paramListB = np.array(paramList)\n",
    "paramListB = np.append(paramListB,['bin1','bin2','bin3','bin4'])\n",
    "paramListB = paramListB[mask]\n",
    "sigmasB = sigmas3[mask]\n",
    "print sigmas3\n",
    "print sigmasB\n",
    "print paramListB\n",
    "print lmin\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from prettytable import PrettyTable\n",
    "print 'wa and bias marginalized'\n",
    "t = PrettyTable(['Parameter', 'sigma_Z(k,g)','sigma_B(k,g)','percent diff.'])\n",
    "for i,p in enumerate(paramList[:8]):\n",
    "    Zi = sigmasB[i]\n",
    "    Bi = Bsigmas[paramList[i]]\n",
    "    diff = (Bi-Zi)/Zi *100\n",
    "    t.add_row([p,\"%5.3g\" % Zi,\"%5.3g\" % Bi, \"%5.3g\" % diff ] )\n",
    "print t"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# This comparison has probably lost its meaning"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.15"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}