import glob import pandas as pd import os import numpy as np import pickle dirs = glob.glob("*/") #dirs = ['DES-SINGLE-RV/'] thingdict = { "DES-SINGLE-RV/":"ORVOEBV", "DES-TWO-RV/":'TRVOEBV', "DES-TWO-RV-TWO-EBV-MULTI-BETA/":"TRVTEBVMB", "DES-TWO-RV-TWO-EBV/":"TRVTEBV", "DES-INTRINSIC-DIFF/":"IGN", "DES-NO-CINT_z0.4/":"IGN", "DES-NO-CINT_sig0/":"IGN", "DES-NO-CINT-NO-MWEBV/":"IGN", "DES-NO-CINT-NO-MWEBV_sig0/":"IGN", "DES-NO-CINT/":"IGN", "CC/training_mass_split/PIa_-99":"CCNOCUT", "CC/training_mass_split/PIa_0.5":"CCSOMECUT", "CC/training_mass_split/PIa_0.9":"CCMORECUT" } HMRVM = [] ; LMRVM = [] ; HMEBV = [] ; LMEBV = [] ; HMRVS = [] ; LMRVS = [] #[HMRVM, LMRVM, HMEBV, LMEBV, HMRVS, LMRVS] def write_out(vals, Ns, thingdict, directory): if thingdict[directory] == "IGN": print(f"This directory {directory} has been set to ignore, skipping") return prefixes = ["HMRVM", "LMRVM", "HMEBV", "LMEBV", "HMRVS", "LMRVS"] with open(f"LATEX-OUT-MG.tex", "a+") as f: for n,blep in enumerate(vals): #blep will be a list of tuples, need to separate it into two try: means, errs = zip(*blep) except ValueError: print(f"This directory, {directory}, is otherwise empty (perhaps this is not yet coded for the mass split results?). Skipping for now.") return print(np.mean(means), np.std(means)) print(np.std(means), np.mean(errs)) val = np.around(np.average(means, weights=1/np.power(errs, 2)), 3) valstderr = np.around(np.std(means) / np.sqrt(len(means)), 3) valstd = np.around(np.std(means), 3) valerr = np.around(np.mean(errs), 3) if val < 0.2: f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}{{${val:0.3f}({valstderr:0.3f})$}}") f.write("\n") f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}std{{${valstd:0.3f}$}}") f.write("\n") f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}err{{${valerr:0.3f}$}}") else: f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}{{${val:0.2f}({valstderr:0.2f})$}}") f.write("\n") f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}std{{${valstd:0.2f}$}}") f.write("\n") f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}err{{${valerr:0.2f}$}}") f.write("\n") for n,blep in enumerate(Ns): f.write(f"{chr(92)}def{chr(92)}{thingdict[directory]}{prefixes[n]}N{{{blep}({2*blep}\%)}}") f.write("\n") return print("Done") try: print("Removing existing output file") os.remove("LATEX-OUT.tex") except FileNotFoundError: pass dirs = thingdict.keys() truths = [ [3, 0.5, 0.08, 3, 0.5, 0.08], [3, 0.5, 0.08, 2, 0.5, 0.08], [3, 0.5, 0.125, 2, 0.5, 0.08], [3, 0.5, 0.125, 2, 0.5, 0.08], [3, 0.5, 0.125, 2, 0.5, 0.08], [], [], [], [], [], [3, 0.5, 0.125, 2, 0.5, 0.08], [3, 0.5, 0.125, 2, 0.5, 0.08], [3, 0.5, 0.125, 2, 0.5, 0.08] ] for i, d in enumerate(dirs): files = glob.glob(f"{d}*training_popRV/*/chains.pkl") if len(files) == 0: files = glob.glob(f"{d}*training_mass_split/*/chains.pkl") if len(files) == 0: files = glob.glob(f"{d}/*/chains.pkl") print(f'Found {len(files)} finished files in {d}') truth = truths[i] HMRVM = [] ; LMRVM = [] ; HMEBV = [] ; LMEBV = [] ; HMRVS = [] ; LMRVS = [] eHMRVM = [] ; eLMRVM = [] ; eHMEBV = [] ; eLMEBV = [] ; eHMRVS = [] ; eLMRVS = [] N95_HMRVM, N95_LMRVM, N95_HMEBV, N95_LMEBV, N95_HMRVS, N95_LMRVS = 0, 0, 0, 0, 0, 0 for f in files: with open(f, "rb") as input_file: e = pickle.load(input_file) #do some processing of the data according to Mattster try: HMRVM.append([np.around(np.mean(e['mu_R']), 6 ), np.around(np.std(e['mu_R']), 6) ]) LMRVM.append([np.around(np.mean(e['mu_R']), 6 ), np.around(np.std(e['mu_R']), 6) ]) HMEBV.append([np.around(np.mean(e['tauA']/e['mu_R']), 6 ), np.around(np.std(e['tauA']/e['mu_R']), 6) ]) LMEBV.append([np.around(np.mean(e['tauA']/e['mu_R']), 6 ), np.around(np.std(e['tauA']/e['mu_R']), 6) ]) HMRVS.append([np.around(np.mean(e['sigma_R']),6 ), np.around(np.std(e['sigma_R']), 6) ]) LMRVS.append([np.around(np.mean(e['sigma_R']),3 ), np.around(np.std(e['sigma_R']), 6) ]) if np.abs((e['mu_R'].mean() - truth[0]) / e['mu_R'].std()) < 2: N95_HMRVM += 1 N95_LMRVM += 1 RVS_l95, RVS_u95 = np.quantile(e['sigma_R'], [0.025, 0.975]) if truth[1] > RVS_l95 and truth[1] < RVS_u95: N95_HMRVS += 1 N95_LMRVS += 1 if np.abs((np.mean(e['tauA']/e['mu_R']) - truth[2]) / np.std(e['tauA']/e['mu_R'])) < 2: N95_HMEBV += 1 N95_LMEBV += 1 except KeyError: HMRVM.append([np.around(np.mean(e['mu_R_HM']), 6 ), np.around(np.std(e['mu_R_HM']), 6) ]) LMRVM.append([np.around(np.mean(e['mu_R_LM']), 6 ), np.around(np.std(e['mu_R_LM']), 6) ]) HMEBV.append([np.around(np.mean(e['tauA_HM']/e['mu_R_HM']), 6 ), np.around(np.std(e['tauA_HM']/e['mu_R_HM']), 6) ]) LMEBV.append([np.around(np.mean(e['tauA_LM']/e['mu_R_LM']), 6 ), np.around(np.std(e['tauA_LM']/e['mu_R_LM']), 6) ]) HMRVS.append([np.around(np.mean(e['sigma_R_HM']),6 ), np.around(np.std(e['sigma_R_HM']), 6) ]) LMRVS.append([np.around(np.mean(e['sigma_R_LM']),6 ), np.around(np.std(e['sigma_R_LM']), 6) ]) if np.abs((e['mu_R_HM'].mean() - truth[0]) / e['mu_R_HM'].std()) < 2: N95_HMRVM += 1 RVS_l95, RVS_u95 = np.quantile(e['sigma_R_HM'], [0.025, 0.975]) if truth[1] > RVS_l95 and truth[1] < RVS_u95: N95_HMRVS += 1 if np.abs((np.mean(e['tauA_HM']/e['mu_R_HM']) - truth[2]) / np.std(e['tauA_HM']/e['mu_R_HM'])) < 2: N95_HMEBV += 1 if np.abs((e['mu_R_LM'].mean() - truth[3]) / e['mu_R_LM'].std()) < 2: N95_LMRVM += 1 RVS_l95, RVS_u95 = np.quantile(e['sigma_R_LM'], [0.025, 0.975]) if truth[4] > RVS_l95 and truth[4] < RVS_u95: N95_LMRVS += 1 if np.abs((np.mean(e['tauA_LM']/e['mu_R_LM']) - truth[5]) / np.std(e['tauA_LM']/e['mu_R_LM'])) < 2: N95_LMEBV += 1 write_out([HMRVM, LMRVM, HMEBV, LMEBV, HMRVS, LMRVS], [N95_HMRVM, N95_HMRVM, N95_HMEBV, N95_LMEBV, N95_HMRVS, N95_LMRVS], thingdict, d)