def filt(x): return ccg.filters(x, edd_method='sch')
ptch = mapp + gmu * srgp #+nsp t0 = time() ptch = blocker(ptch, 512, 512) btime.append(time() - t0) npp = len(ptch) for i in range(npp): mp = ptch[i] for rs in range(4, 8): keyn = 'ip{}_ipp{}_c{}_'.format(i_patch, i, rs) t0 = time() mc = ccg.curvelet(mp, r_scale=rs) ctime.append(time() - t0) t0 = time() m1 = ccg.filters(mc, edd_method='sob') m2 = ccg.filters(mc, edd_method='sch') ftime.append(time() - t0) res[keyn + 'sob'] = np.std(m1) res[keyn + 'sch'] = np.std(m2) ccg.save(fname, res) dtt = time() - t00 print(np.mean(btime), np.mean(ctime), np.mean(ftime)) print(100. * np.sum(btime) / dtt, 100. * np.sum(ctime) / dtt, 100. * np.sum(ftime) / dtt) exit() # IN CASE YOU WANT TO RUN IT parallel #igmu = int(sys.argv[1]) #gmulist = gmulist[igmu:igmu+1]
tadd = '../data/test_set/healpix_p/' sim_name = tadd.split('/')[3] res_add = '../classic/'+sim_name+'/' ccg.ch_mkdir(res_add) print('Buiding classical method results:') for j,gmu in enumerate(gmulist): dir_name = '{:3.2e}'.format(gmu) add = tadd+dir_name+'/' stds = [] for i in range(480): ccg.pop_percent(j*480+i,480*ngmu) m = np.load(add+str(i)+'.npy') m = ccg.curvelet(m,r_scale = 7) m = ccg.filters(m, edd_method = 'sch') std = np.std(m) stds.append(std) np.save(res_add+dir_name,stds) lst1 = [] for gmu in gmulist: file_name = '{:3.2e}'.format(gmu) var_file = res_add+file_name+'.npy' lst1.append(np.load(var_file)) lst1 = np.array(lst1) result = p_value(lst1) np.save(res_add[:-1],np.array([gmulist,result]))