from barry.config import setup
from barry.models import PowerSpectrumFit, PowerSeo2016, PowerBeutler2017, PowerDing2018, PowerNoda2019
from barry.samplers import DynestySampler
from barry.fitter import Fitter
if __name__ == "__main__":
pfn, dir_name, file = setup(__file__)
sampler = DynestySampler(temp_dir=dir_name, nlive=1000)
fitter = Fitter(dir_name)
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
d = PowerSpectrum_DESIMockChallenge_Handshake(min_k=0.005, max_k=0.3, isotropic=False, realisation="data", fit_poles=[0, 2])
fitter.add_model_and_dataset(PowerBeutler2017(isotropic=False), d, name=f"Beutler 2017 Prerecon", color=cs[0])
fitter.add_model_and_dataset(PowerSeo2016(isotropic=False), d, name=f"Seo 2016 Prerecon", color=cs[1])
fitter.add_model_and_dataset(PowerDing2018(isotropic=False), d, name=f"Ding 2018 Prerecon", color=cs[2])
fitter.add_model_and_dataset(PowerNoda2019(isotropic=False), d, name=f"Noda 2019 Prerecon", color=cs[3])
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
import logging
logging.info("Creating plots")
res = fitter.load()
from chainconsumer import ChainConsumer
from barry.fitter import Fitter
from barry.models.test import TestModel
from barry.datasets.test import TestDataset
from barry.samplers import DynestySampler
if __name__ == "__main__":
pfn, dir_name, file = setup(__file__)
model = TestModel()
data = TestDataset()
sampler = DynestySampler(temp_dir=dir_name, max_iter=None)
fitter = Fitter(dir_name)
fitter.add_model_and_dataset(model, data)
fitter.set_sampler(sampler)
fitter.set_num_walkers(1)
fitter.fit(file)
if fitter.should_plot():
res, = fitter.load()
posterior, weight, chain, evidence, model, data, extra = res
import matplotlib.pyplot as plt
fig, ax = plt.subplots(nrows=2)
ax[0].plot(weight)
ax[1].plot(evidence)
plt.show()
fix_params=[
"om", "b", f"a{{0}}_1", f"a{{0}}_2", f"a{{0}}_3", f"a{{0}}_4",
f"a{{0}}_5"
],
correction=Correction.HARTLAP,
)
model_fixed.set_default("b", 1.591)
model_fixed.set_default(f"a{{0}}_1", 4651.0)
model_fixed.set_default(f"a{{0}}_2", -4882.0)
model_fixed.set_default(f"a{{0}}_3", 2137.0)
model_fixed.set_default(f"a{{0}}_4", -25.43)
model_fixed.set_default(f"a{{0}}_5", 0.01628)
fitter.add_model_and_dataset(model,
d,
name=f"Full Fit",
linestyle=ls,
color=cs[0])
fitter.add_model_and_dataset(model_marg_full,
d,
name=f"Full Analytic",
linestyle=ls,
color=cs[1])
fitter.add_model_and_dataset(model_marg_partial,
d,
name=f"Partial Analytic",
linestyle=ls,
color=cs[2])
fitter.add_model_and_dataset(model_fixed,
d,
name=f"Fixed Bias+Poly",
# Set up the model we'll use. Fix Omega_m and beta. 5 polynomials (default)
# for each of the fitted multipoles. Use full analytic marginalisation for speed
# Apply the Hartlap correction to the covariance matrix.
model = PowerBeutler2017(
recon=dataset.recon,
isotropic=dataset.isotropic,
marg="full",
fix_params=["om", "beta"],
poly_poles=dataset.fit_poles,
correction=Correction.HARTLAP,
n_poly=5,
)
# Create a unique name for the fit and add it to the list
name = dataset.name + " mock mean"
fitter.add_model_and_dataset(model, dataset, name=name)
allnames.append(name)
# Now add the individual realisations to the list
for i in range(len(dataset.mock_data)):
dataset.set_realisation(i)
name = dataset.name + f" realisation {i}"
fitter.add_model_and_dataset(model, dataset, name=name)
allnames.append(name)
# Submit all the jobs to NERSC. We have quite a few (78), so we'll
# only assign 1 walker (processor) to each. Note that this will only run if the
# directory is empty (i.e., it won't overwrite existing chains)
fitter.set_sampler(sampler)
fitter.set_num_walkers(1)
fitter.fit(file)
fitter = Fitter(dir_name)
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
for r in [False]:
t = "Recon" if r else "Prerecon"
# Fix sigma_nl for one of the Beutler models
model = PowerBeutler2017(recon=r, isotropic=False, correction=Correction.NONE)
model.set_default("sigma_nl_par", 10.9)
model.set_default("sigma_nl_perp", 5.98)
model.set_fix_params(["om", "sigma_nl_par", "sigma_nl_perp"])
for mink, maxk in [(0.02, 0.3), (0.02, 0.4), (0.02, 0.5), (0.03, 0.3), (0.03, 0.5), (0.03, 0.4), (0.04, 0.3), (0.04, 0.4), (0.04, 0.5)]:
d = PowerSpectrum_DESIMockChallenge0_Z01(recon=r, isotropic=False, realisation="data", min_k=mink, max_k=maxk)
fitter.add_model_and_dataset(model, d, name=f"B17 Fixed {t}, k = ({mink:0.2f}, {maxk:0.2f})")
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
import logging
logging.info("Creating plots")
from chainconsumer import ChainConsumer
output = []
bestfit_model = []
pks = []
c = getCambGenerator()
r_s = c.get_data()["r_s"]
fitter = Fitter(dir_name)
p = BAOExtractor(r_s)
cs = ["#262232", "#116A71", "#48AB75", "#b7c742"]
for r in [True]:
t = "Recon" if r else "Prerecon"
datae = PowerSpectrum_SDSS_DR12_Z061_NGC(recon=r, min_k=0.03, max_k=0.30, postprocess=p)
for ls, n in zip(["-", ":"], ["", " (No Poly)"]):
if n:
fix = ["om", "f", "a1", "a2", "a3", "a4", "a5"]
else:
fix = ["om", "f"]
fitter.add_model_and_dataset(PowerBeutler2017(postprocess=p, recon=r, fix_params=fix), datae, name=f"Beutler 2017{n}", linestyle=ls, color=cs[0])
fitter.add_model_and_dataset(PowerSeo2016(postprocess=p, recon=r, fix_params=fix), datae, name=f"Seo 2016{n}", linestyle=ls, color=cs[1])
fitter.add_model_and_dataset(PowerDing2018(postprocess=p, recon=r, fix_params=fix), datae, name=f"Ding 2018{n}", linestyle=ls, color=cs[2])
fitter.add_model_and_dataset(PowerNoda2019(postprocess=p, recon=r), datae, name=f"Noda 2019", color=cs[3])
sampler = DynestySampler(temp_dir=dir_name, nlive=300)
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
from chainconsumer import ChainConsumer
c = ChainConsumer()
for posterior, weight, chain, evidence, model, data, extra in fitter.load():
c.add_chain(chain, weights=weight, parameters=model.get_labels(), **extra)
sigma_nl = 6.0 if r else 9.3
beutler.set_default("sigma_nl", sigma_nl)
beutler.set_fix_params(["om", "sigma_nl"])
seo = PowerSeo2016(recon=r)
ding = PowerDing2018(recon=r)
noda = PowerNoda2019(recon=r, postprocess=p)
for i in range(999):
d.set_realisation(i)
de.set_realisation(i)
fitter.add_model_and_dataset(
beutler_not_fixed,
d,
name=f"Beutler 2017 {t}, mock number {i}",
linestyle=ls,
color="p",
realisation=i)
fitter.add_model_and_dataset(
beutler,
d,
name=
f"Beutler 2017 Fixed $\\Sigma_{{nl}}$ {t}, mock number {i}",
linestyle=ls,
color="p",
realisation=i)
fitter.add_model_and_dataset(seo,
d,
name=f"Seo 2016 {t}, mock number {i}",
linestyle=ls,
sampler = DynestySampler(temp_dir=dir_name, nlive=1000)
fitter = Fitter(dir_name)
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
for r in [False]:
t = "Recon" if r else "Prerecon"
d = PowerSpectrum_DESIMockChallenge0_Z01(recon=r, isotropic=False, realisation="data")
# Fix sigma_nl for one of the Beutler models
model = PowerBeutler2017(recon=r, isotropic=False, correction=Correction.NONE)
model.set_default("sigma_nl_par", 10.9)
model.set_default("sigma_nl_perp", 5.98)
model.set_fix_params(["om", "sigma_nl_par", "sigma_nl_perp"])
fitter.add_model_and_dataset(PowerBeutler2017(recon=r, isotropic=False, correction=Correction.NONE, fix_params=["om"]), d, name=f"Beutler 2017 {t}")
fitter.add_model_and_dataset(model, d, name=f"Beutler 2017 Fixed $\\Sigma_{{nl}}$ {t}")
# fitter.add_model_and_dataset(PowerSeo2016(recon=r, isotropic=False), d, name=f"Seo 2016 {t}", linestyle=ls, color=cs[1])
# fitter.add_model_and_dataset(PowerDing2018(recon=r, isotropic=False), d, name=f"Ding 2018 {t}", linestyle=ls, color=cs[2])
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
import logging
logging.info("Creating plots")
from chainconsumer import ChainConsumer
model = PowerBeutler2017(recon=True, isotropic=False, fix_params=["om"], poly_poles=[0, 2], correction=Correction.NONE, marg="full")
pklin = np.array(pd.read_csv("../barry/data/desi_mock_challenge_post_recon/Pk_Planck15_Table4.txt", delim_whitespace=True, header=None))
model.set_fix_params(["om"])
model.set_data(data.get_data())
model.kvals = pklin[:, 0]
model.pksmooth = smooth(model.kvals, pklin[:, 1])
model.pkratio = pklin[:, 1] / model.pksmooth - 1.0
# Power spectrum
ls = "-"
names = [f"Xinyi-std Pk", f"Pedro-std Pk", f"Baojiu-std Pk", f"Xinyi-Hada Pk", f"Hee-Jong-std Pk", f"Yu-Yu-std Pk", f"Javier-std Pk"]
for i in range(7):
data.set_realisation(i)
fitter.add_model_and_dataset(model, data, name=names[i], color=cs[i], realisation=i, ls=ls)
# Correlation Function
data = CorrelationFunction_DESIMockChallenge(recon=True, isotropic=False, fit_poles=[0, 2])
model = CorrBeutler2017(recon=True, isotropic=False, fix_params=["om"], poly_poles=[0, 2], correction=Correction.NONE, marg="full")
ls = "--"
names = [f"Xinyi-std Xi", f"Pedro-std Xi", f"Baojiu-std Xi", f"Xinyi-Hada Xi", f"Hee-Jong-std Xi", f"Yu-Yu-std Xi", f"Javier-std Xi"]
for i in range(7):
data.set_realisation(i)
fitter.add_model_and_dataset(model, data, name=names[i], color=cs[i], realisation=i, ls=ls)
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
sampler = DynestySampler(temp_dir=dir_name)
fitter = Fitter(dir_name)
for r in [True, False]:
rt = "Recon" if r else "Prerecon"
data = PowerSpectrum_SDSS_DR12_Z061_NGC(recon=r,
postprocess=postprocess)
n = PowerNoda2019(postprocess=postprocess,
recon=r,
fix_params=["om", "f", "gamma", "b"])
n.param_dict["b"].default = 1.992 if r else 1.996
fitter.add_model_and_dataset(
n,
data,
name=f"N19 {rt} fixed $f$, $\\gamma_{{rec}}$, $b$",
linestyle="-" if r else "--",
color="o",
shade_alpha=0.7,
zorder=10)
fitter.add_model_and_dataset(
PowerNoda2019(postprocess=postprocess,
recon=r,
fix_params=["om", "f", "gamma"]),
data,
name=f"N19 {rt} fixed $f$, $\\gamma_{{rec}}$",
linestyle="-" if r else "--",
color="r",
shade_alpha=0.2,
)
fitter.add_model_and_dataset(
PowerNoda2019(postprocess=postprocess,
model_hexa = PowerBeutler2017(recon=r,
isotropic=False,
fix_params=["om"],
poly_poles=[0, 2, 4],
correction=Correction.HARTLAP,
marg="full")
model_all = PowerBeutler2017(recon=r,
isotropic=False,
fix_params=["om"],
poly_poles=[0, 1, 2, 3, 4],
correction=Correction.HARTLAP,
marg="full")
fitter.add_model_and_dataset(model_quad,
d_quad,
name=r"$P_{0}+P_{2}$",
linestyle=ls,
color=cs[0])
fitter.add_model_and_dataset(model_odd,
d_odd,
name=r"$P_{0}+P_{1}+P_{2}$",
linestyle=ls,
color=cs[1])
fitter.add_model_and_dataset(model_hexa,
d_hexa,
name=r"$P_{0}+P_{2}+P_{4}$",
linestyle=ls,
color=cs[2])
fitter.add_model_and_dataset(model_all,
d_all,
name=r"$P_{0}+P_{1}+P_{2}+P_{3}+P_{4}$",
ls = "-" if r else "--"
d = CorrelationFunction_SDSS_DR12_Z061_NGC(recon=r, realisation=0)
beutler_not_fixed = CorrBeutler2017()
beutler = CorrBeutler2017()
sigma_nl = 5.0 if r else 8.0
beutler.set_default("sigma_nl", sigma_nl)
beutler.set_fix_params(["om", "sigma_nl"])
seo = CorrSeo2016(recon=r)
ding = CorrDing2018(recon=r)
for i in range(999):
d.set_realisation(i)
fitter.add_model_and_dataset(beutler_not_fixed, d, name=f"Beutler 2017 {t}, mock number {i}", linestyle=ls, color=c4[0], realisation=i)
fitter.add_model_and_dataset(beutler, d, name=f"Beutler 2017 Fixed $\\Sigma_{{nl}}$ {t}, mock number {i}", linestyle=ls, color=c4[0], realisation=i)
fitter.add_model_and_dataset(seo, d, name=f"Seo 2016 {t}, mock number {i}", linestyle=ls, color=c4[1], realisation=i)
fitter.add_model_and_dataset(ding, d, name=f"Ding 2018 {t}, mock number {i}", linestyle=ls, color=c4[2], realisation=i)
fitter.set_sampler(sampler)
fitter.set_num_walkers(1)
fitter.set_num_concurrent(400)
if not fitter.should_plot():
fitter.fit(file)
# Everything below is nasty plotting code ###########################################################
if fitter.should_plot():
from os import path
import matplotlib.pyplot as plt
d_xi = CorrelationFunction_SDSS_DR12_Z061_NGC(recon=r, realisation=0)
ding_pk = PowerDing2018(recon=r)
ding_pk_smooth = PowerDing2018(recon=r, smooth=True)
ding_xi = CorrDing2018(recon=r)
ding_xi_smooth = CorrDing2018(recon=r, smooth=True)
for i in range(999):
d_pk.set_realisation(i)
d_xi.set_realisation(i)
fitter.add_model_and_dataset(
ding_pk,
d_pk,
name=f"Ding 2018 $P(k)$, mock number {i}",
linestyle="-",
color="p",
realisation=i)
fitter.add_model_and_dataset(
ding_pk_smooth,
d_pk,
name=f"Ding 2018 $P(k)$ Smooth, mock number {i}",
linestyle="-",
color="p",
realisation=i)
fitter.add_model_and_dataset(
ding_xi,
d_xi,
name=f"Ding 2018 $\\xi(s)$, mock number {i}",
if __name__ == "__main__":
pfn, dir_name, file = setup(__file__)
sampler = DynestySampler(temp_dir=dir_name, nlive=1000)
fitter = Fitter(dir_name)
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
d = PowerSpectrum_DESIMockChallenge_Handshake(min_k=0.005,
max_k=0.3,
isotropic=False,
realisation="data",
fit_poles=[0, 2])
fitter.add_model_and_dataset(PowerBeutler2017(isotropic=False),
d,
name=f"Cullan's Pk",
color=cs[0])
# Re-do the desi fits, but now using the linear pk provided for the UNIT sim
pklin = np.array(
pd.read_csv(
"../barry/data/desi_mock_challenge_post_recon/Pk_Planck15_Table4.txt",
delim_whitespace=True,
header=None))
model = PowerBeutler2017(isotropic=False)
model.set_fix_params(["om"])
model.set_data(d.get_data())
model.kvals = pklin[:, 0]
model.pksmooth = smooth(model.kvals, pklin[:, 1])
model.pkratio = pklin[:, 1] / model.pksmooth - 1.0
print(i, type, j, realisations[i])
data = PowerSpectrum_DESIMockChallenge_Post(isotropic=False,
recon=recons[i],
realisation=j,
fit_poles=[0, 2, 4],
min_k=0.02,
max_k=0.30,
num_mocks=1000,
type=type)
model = PowerBeutler2017(recon=recon_types[i],
isotropic=False,
fix_params=["om"],
poly_poles=[0, 2, 4],
correction=Correction.HARTLAP,
marg="full")
fitter.add_model_and_dataset(model, data, name=names[counter])
counter += 1
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
# Everything below is nasty plotting code ###########################################################
if fitter.should_plot():
import logging
logging.info("Creating plots")
from chainconsumer import ChainConsumer
splits = ["PreRecon"]
colors = ["#CAF270", "#66C57F", "#219180", "#205C68", "#232C3B"]
types = ["julian_reciso", "julian_reciso", "julian_recsym", "martin_reciso", "martin_recsym"]
recons = [False, True, True, True, True]
recon_types = ["None", "iso", "ani", "iso", "ani"]
realisations = ["data", "data", "data", "data", "data"]
for i, (name, type, recon, recon_type, realisation, color) in enumerate(zip(names, types, recons, recon_types, realisations, colors)):
print(i, name, type, recon, recon_type, realisation)
data = PowerSpectrum_DESILightcone_Mocks_Recon(
isotropic=False, recon=recon, realisation=realisation, fit_poles=[0, 2], min_k=0.02, max_k=0.30, type=type
)
model = PowerBeutler2017(
recon=recon_type, isotropic=False, fix_params=["om"], poly_poles=[0, 2], correction=Correction.NONE, marg="full"
)
fitter.add_model_and_dataset(model, data, name=name, color=color)
fitter.set_sampler(sampler)
fitter.set_num_walkers(5)
fitter.fit(file)
# Everything below is nasty plotting code ###########################################################
if fitter.should_plot():
import logging
logging.info("Creating plots")
from chainconsumer import ChainConsumer
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
poly_poles=[0, 2],
correction=Correction.HARTLAP,
marg="full")
model4 = PowerBeutler2017(recon="iso",
isotropic=False,
fix_params=["om"],
poly_poles=[0, 2, 4],
correction=Correction.HARTLAP,
marg="full")
for d in datasets:
for i in range(999):
d.set_realisation(i)
if 4 in d.fit_poles:
fitter.add_model_and_dataset(model4,
d,
name=d.name + " " + str(i),
realisation=i)
else:
fitter.add_model_and_dataset(model,
d,
name=d.name + " " + str(i),
realisation=i)
fitter.set_sampler(sampler)
fitter.set_num_walkers(1)
fitter.set_num_concurrent(100)
fitter.fit(file)
# Everything below is nasty plotting code ###########################################################
if fitter.should_plot():
from os import path
# redorder cov matrix, pk goes first, then the extracted part
mask_extracted = postprocess.get_is_extracted(datas[0].ks)
index_sort = np.arange(mask_extracted.size).astype(
np.int) + (1000 * mask_extracted)
a = np.argsort(index_sort)
cov_noda = cov_noda[:, a]
cov_noda = cov_noda[a, :]
cov_noda_diag = cov_noda_diag[:, a]
cov_noda_diag = cov_noda_diag[a, :]
datas[1].set_cov(cov_noda)
datas[2].set_cov(cov_noda_diag)
sampler = DynestySampler(temp_dir=dir_name)
fitter = Fitter(dir_name)
fitter.add_model_and_dataset(model, datas[0], name="Mock-computed")
fitter.add_model_and_dataset(model, datas[1], name="Nishimichi 2018, full")
fitter.add_model_and_dataset(model,
datas[2],
name="Nishimichi 2018, diagonal")
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
from chainconsumer import ChainConsumer
c = ChainConsumer()
for posterior, weight, chain, evidence, model, data, extra in fitter.load(
):
c.add_chain(chain,
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
d = PowerSpectrum_DESIMockChallenge_Handshake(min_k=0.005,
max_k=0.3,
isotropic=False,
realisation="data",
fit_poles=[0, 2])
d2 = PowerSpectrum_DESIMockChallenge_Handshake(min_k=0.005,
max_k=0.3,
isotropic=False,
realisation="data",
fit_poles=[0, 2, 4])
fitter.add_model_and_dataset(PowerBeutler2017(isotropic=False),
d,
name=r"$P_{0}$+P_{2}$",
color=cs[0])
fitter.add_model_and_dataset(
PowerBeutler2017(isotropic=False),
d2,
name=r"$P_{0}$+P_{2}+P_{4},\,No\,a_{4}\,poly$",
color=cs[1])
fitter.add_model_and_dataset(PowerBeutler2017(isotropic=False,
poly_poles=[0, 2, 4]),
d2,
name=r"$P_{0}$+P_{2}+P_{4}$",
color=cs[2])
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
for r in [False]:
t = "Recon" if r else "Prerecon"
ls = "-" # if r else "--"
d = PowerSpectrum_Beutler2019_Z061_SGC(recon=r, isotropic=False)
# Fix sigma_nl for one of the Beutler models
model = PowerBeutler2017(recon=r, isotropic=False)
sigma_nl = 6.0 if r else 9.3
model.set_default("sigma_nl_par", 14.1)
model.set_default("sigma_nl_perp", 2.64)
model.set_fix_params(["om", "f", "sigma_nl_par", "sigma_nl_perp"])
fitter.add_model_and_dataset(PowerBeutler2017(recon=r, isotropic=False), d, name=f"Beutler 2017 {t}", linestyle=ls, color=cs[0])
fitter.add_model_and_dataset(model, d, name=f"Beutler 2017 Fixed $\\Sigma_{{nl}}$ {t}", linestyle=ls, color=cs[0])
fitter.add_model_and_dataset(PowerSeo2016(recon=r, isotropic=False), d, name=f"Seo 2016 {t}", linestyle=ls, color=cs[1])
fitter.add_model_and_dataset(PowerDing2018(recon=r, isotropic=False), d, name=f"Ding 2018 {t}", linestyle=ls, color=cs[2])
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
import logging
logging.info("Creating plots")
from chainconsumer import ChainConsumer
for r in [True, False]:
rt = "Recon" if r else "Prerecon"
data = PowerSpectrum_SDSS_DR12_Z061_NGC(recon=r,
postprocess=postprocess)
n = PowerNoda2019(postprocess=postprocess,
recon=r,
nonlinear_type="spt")
n2 = PowerNoda2019(postprocess=postprocess,
recon=r,
nonlinear_type="halofit")
fitter.add_model_and_dataset(n,
data,
name=f"N19 {rt} SPT",
color="r",
shade_alpha=0.7,
linestyle="-" if r else "--",
zorder=10 if r else 2)
fitter.add_model_and_dataset(n2,
data,
name=f"N19 {rt} Halofit",
color="lb",
shade_alpha=0.7,
linestyle="-" if r else "--",
zorder=10 if r else 2)
fitter.set_sampler(sampler)
fitter.set_num_walkers(20)
fitter.fit(file)
ls = "-" if r else "--"
d = PowerSpectrum_SDSS_DR12_Z061_NGC(recon=r, min_k=0.03, max_k=0.21)
de = PowerSpectrum_SDSS_DR12_Z061_NGC(recon=r,
postprocess=p,
min_k=0.03,
max_k=0.21)
# Fix sigma_nl for one of the Beutler models
model = PowerBeutler2017(recon=r)
sigma_nl = 6.0 if r else 9.3
model.set_default("sigma_nl", sigma_nl)
model.set_fix_params(["om", "sigma_nl"])
fitter.add_model_and_dataset(PowerBeutler2017(recon=r),
d,
name=f"Beutler 2017 {t}",
linestyle=ls,
color=cs[0])
fitter.add_model_and_dataset(
model,
d,
name=f"Beutler 2017 Fixed $\\Sigma_{{nl}}$ {t}",
linestyle=ls,
color=cs[0])
fitter.add_model_and_dataset(PowerSeo2016(recon=r),
d,
name=f"Seo 2016 {t}",
linestyle=ls,
color=cs[1])
fitter.add_model_and_dataset(PowerDing2018(recon=r),
d,
# Changing fitting range and sigma values to match Hee-Jong
d = PowerSpectrum_DESIMockChallenge0_Z01(recon=r,
isotropic=False,
realisation="data",
min_k=0.001,
max_k=0.30)
model = PowerBeutler2017(recon=r,
isotropic=False,
correction=Correction.NONE)
model.set_default("sigma_nl_par", 6.2)
model.set_default("sigma_nl_perp", 2.9)
model.set_default("sigma_s", 0.0)
model.set_fix_params(
["om", "sigma_nl_par", "sigma_nl_perp", "sigma_s"])
fitter.add_model_and_dataset(
model, d, name=f"Beutler 2017 New $\\Sigma_{{nl}}$ {t}")
# Now change linear and smooth spectra to Hee-Jong's inputs
pklin = np.array(
pd.read_csv(
"../barry/data/desi_mock_challenge_0/mylinearmatterpkL900.dat",
delim_whitespace=True,
header=None))
pksmooth = np.array(
pd.read_csv(
"../barry/data/desi_mock_challenge_0/Psh_mylinearmatterpkL900.dat",
delim_whitespace=True,
header=None,
skiprows=2))
model2 = PowerBeutler2017(recon=False,
isotropic=False,
model = PowerBeutler2017(recon="iso",
isotropic=False,
fix_params=["om"],
poly_poles=[0, 2],
correction=Correction.HARTLAP,
marg="full")
model4 = PowerBeutler2017(recon="iso",
isotropic=False,
fix_params=["om"],
poly_poles=[0, 2, 4],
correction=Correction.HARTLAP,
marg="full")
for d in datasets:
if 4 in d.fit_poles:
fitter.add_model_and_dataset(model4, d, name=d.name)
else:
fitter.add_model_and_dataset(model, d, name=d.name)
fitter.set_sampler(sampler)
fitter.set_num_walkers(5)
fitter.set_num_concurrent(100)
fitter.fit(file)
# Everything below is nasty plotting code ###########################################################
if fitter.should_plot():
import logging
logging.info("Creating plots")
from chainconsumer import ChainConsumer
sampler = DynestySampler(temp_dir=dir_name, nlive=1000)
fitter = Fitter(dir_name, remove_output=False)
cs = ["#262232", "#116A71", "#48AB75", "#D1E05B"]
for r in [True, False]:
t = "Recon" if r else "Prerecon"
ls = "-" if r else "--"
d = CorrelationFunction_SDSS_DR12_Z061_NGC(recon=r)
# Fix sigma_nl for one of the Beutler models
model = CorrBeutler2017()
sigma_nl = 6.0 if r else 9.3
model.set_default("sigma_nl", sigma_nl)
model.set_fix_params(["om", "sigma_nl"])
fitter.add_model_and_dataset(CorrBeutler2017(), d, name=f"Beutler 2017 {t}", linestyle=ls, color=cs[0])
fitter.add_model_and_dataset(model, d, name=f"Beutler 2017 Fixed $\\Sigma_{{nl}}$ {t}", linestyle=ls, color=cs[0])
fitter.add_model_and_dataset(CorrSeo2016(recon=r), d, name=f"Seo 2016 {t}", linestyle=ls, color=cs[1])
fitter.add_model_and_dataset(CorrDing2018(recon=r), d, name=f"Ding 2018 {t}", linestyle=ls, color=cs[2])
fitter.set_sampler(sampler)
fitter.set_num_walkers(30)
fitter.fit(file)
if fitter.should_plot():
import logging
logging.info("Creating plots")
res = fitter.load()
if True:
from chainconsumer import ChainConsumer
beutler = PowerBeutler2017(recon=r)
sigma_nl = 6.0
beutler.set_default("sigma_nl", sigma_nl)
beutler.set_fix_params(["om", "sigma_nl"])
beutler_extracted = PowerBeutler2017(recon=r, postprocess=p)
beutler_extracted.set_default("sigma_nl", sigma_nl)
beutler_extracted.set_fix_params(["om", "sigma_nl"])
ding_extracted = PowerDing2018(recon=r, postprocess=p)
for i in range(999):
d.set_realisation(i)
de.set_realisation(i)
fitter.add_model_and_dataset(ding,
d,
name=f"D18",
linestyle=ls,
color="p",
realisation=i)
fitter.add_model_and_dataset(beutler,
d,
name=f"B17",
linestyle=ls,
color="p",
realisation=i)
fitter.add_model_and_dataset(ding_extracted,
de,
name=f"D18 + Extractor",
linestyle=ls,
color="p",
realisation=i)
fitter.add_model_and_dataset(beutler_extracted,
if __name__ == "__main__":
pfn, dir_name, file = setup(__file__)
r = True
models = [
PowerBeutler2017(recon=r, smooth_type="hinton2017", name="Hinton2017"),
PowerBeutler2017(recon=r, smooth_type="eh1998", name="EH1998")
]
data = PowerSpectrum_SDSS_DR12_Z061_NGC(name="Recon mean",
recon=r,
min_k=0.02,
max_k=0.30)
sampler = DynestySampler(temp_dir=dir_name)
fitter = Fitter(dir_name)
fitter.add_model_and_dataset(models[0], data, name="Hinton2017")
fitter.add_model_and_dataset(models[1], data, name="EH1998")
fitter.set_sampler(sampler)
fitter.set_num_walkers(10)
fitter.fit(file)
if fitter.should_plot():
from chainconsumer import ChainConsumer
c = ChainConsumer()
pks = {}
for posterior, weight, chain, model, data, extra in fitter.load():
c.add_chain(chain,
weights=weight,
parameters=model.get_labels(),
**extra)
# Run a not so quick test submitting 128 jobs each with a different mock realisation.
# If run on NERSC, this will actually submit 128/num_per_task jobs, and each job will
# do num_per_task.
if __name__ == "__main__":
pfn, dir_name, file = setup(__file__)
data = PowerSpectrum_SDSS_DR12(isotropic=True, recon="iso")
model = PowerBeutler2017(isotropic=data.isotropic, recon=data.recon, marg="full")
sampler = EnsembleSampler(num_walkers=16, num_steps=5000, num_burn=300, temp_dir=dir_name)
fitter = Fitter(dir_name)
for i in range(128):
data.set_realisation(i)
fitter.add_model_and_dataset(model, data, realisation=i)
fitter.set_sampler(sampler)
fitter.set_num_walkers(1)
fitter.set_num_concurrent(128)
fitter.fit(file)
if fitter.should_plot():
from chainconsumer import ChainConsumer
posterior, weight, chain, evidence, model, data, extra = fitter.load()[0]
model_results, summary = get_model_comparison_dataframe(fitter)
c = ChainConsumer()
c.add_chain(chain, weights=weight, parameters=model.get_labels())
c.plotter.plot(filename=pfn + "_contour.pdf")
