Euclid galaxy clustering simulation requirements: the basics Gigi - - PowerPoint PPT Presentation

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 1

Euclid Consortium

Euclid galaxy clustering simulation requirements: the basics

Gigi Guzzo & Will Percival (for the GC SWG)

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 2

Euclid Consortium

Why redshift surveys need simulations…

• 1. Understand and fully characterize survey selection

function(s) (ideal vs observed Universe):

• Verify that experiment will deliver as planned
• Quantify and cure instrumental systematic effects
• Quantify and minimize modeling systematic effects
• Need very realistic mock catalogues in terms of size,

clustering and galaxy properties (mass resolution, photometric and spectral properties, …)

• 2. Estimate statistical errors
• Build representative and reliable covariance matrices
• Need possibly less detailed mock catalogues, but in very

large number (e.g. PT-halos, Pinocchio, COLA…)

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 4

Euclid Consortium

Typical survey idiosyncrasies… holes, gaps

1 1.005 1.01 1.015 1.02 1.025 1.03 1.035 1.04 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1

()/()full v/vfull

5 arcmin2 holes 10 arcmin2 holes single contiguous 7% lost area

• D. Bianchi & LG, Red Book

Fisher matrix prediction

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 5

Euclid Consortium

…but ¡there ¡is ¡more ¡to ¡Euclid… ¡

Will it work at all? A slitless experiment…

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 6

Euclid Consortium

1deg2 of (simulated) Euclid spectroscopy

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 7

Euclid Consortium

NISP performance e2e simulations (Garilli et al.)

Right Ascension

Declination

Step 1: catalog, observing strategy, instrument parameters Step 2: for each object (K<22.5), simulate image and incident spectrum (+ stars) Step 3: simulate dispersed

• bjects (TIPS) in all

dithers Step 4: extract spectra and combine all dithers (AXE -> OU-SIR) Step 5: Measure redshift and reliability (OU- SPE), compute completeness and purity Step 6: Apply Completeness and Purity to expected counts, produce dN/dz

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 8

Euclid Consortium

The key ingredient: the input catalogue…

Need to reproduce accurately (to z=2!):

• a. Large-scale structure (for realistic projection/confusion)
• b. Galaxy spectral properties (for realistic population

abundance and distribution) c. LSS and galaxy properties need to be correlated as in real Universe, as this is crucial for performance test

• 1. Use n-body+semi-analytic mock (e.g. Durham)
• Precise redshifts, area ~10-100 deg2 , as deep as we like
• LSS description ~OK (we know P(k), halos vs galaxies?)
• Galaxy population and connection to LSS not fully realistic

(at the time): how many Ha emitters? Where do they form?

• 2. Use a real, observed multi-band catalogue
• COSMOS field: 30 photometric bands from radio to X-ray
• Accurate photo-z to z=2
• Very good knowledge of galaxy SEDs robust identification
• f flux-limited Ha population + correct location within LSS
• Very small field (<2 deg2)

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 9

Euclid Consortium

Cucciati & VIPERS team, in prep.

Colours: (U-B) rest frame

Simulations need to reproduce this accurately…

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 10

Euclid Consortium

How is the input catalogue built in detail

• Starting point: COSMOS IRAC Catalogue (Ilbert

2010), 1.4x1.4 deg

• Accurate photo-z known, very extended SED known
• Derive galaxy type and stellar mass from SED

fitting, direct association with spectral template from Bruzual & Charlot (no emission lines)

• Spectral type SFR Ha luminosity: add H lines to

spectrum

• Derive metallicity from stellar mass and SFR build

line ratios accordingly (Dopita 2006 models) add metal lines to spectrum

• Derive reddening from EB-V

(Roche and Zamorani 2011 see document on RedMine

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 11

Euclid Consortium

e.g. GC Interim Science Review simulations…

*4-red 0.9 start *3-red 0.9 start

• red+blue baseline

Ngal= 3141 (F>3e-16) Ngal= 3667 (F>2e-16) UPDATED Interim Science review fig. 8.2

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 12

Euclid Consortium

Where do we go from here…

Combine advantages of both approaches (work

• ngoing, as seen earlier today):
• 1. Need larger area mocks
• Push down cosmic variance
• 2. Improved semi-analytic methods?
• How close to reality can they be?
• Especially in correctly placing star-forming objects within the

surrounding structure

• And following the evolution of this through time?
• 3. A hybrid approach?
• E.g. “dress” a large, good resolution n-body sim with

galaxies built with the right SED and SFR-density relations, taken from a real catalogue like COSMOS, in probabilistic sense (in particular, correct type-density relation as a function of redshift…)

• Enough information?

Euclid Simulation SWG meeting Barcelona 4-5 Dec 2014 13

Euclid Consortium

• 2. Errors / Likelihood / Covariance
• e.g. Alina’s talk this morning…
• Needed as to optimize all stages of data analysis
• Including approach to likelihood calculation and error determination
• For this we need to test with many realizations of “Euclid universes”
• We cannot rely on N-body simulations for ~10,000 mock surveys
• Approximate methods - fast N-body, Lagrangian PT-based (e.g.

PTHalos, Pinocchio, COLA, …)

• Need to balance accuracy and speed
• As with all mocks, the sooner we have these, the better
• These are not just required on the same time-scale as the data, but

are required now!

• Existing expertise and activity (WP in OU-LE3, Monaco et al.): time

scales?