CPLUOS Astrophysics team meeting

Friday 24 Jun 2022, 14:00 → 15:10 Asia/Seoul

물리학과

https://uos-ac-kr.zoom.us/j/8264902652

14:00 → 15:00 COSMOLOGY¶

14:00 Hyeonmo's Report¶

Speaker: Hyeonmo Koo (University of Seoul)

Head-on Collsion of FDM/CDM Halos

• Snapshot: FDM vs CDM

14:10 Young's report. FoF and MST algorithm¶

Speaker: Young Ju

FoF and MST algorithm

• Controlled data 1
  • Number of clusters: 50
  • sig = 1: separated data
  • sig = 10: diffused data
  • Aim: MGS algorithm can find clusters as similar to other algorithms
• Controlled data 2
  • Number of clusters: 50
  • Distribution of galaxy: power-law
  • Aim: MGS has different behavior
• Controlled data 3
  • Number of clusters: 500
  • Generate cluster from halo
  • mass function and HOD
  • Aim: Test algorithms for more complex system
• Real data: kias vagc data
  • Linking length (h^-1Mpc) - Number of clusters graph
• Real data: kias vagc data
  • 1st cluster: the biggest cluster
    • Percolation: MGS
  • 2nd cluster

14:20 Hannah's Report¶

See the slides from 8p

Speaker: Hannah Jhee (University of Seoul)

Filament

• Snapshot: Fig R1
  • Filament structures from  z = 1 ~ z = 0 with halo distribution (green triangles).
  • Filament evolution as a function of time (see the color scheme).

14:30 Sumi's report¶

Speaker: Sumi Kim (University of Seoul)

Calculating Bright Galaxy Survey (BGS) 2 pcf

• DA02 BGS EZmocks 2 pcf I = 0, Nran = 20, used W_fkp calculation
  • s [Mpc/h] - ξ₂ (s)s² [Mpc/h]² graph
    • 0.1 < z < 0.5
    • 0.1 < z < 0.2
    • 0.2 < z < 0.3
    • 0.3 < z < 0.4 is bigger than other things
    • 0.4 < z < 0.5
    • Plot → Strange!
  • W_FKP: FKP weight
    • Feldman, Kaiser, and Peacock
  • Function of number density, to optimize the clustering measurements facing shot noise and cosmic variance.
• Feedback
  • 1. See the coverage of data and random
  • 2. Calculate with 5 Mpc bin, not 1 Mpc
  • 3. Compare individuals, not average

14:40 CLML¶

Cosmology with Large scale structure using Machine Learning

Speaker: Se Yeon Hwang

Cosmology with Large scale structure using Machine Learning (CLML)

• Outline
  • input data
  • ViT
• Input data
  • Using comoving position (x, y, z) from PINOCCHIO catalogs
    • There are 164,195 halos
    • (boxsize = 1 Gpc, number of particles = 512³ , z_start = 0.3)
  • Using theta and phi from PINOCCHIO catalogs
    • Theta and phi of light cone
  • We rotate the coordinate like this
  • Cut range
  • Make 3-D data using theta, phi and observed  redshift (which is on the redshift-space)
    • theta - z
    • phi - z
    • phi - theta
• Check minimum mass
  • Every simulation has different minimum halo mass because of Ω_m
  • We have to set the same minimum halo mass for all dataset, so we found maximum of minimums applied the mass cut
• Parameter test with each range cut
  • Reference parameter (Abacus)
    • (Ω_m, Ω_b, Ω_cdm, σ₈, w₀, w_a, h, n_s)
      • (0.3133, 0.0493, 0.264, 0.8079, -1, 0, 0.6736, 0.9649)
      • Omega_m = [0.25, 0.40]
      • sigma_8 = [0.74, 0.85]
      • w0 = [-1.3, 0.7]
      • wa = [-0.3, 0.3]
      • ns = [0.9, 1.1]
      • h = [0.6, 0.8]
    • For the lower (blue) and upper (orange), I used same random seed for each parameters and for the reference I used different random seed
    • x axis - k
    • y axis - p (k_taret) - p (k_ref)
• Vision Transformer (ViT)
• Making patches
  • In 2-D
    • Using tensorflow.image.extract_patches
  • In 3-D
    • Using tensorflow.extract_volume_patches
• ViT Tutorial
  • https://keras.io/examples/vision/image_classification_with_vision_transformer/
  • We are doing test running

14:50 Dr.Sabiu's Report¶

Speaker: Dr Cristiano Sabiu (University of Seoul)

Deep Learning the Deep Sky: Recovering low surface brightness objects with machine learning

• No progress

15:00 → 15:10 Discussion¶