Brief Introduction of the Calculation inside GADGET2 Simulation & Plotting
No progress
14:10
Young's report. FoF and MST algorithm10m
Speaker:
Young Ju
FoF and MST algorithm (Writing a paper)
Mulguisin Clustering Algorithm I. - Comparison of Clustering Algorithms for Study of Cosmic Structure Finding
14:20
Hannah's Report10m
See the slides from 8p
Speaker:
Hannah Jhee(University of Seoul)
Axionyx: CDM Mixed Simulation
One plot
dv graph
Right side: Over estimated value
14:30
Seyeon's Report - CLML10m
Speaker:
Se Yeon Hwang(Universe of Seoul)
Cosmology with Large scale structure using Machine Learning (CLML)
Outline
Checking σ8 effect in the simulation
CNN training visualization (for KAML)
ViT result (update)
1. Checking σ8 effect on the input data
Our 6 parameters simulation with same random seed
(Box size, number of particles)
(1 (Gpc), 10243)
(Box size, number of particles)
(2 (Gpc), 10243)
Other parameters..
(Omegabaryon, h, ns, w0, wa)
(0.0493, 0.6736, 0.9649, -1, 0)
Different parameters..
(Sigmalow, Sigmahigh)
(0.75, 0.85)
Light cone
3d histogram axis = RA, Dec and observed redshift
Color = The number of dark matter halo inside the bins
1Gpc, redshift = 0.1 ~ 0.3
σ8 = 0.75 (1 Gpc)
Maximum number of halo = 79
σ8 = 0.85 (1 Gpc)
Maximum number of halo = 82
2Gpc, redshift = 0.1 ~ 0.8
σ8 = 0.75 (2 Gpc)
Maximum number of halo = 86
σ8 = 0.85 (2 Gpc)
Maximum number of halo = 102
Snap shot
3d histogram axis = x, y, z (Mpc/h)
Color = The number of dark matter halo inside the bins
2. CNN training visualization (for KAML)
Feature map while doing CNN...
Bin size = 16
Final result of CNN
Bin size = 32
3. ViT result (update)
About attention mechanism
Vit is algorithm that calculates score between words and words independently while CNN has filter that can moving around each pixels.
Therefore, how to make words (in my case patches) is very important in ViT.
CNN
Ωm = 0.393707, σ8 = 0.695878
Ωm = 0.187478, σ8 = 0.734768
14:40
Sumi's report10m
Speaker:
Sumi Kim(University of Seoul)
Cosmology of High-Order Statistics (CHOA)
Preparing for the KAS
Previously on IAU...
Goal: Npcf result, 2 pcf, 3 pcf, 4 pcf → Decision tree, random forest → Parameter estimation, Ωm, σ8, w0, wa, ns, h → To see 4pcf gets better information than low order statistics
Using 2 pcf + 3 pcf + 4 pcf calculation... → Ωm, σ8, w0, wa, ns, h, parameters are not trained well → Ωm, σ8 parameter space, Ωm, σ8, parameter contour is biased.
1. 2 pcf, 3 pcf, 4 pcf calculation
Why the data is not trained at last time?
2 pcf: 110mpc, anisotropic inclued
3 pcf: Isotropic
4 pcf: Isotropic
2. Machine learning to get parameter contour.
Anisotropic: Put 2D array in random forest
How to correct the bias?
+α using DESI data?
Will it make it in time?
Which data to use?
Isn't there an embargo?
1. 2 pcf, 3 pcf, 4 pcf calculation
Why the data is not trained at last time?
2pcf: 110mpc, anisotropic included
3pcf: Isotropic
4pcf: Isotropic
Previous data (800 mpc, 0.1 < z < 0.3) → New data (2 Gpc, 0 < z < 0.8)
Calcution time check (minute)
2 pcf: 110 mpc, ~ 12 m
3 pcf: 65 mpc, ~ 33 m
4 pcf: 40 mpc, ~ 82 m
Anisotropic example
mu - sig graph
Pinocchio data redshift distribution
z graph
Cosmology parameters
0
100
500
2. Machine learning to get parameter contour.
Anisotropic: Put 2D array in random forest
How to correct the bias?
Isotropic result: 1D array
Anisotropic result: 2D array → Flatten, 1D array
Ωm, σ8 parameter space
Ωm - σ8 graph
+α using DESI data?
Will it make it in time?
Which data to use?
Isn't there an embargo?
Abstract
The N-point spatial correlation function is a commonly used method of compressing the information of the large-scale distribution of galaxies. It can be used to extract the background expansion information via a standard (scale) known as the BAO, and growth of structure information via redshift-space distortions. Theoretical models of the higher order statistics are notoriously difficult to predict. In this work we will look at the cosmological information contained in the higher order clustering statistics upto the fourth order using a suite of N-body simulations as our theoretical model. We apply this methodology to the SDSS CMASS sample of galaxies, and compare with previous results. We also make forecasts for future data from DESI.
Some mock simulations if time allows.
LRG Ezmock simulation
RA - DEC graph
LRG Ezmock redshift
Z graph
14:50
In's Report10m
Speaker:
In Hwang(University of Seoul)
Kerr Black Hole Mechanics using EinsteinPy
Explaining concepts
What is the black hole?
A celestial body whose escape velocity is greater than the speed of light.
Escape velocity means the speed at which the kinetic energy (velocity) of an object exceeds the gravitational potential energy (orbital velocity) of a planet, etc. It is usually understood as the action of orbiting or the speed of passing through a gravitational field such as a satellite.
A black hole is a region of spacetime where gravity is so strong that nothing – no particles or even electromagnetic radiation such as light – can escape from it.
Classification and physical meaning of black holes
Angular momentum (X), electric charge (X)
Schwarzschild black hole
Angular momentum (O), electric charge (X)
Kerr black hole
Angular momentum (X), electric charge (O)
Reissner-Nordström black hole
Angular momentum (O), electric charge (O)
Kerr-Newman black hole
The basic elements of a black hole include ‘mass’, ‘angular momentum’, and ‘electric charge’, but the author creatively added a fourth property called ‘temperature’.
Schwarzchild black hole: It is spherical and all mass is concentrated in the singularity. Since it is at rest, there is no momentum
Kerr black hole: It is elliptical and rotates, so it has momentum and the momentum is proportional to the black hole's mass.
Reissner-Nordström black hole: The charges accumulated in the star become neutral so quickly that it is almost impossible for charged black holes to exist.
Kerr-Newman black hole: Mass, angular momentum, and electric charge have all the basic elements of a black hole.
Classification and physical meaning of black holes
Static limit: It is the outermost surface of the black hole. If we are approaching the black hole in a spaceship, it will be pulled in the direction of rotation around the black hole due to frame-dragging.
Ergosphere: In the region between the static limit and the outer event horizon, the particles continue to rotate without being stationary due to frame-dragging. Escape from the Ergosphere is possible.
Event horizon: It is a virtual area that cannot be touched by the surface of a black hole. If any material crosses the surface of a black hole, it is sucked into the center of the black hole and becomes part of the black hole singularity. The outer event horizon and the inner event horizon have the same properties. And neither can get out.
Singularity: When a massive star collapses into a point at the center of a black hole, all the material in the star will disappear into the singularity, so it will have infinite density. Kerr black hole has a ring-shaped singularity.
Visualizing Event Horizon and Ergosphere (Singularities) of Kerr Metric or Black Hole
Setting parameters
Metric or black hole parameters: M = Sun-mass black hole’s mass, a = Dimensionless spin parameter
M = 1.9891e30 kg
Let's look at the four spin parameter a.
Radius [m] - radius [m] graph
a1 = 0.1
a2 = 0.4
a3 = 0.9
a4 = 0.99 → Extremal Kerr black hole
Schwarzschild radius, rs = 2GM/c2 = 2948.3 m
0 ≤ a = cJ/GM2 ≤ 1
Physical meaning of spin parameter a
Maximum angular momentum of a Kerr black hole corresponds to a spin parameter a = 1.
Cannot spin a Kerr black hole up beyond this limit a = 1.
The value of cJ/GM2 can exceed 1 for objects other than black holes.
→ M is the sun-mass black hole's mass.
Plotting four black holes
The surfaces are clearly visible in the plots. Going radially inward, we have Outer Ergosphere, Outer Event Horizon, Inner Event Horizon and Inner Ergosphere. Here, Inner Ergosphere is a mathematical drawing. We can also observe the following
As 𝑎 → 1 (its maximum attainable value), the individual singularities become prominent.
As 𝑎 → 0, some singularities appear to fade away, leaving us with a single surface, that is the Event Horizon of a Schwarzschild black hole.
Visualizing Frame-dragging in Kerr Space-time
What is frame-dragging?
Frame-dragging is a phenomenon predicted by the general theory of relativity.
This phenomenon is a gravitational effect that occurs when an object with very large mass rotates, and the surrounding space-time also rotates accordingly.
In other words, it refers to the fact that the moving mass-energy distribution affects time and space, resulting in a drag effect on the inertial system.
Frame-dragging phenomenon appears between the static limit and the outer event horizon. In other words, frame-dragging phenomenon appears in the ergospehre.
Setting parameters
rs = 2GM/c2
Spin parameter a = 0.99
Setting momentum's ϕ-component to negative, implies an initial retrograde trajectory
GM/c2 - GM/c2 graph, used rs = 2GM/c2.
→ Physical meaning: As can be seen in the plot on the left, the photon's trajectory is reversed, due to frame-dragging effects, so that, it moves in the direction of the black hole's spin, before eventually falling into the black hole.
Future work
Try making some more black hole plots using EinsteinPy.
The Hawking temperature determines the mass of the black hole, and the mass of the black hole determines the evaporation time of the black hole.
Find the masses of the black holes when the evaporation time of the black holes is Planck time.
This condition means the initial masses of the black holes where the effects of quantum gravity can appear. Calculate the condition to find the critical masses.
The initial mass is the mass before Hawking radiation occurs.
The final mass is the mass after the Hawking radiation has finished.
Here, the final masses of the black holes are zero. The reason is that Hawking radiation reduces mass.
15:00
John Suarez's report10m
Research Projects
1. From galaxies to cosmic web DM environments
Abstract
Problem: One of the main goals in cosmology is understanding the distribution of Dark Matter (DM) in the local Universe. The distribution of DM is not possible to observe directly.
Proposal: To make an inference of the DM distribution from observational measurements of galaxies distributions like the DESI experiment.
1. To predict the DM cosmic web from the galaxies distribution training a ML model with simulations.
2. To extrapolate the model to observational measurements.
The illustris TNG project
The next generation of cosmological hydrodynamical simulations.
1) CIC
2) Smoothing
3) Overdensity
4) Potential FFT
5) Tidal tensor
6) Eigenvalues
7) T-web class
T_αβ = ∂2ϕ/((∂r_α)(∂r_β))
2. Alignment between galaxies and filaments
Abstract
Goal: To study the environmental dependent evolution of local galaxies
Proposal: To understand the correlations between the galaxy distribution and its environment.
To identify the alignment of galaxies and their nearest cosmic web filament.
Does this alignment depend on some environmental features?
To create valued added catalogs for the DESI collaboration.
15:10
Dr.Sabiu's Report10m
Speaker:
DrCristiano Sabiu(University of Seoul)
Deep Learning the Deep Sky: Recovering low surface brightness objects with machine learning