KAGRA data management and analysis N.Kanda, on behalf of KAGRA - - PowerPoint PPT Presentation

kagra data management and analysis
SMART_READER_LITE
LIVE PREVIEW

KAGRA data management and analysis N.Kanda, on behalf of KAGRA - - PowerPoint PPT Presentation

1 JGW-G1706749-v2 KAGRA data management and analysis N.Kanda, on behalf of KAGRA collaboration The 3rd KAGRA International Workshop (KIW3) May 21-22, 2017 Academia Sinica (NTU campus), Taipei 2 KAGRA data related subgroups CAL


slide-1
SLIDE 1

KAGRA data 
 management and analysis

N.Kanda, on behalf of KAGRA collaboration The 3rd KAGRA International Workshop (KIW3) May 21-22, 2017 Academia Sinica (NTU campus), Taipei

1

JGW-G1706749-v2

slide-2
SLIDE 2

KAGRA data related subgroups

CAL

: Calibration

DGS

: Digital System

DMG

: Data Management

DAS

: Data Analysis

2

Interferometer

computing environment

storage CPUs softwares Data Results

slide-3
SLIDE 3

DMG (Data Management subgroup)

Targets / Tasks

to manage and to operate ‘KAGRA data Tier’

  • Data Transfer
  • Data Archive

+ Development of softwares for these operations

  • Data Distribution

mirror sites and end users

Members

leader : N.Kanda
 sub-leader : K.Oohara effort members : 
 S.Haino, K.Hayama, Y.Inoue, Y.Itoh, M.Kaneyama, G. Kang, C-Y Lin, O.Miyakawa, A.Miyamoto, S.Miyoki, S.Oh, K.Sakai, Y.Sasaki, E. J. Son, H.Tagoshi, H.Takahashi, K.Tanaka, S.Ueki, T.Yamamoto, T.Yokozawa, H.Yuzurihara

3

slide-4
SLIDE 4

Overall of KAGRA data transfer

4

Tier-0 Tier-0.5 Tier-2 Tier-1 Tier-3 Data Sharing with Other

  • bs.

faster (upstream) ↔ later (downstream) less amount ↔ larger amount

Overview of KAGRA data flow

Event Alert to Counterparts/ Follow-ups in KAGRA Mirrors: Academia SINICA (Taiwan) KISTI (Korea)

Tier-1 archive Kashiwa Tier-0 archive Oversea GW experiments (low latency h(t)) Oversea GW experiments (bulk of data) Kamioka Osaka City U. Tier-0.5 for low latency Niigata U, RESCEU Detector (tunnel) raw data ~20MB/s raw data ~20MB/s +

  • Proc. data

~1MB/s raw + Proc. data

  • Proc. data +

partial raw data partial raw&proc. data set

  • Proc. data

~1MB/s

(option : raw data without permanent store)

end user sites

alert data base follow-ups / counterparts

… …

socket (KAGRA DMG software) GRID GRID or alternative low latency (h(t)) Alart in GCN format

Nagaoka Tech.

slide-5
SLIDE 5

KAGRA data

All data sets are in ‘frame’ format.

RAW

Data files derived from DGS’s frame writer Tier-0 : primary fully and permanently archive at Kashiwa Tier-1 : mirror of Tier-1 at Academia SINICA (Taiwan) and KISTI (Korea)

KISTI has iKAGRA data now. It will be extend to ‘full’ data in near future.

Tier-0.5 : low latency transfer to OCU (Osaka) same to Tier-0, but will not keep full sets permanently.

Proc.

Processed data set that consists of calibrated strain h(t) and some related channels for GW event analysis. at Tier-0, 1, 0.5 at Tier-2 : RESCEU (Tokyo), Niigata, +more future

——— iKAGRA data

2016/3/25-31 : lock duration 101.93 hours 2016/4/11-25 : lock duration 296.18 hours

~7.5 TB of raw data, 756 GB of proc data

5

slide-6
SLIDE 6

A bucket brigade of the data

6

HUB 10G

IFO, DSG frontend Environmental Monitor (EPICS layer)

IP address low latency hostname IP address frame writer k1fw1 IP address frame writer k1fw0 IP address NDS hostname IP address DetChar hostname HUB 1G monitor hosts in control room IP address data concentrator hostname 20TB 20TB IP address login / job man. taurus-01 IP address login / job man. taurus-02 disk array Infiniband SW IP address data transfer aldebaran IP address calculation pleiades-01 IP address calculation pleiades-02 IP address calculation pleiades-03 IP address calculation pleiades-04 IP address primary data server k1dm0 / hyades-0 IP address primary data server k1dm1 / hyades-1 HUB 10G HUB 10G / 1G

Dedicated optical fiber 4.5 km, tunnel <-> surface build.

IP address VPN gwave_kamioka Infiniband SW IP address MDS crab-mds-01 IP address MDS crab-mds-02 IP address OSS crab-oss-01 IP address OSS crab-oss-02 disk array MDT disk array OST disk array OST IP address VPN gwave_kashiwa HUB 10G / 1G IP address login server perseus-01 IP address login server perseus-02 Infiniband SW IP address MDS/OSS algol-01 disk array MDT/OST IP address ICRR interoperable computer system

SINET

iKAGRA data system overview

Drawn by N.Kanda last update : 2014/8/19

KAGRA Tunnel site Surface building at Kamioka ICRR, U Tokyo. (Kashiwa campus)

x2sets frame writer

  • -> data server

200 TiB luster file system mid spool of the data, HUB of data transfer Tier-0 archive iKAGRA :100TiB bKAGRA : 3PB

KAGRA’s VPN connection

slide-7
SLIDE 7

Data transfer

We developed the transfer software

We writing the code fully. (No black-box!) Using Linux standard tools as possible as.

Speed and Stability is good for bKAGRA requirement.

bKAGRA requirement : >20MB/s Performance : >40MB/s between Kamika and Kashiwa/Osaka

7

最頻値は以下の値となった( 幅:

転送遅延のヒストグラム

—> Performance test for 1sec data chunk from Kamioka to Osaka (by K.Sakai)

slide-8
SLIDE 8

Linux cluster(s) at Osaka City University

for the development of KAGRA analysis for low-latency search

8

# of cores
 Storage [TB] 200 400 600 800 Fiscal Year 2012 2013 2014 2015 2016

CPU [cores] (for calculation) Storage [TB] (/data, /home)

cluster system CPU/Storage

760 
 cores 304 TB ↑iKAGRA run

KAGRA’s VPN

to Osaka

slide-9
SLIDE 9

Tier-1 Mirroring

Academia SINICA (Taiwan)

started continuous mirroring at iKAGRA 2nd half with GRID computing technique.

KISTI (Korea)

archived iKAGRA data sets. We will establish automatic mirroring soon.

9

KAGRA DMG VPN
 (Kamioka-Kashiwa) Tier-0 archive ICRR VPN
 (Kashiwa) Prototype PC GRID server
 (Tier-1) Checksum verification SSH access over VPN GRID transfer ASGC (Taiwan) drawing by S.Haino

slide-10
SLIDE 10

KAGRA data systems(iKAGRA & New main storge)

10

New main storage (System-A) iKAGRA data storage

VPN

slide-11
SLIDE 11

New Peta-byte class system for bKAGRA era

Working since March 2017

2.4 PiB (HHD) for

  • bservational

data storage gpfs file system 12.8 TFLOPS

11

Storage: DDN SFA7700X + SS8460 Servers: HP ProLiant DL180 G9 HP ProLiant DL20 Gen9 HP ProLiant XL170 Gen9 Internal network: Infiniband FDR

slide-12
SLIDE 12

(working) VPN router KAGRA Private Network Internal Network (working) ICRR shared computer (working) iKAGRA data transfer and storage system

The internat (SINET)

Kamioka⇐

Kashiwa Campus

Hi-Speed Network Login Server Data Reciever Login Server (Disk Array) Mass Storage Data Sender

Compute Server Gateway Server

KAGRA Main Data Storage (System A)

Fire wall

Gateway Server Compute Server Compute Server Compute Server Compute Server Compute Server Compute Server Compute Server

2.4 PiB data storage 100 TiB for iKAGRA era

Extend the data transfer to new system

12

Extend in 2017 push==> (current) push==>

all data

spool ~100TiB

==> pull

partial data (h(t), latest raw) h(t)

slide-13
SLIDE 13

Current storages

13

Site Capacity Main Usage Kamioka (surface) 200 TiB spool, On-site analysis Kashiwa 100 TiB + 2.4 PiB iKAGRA data storeage Osaka City Univ. 304 TiB CBC, Burst, low latency search RESCEU 80 TiB CW Niigata Univ. 77 TiB

  • misc. analysis

Academia SINICA (Taiwan Group) 220 TiB Tier-1 mirroring, etc. KISTI-GSDC (Korea Group) 150TB (800TB in 2018 and 2019) Tier-1, Detector characterization (total) ~ 3.5 PiB

  • まとめ

( ) Site Capacity Main Usage ICRR computer for cooperative researches proc data, event analysis KEK computer center (not yet decided)

まとめ

We also cooperate with:

slide-14
SLIDE 14

Addendum

14

“M31” is system-A’s login servers Osaka Kashiwa Tokyo, RESCEU “KANBAI(寒梅)”

slide-15
SLIDE 15

Data Analysis

Target:

Search for Gravitational Waves in KAGRA data Extract Science of GWs Development of Tools for them (=Software, Computers, etc.)

Data analysis is one of cooperation channels between

  • ther observations:

LIGO (including Inidia), Virgo, GEO J-GEM (Japanese collaboration for Gravitational-wave Electro-Magnetic

follow-up)

Neutrino observations etc.

We have been cooperated with these partners in these several years.

15

slide-16
SLIDE 16

16

2 Chief:&H.Tagoshi SubJchiefs:&Y.Itoh,&H.Takahashi Core&members:&N.Kanda,&K.Oohara,&K.Hayama Korean&subgroup Leader:&Hyung Won&Lee

Osaka&City&Univ(13)&:&&N.Kanda,&H.&Tagoshi,&T.Narikawa,&T.&Yokozawa,& M.&Kaneyama,&N.&Uchikata,&S.&Tsuchida,& H.Yuzurihara,&T.&Yamamoto,&K.Tanaka, A.&Miyamoto,&T.&Kaji,&Y.&Kitaoka RESCEU(6)&&&&&&&&&&&&&:& Y.&Itoh,&K.&Eda,&J.&Yokoyama,&S.&Morisaki,&R.&Tsukada K.&Cannon ICRR&(1)&&&&&&&&&&&&&&&&&: K.Hayama Nagaoka&Tech&(4)&: H.Takahashi,&K.&Sakai,&Y.&Sasaki,&S.&Ueki Niigata&Univ (4)&&: K.Oohara,&Y.Hiranuma,&T.&Wakamatsu,&Y.&Watanabe Toyama&Univ (3)&&:& S.&Hirobayashi,&Dongbao Jia,&K.&Miyake NAOJ&&(3)&&&&&&&&&&&&&&:& N.&Ohishi,&A.&Shoda,&Y.&Fujii ISM&&&(1)&&&&&&&&&&&&&&&&: S.&Mano OIT&&&&(2)&&&&&&&&&&&&&&&&:&&H.&Shinkai,&S.&Yamamoto UWM&(1)&&&&&&&&&&&&&&:&&K.&Ueno CUHK(2)&&&&&&&&&&&&&&&:&&T.G.F&Li,&R.&Yeung Inje Univ.&:&Hyung Won&Lee Jeongcho Kim KASI:&&Chunglee Kim

Data&Analysis&Subsystem&(DAS)

2016/07/26

Total:&43&(Undergraduate&students&are&not&included)&

DAS viewgraphs are drawn by Tagoshi.

slide-17
SLIDE 17

17

Ongoing&iKAGRA analysis

3

Projects Members

CBC&offline

Yuzurihara, Tagoshi,&(Ueno)

CBCJPE&in&KAGALI

H.W.&Lee,&Jeongcho Kim, Chunglee Kim&

Burst

Hayama

CW

Eda,&Itoh

Radiometry&for&CW&

K.Tanaka,&Kanda,&Itoh

CBCJPE&on&injected signals

Narikawa, (Tagoshi)

HHT&on&injected&signals

Ueki, Takahashi,&Oohara,&Kanda,&Yokozawa

HW&injection&signals

Yokozawa,&……

Gaussianity,& CBC&nonJGaussian&triggers,…

Kitaoka,&Sasaki,&Kanda,&Tagoshi,&…

slide-18
SLIDE 18

Searches

CBC

Method: Frequency domain, matched filter - chi square analysis Mass range: 1-3Msun, no spin Template: TaylorF2 Template bank: Hexagonal template placement method

Burst

Search method : Excess power based method Event reconstruction :clustering based on mathematical morphology

Continuous

Continuous waves from 62 isolated known pulsars Analysis frequency range : 50 Hz – 1,000 Hz Search method : F-statistic

18

Horizon&distance&of&CBC&by&iKAGRA

6

10 0 10 1 10 2

Component mass [Msun]

10 0 10 1

[pc] Horizon distance of iKAGRA

ρ=8 fmin=5Hz m1=m2

(1.4,1.4)Msun =>3.2pc (10,10)Msun =>12.2pc

slide-19
SLIDE 19

19

Signal&Injection&test&@&iKAGRA

9

This&is&often&called&"Hardware&Injection&test". This&is&done&by&shaking&mirrors.& (The&signals&injected&are&not&real&gravitational&wave&signals!)&

Laser

Injected waveforms (Sine Gaussian)

○Sine Gaussian (test run) Q = 20(constant), f0 = [30, 100, 300, 1000]Hz, A = various XY differential, X/Y arm only(30,100Hz), 1count = 7.5e-13[strain] time[index] count count time[index] 100Hz SG 33Hz SG ○Sine Gaussian (injection run) Q = 20(constant), f0 = [33, 100, 333], A = 3000/f^2. 1500/f^2, XY differential

6

Injected waveforms (Supernova)

○Supernova waveform Provided Y.Suwa(YTIP, Kyoto), 2D Newtonian numerical simulation (KAGRA original) t=0 : Start time of gravity collapse Progenitor mass = 11.2M , different core rotation, matter eject memory effect 0.0 π rad/s 0.2 π rad/s 0.5 π rad/s 1.0 π rad/s No core rotation strong SASI motion(from entropy motion) 50M 80M 6Models(2s) × 3 amplitude scale 18 waveforms injection

7

Injected waveforms (CBC)

○CBC waveform(prepare waveforms by H.Yuzurihara, H.Tagoshi)

  • 1. Effective One Body (EOB) waveforms (generated by LAL)

mass combination : 20-20, 25-25, 30-30, 36.3-28.6 and 50-50M , no spin 3 amplitude(SNR) scales, low frequency cutoff : 10Hz To connect smoothly between 0 and waveform, we used error function

  • 2. NS-NS : Numerical Relativity simulation waveform+EOB waveform(by M.Shibata)

1.35-1.35 M , 2 amplitude(Scales) (KAGRA original) 20-20M “GW150914” 1.35-1.35M 36.3-28.6M⊙ Total 17 waveforms are injected

8

Hardware injection studies on

  • Parameter estimation in the case of CBC waveform
  • Checking the transfer function of the interferometer system
slide-20
SLIDE 20

20

KAGALI

KAGRA&Algorithmic&LIbrary current&version:&kagaliJv0r4a

  • Coding&style&guide&(Itoh)&
  • Git repository&at&RESCEU,&U.Tokyo (Itoh)

Nightly&build,&Doxygen for&documentation

  • Basic&design&(Oohara)

Autotools for&installation& Error&handing&mechanism& OpenBLAS,&Atlas&capability

  • Several&basic&components&(Ueno,&Yuzurihara,&Tagoshi,&…)

FFTW,&Frame,… KAGALIApps

  • CBC&pipeline&and&CBC&PE&pipeline

13

slide-21
SLIDE 21

21

KAGALI

14

  • Manuals
  • Coding&style&guide&(English/Japanese)&&&JGWJT1302019
  • KAGALI&Simple&Manual&(Japanese)&&&&&&&&&JGWJT1605584
  • Tips&on&C&Programming&(Japanese)&&&&&&&&&JGWJT1605585
  • KAGALI&Developer’s&Manual&(English/Japanese) JGWJT1605588

We&are&planning&to&revise&the&basic&structure&of&KAGALI,& and&to&revise&existing&code&which&are&not&wellJwritten&soon.&

slide-22
SLIDE 22

Summary

Data Management

We constructed and succeeded in iKAGRA

  • peration.

We are now extending / improving the system toward bKAGRA.

Data Analysis

We are developing our GW search pipelines. Some searches (CBC, Burst, CW) are executed with iKAGRA data. KAGALI

22