Hydrodynamic Collectivity in pp collisions Huichao Song Peking

Hydrodynamic Collectivity in pp collisions Huichao Song 宋慧超 Peking University Workshop on exploring the perfect liquid Munich, Sep 6 -8, 2018 Sep. 8, 2018

Collectively & Hydrodynamics in Pb+Pb collisions at the LHC

Fluctuations & collectivity in Pb+Pb collisions The initial state of Pb+Pb collisions fluctuate event-by-event, collective expansion of the system transform the initial state fluctuations to final state correlations associated with various flow observables

Hydrodynamics & Hybrid Model QGP Initial conditions Hadron Gas viscous hydro hadron cascade Conservation laws: - Israel-Stewart eqns. Input: “EOS” initial and final conditions

The Success of Hydrodynamics in Pb+Pb collisions (I) ALICE Paper JHEP 1609 164 (2016) -Hydro + IP-Glasma Gale, et. Al, PRL 2013 -i. EBE-VISHNU + AMPT Xu, Li, Song, PRC 2016 -hydrodynamics nice describe of integrated and differential Vn of all charged and identified hadrons

The Success of Hydrodynamics in Pb+Pb collisions (II) -Vn distributions -non-linear response coeff. -Correlations of Flow Harmonics -Hydrodynamics can quantitatively / qualitatively describe / predict various flow data

Correlations & Collectively in p+Pb collisions at 5 Te. V

Correlations & Flow in p- PLB 2013 -Many flow-like signals have been observed in high multiplicity p-Pb collisions

Flow like signals –theoretical interpolations Initial state effects: -K. Dusling and R. Venugopalan, PRL 2012, PRD 2013, NPA 2014 -A. Dumitru and A. V. Giannini, NPA 2015, A. Dumitru and V. Skokov PRD 2015 -B. Schenke, S. Schlichting, P. Tribedy, and R. Venugopalan, PRL 2016 … … Final state interactions: -P. Bozek, W. Broniowski, G. Torrieri, PRL 2013 -K. Werner, et. Al. , PRL 2014 G. -Y. Qin, B. Muller. PRC 2014 -Y. Zhou, X. Zhu, P. Li, and H. Song, PRC 2015 -P. Bozek, A. Bzdak, and G. -L. Ma, PLB 2015 -H. Li, L. He, Z. -W. Lin, D. Molnar, F. Wang, and W. Xie, 1604. 07387. … … Combinations of initial & Final state effects -H. M antysaari, B. Schenke, C. Shen, and P. Tribedy, PLB 2017

Flow signal in p-Pb -- Hydrodynamics Simulations in p+Pb collisions at 5. 02 Te. V G. -Y. Qin, B. Muller. PRC 2014 K. Werner, et. Al. , PRL 2014 P. Bozek, W. Broniowski, G. Torrieri, PRL 2013

Correlations & flow-like signals in p+p collisions at 13 Te. V

2 particle correlations in p+p collisions -Similar double ridge structure, but with smaller magnitudes in p-p collisions -measured v 2{2}, v 3{2} from 2 particle correlations [CMS Collaboration], Phys. Lett. B 765, 193 (2017). -clear v 2 mass ordering observed in experiments

Multi-particle correlations with standard method CMS [CMS Collaboration], Phys. Lett. B 765, 193 (2017) [ATLAS Collaboration], Phys. Rev. C 97, no. 2, 024904 (2018). Wenbin

Multi-particle correlations with 2 - & 3 -subevent ATLAS 3 subevent cumulant can largely suppress the nonflow effects. [ATLAS Collaboration], Phys. Rev. C 97, no. 2, 024904 (2018). Wenbin

Symmetric Cumulants ATLAS The ATLAS collaboration [ATLAS Collaboration], ATLAS-CONF-2018 -012.

Hydrodynamic Collectivity in p+p collisions at 13 Te. V

i. EBE-VISHNU hybrid model QGP Initial conditions Hadron Gas viscous hydro hadron cascade HIJING initial condition -produced jets pairs & excited nucleus independent strings break into partons form hot spots for succeeding hydro. 1)The center positions of strings (xc ; yc ) are sampled by Saxon-Woods distribution 2) positions of partons within the strings are sampled by 3) Energy decompositions of individual partons with a Gaussian smearing: Four parameter sets for the calculations W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018)

Spectra & 2 -particle correlation W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018) -In general, i. EBE-VISHNU + HIJING can describe the v 2{2}, v 3{2} and v 4{2}, from ATLAS and CMS. - Measured v 3{2} requires large initial state fluctuations of p-p collisions HIJING initial condition

Differential elliptic flow -i. EBE-VISHNU + HIJING intial conditions can describe the ATLAS and CMS well. -Clear v 2(p. T ) from v 2 mass ordering, as measured in experiment. W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018)

4 -particle correlations C 2{4} i. EBE-VISHNU + HIJING can not get the negative c 2{4}. W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018)

More details on C 2{4} calculations Minimize multiplicity fluctuations: (same method as used by ATLAS) ATLAS-CONF-2017 -002 Check standard, 2 -, 3 -subevent C 2{4} In i. EBE-VISHNU, no jets, non-flow mainly from resonance decays, standard method gives same results as 2 - and 3 - subevent methods. W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018)

C 2{4} & initial Ɛ distributions P(Ɛ ) 2 2 W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018)

-Cubic response: W. Zhao and H. Song private notes

Initial conditions for p+p collisions at 13 Te. V

Other initial conditions models HIJING, super-MC and TRENTo initial model neither can get negative c 2{4} HIJING: Zhao, Zhou, Xu, Deng, Song, Phys. Lett. B 780, 495(2018) Super-MC: Welsh, Singer, Heinz, Phys. Rev. C 94, no. 2, 024919 (2016) TRENTo: J. S. M ORELAND, Quark Matter 2018.

Some mock initial conditions

A Short Summary for hydrodynamic collectivity in pp collisions

-With tuned parameters, i. EBE-VISHNU with HIJING initial conditions. , can describe the measured 2 particle correlations v 2{2}, v 3{2} (for all charged and identified hadrons) W. Zhao, Y. Zhou, H. Xu, W. Deng and H. Song, Phys. Lett. B 780, 495 (2018) & paper in preparation -However, i. EBE-VISHNU with HIJING & other available initial conditions fails to describe the measured c 2{4} data -We still need better understanding on the initial conditions for pp collisions

Applications of Deep Learning to hydrodynamics Huang, Xiao, Xiong, Wu, Mu, Song, ar. Xiv: 1801. 03334 Exploring the perfect liquid

Applications of Deep Learning in Physics

Why Deep Learning in Physics? “Unlike earlier attempts … Deep Learning systems can see patterns and spot anomalies in data sets far larger and messier than human beings can cope with. ” Can “Black-box” models learn patterns and models solely from data without relying on scientific knowledge?

Image generation For hydrodynamics can we use deep learning to learn/predict the pattern transformation between initial and final profiles? Initial energy density profiles ---- > final energy density velocity profiles For the non-linear hydro system, can the black-box network could learn pattern transformations solely from data without relying on scientific knowledge? ( conservation laws)

Deep Learning Step 1）Generate the training/testing data sets from hydro Step 2）Design & train the deep neural network The Training Data Sets hydro VISH 2+1 MC-Gl 10000 Step 3）Test the deep neural network The Testing Data Sets hydro MC-Gl MC-KLN VISH 2+1 10000 AMPT 10000 Trento 10000

s. Unet prediction vs. hydro simulations

s. Unet prediction vs. hydro simulations -for a closer look

s. Unet prediction vs. hydro simulations Eccentricity distributions:

Summary & outlook

Traditional hydrodynamics Deep Learning More to explore, have fun！

Thank You

Multi-particle correlations with 2 - & 3 -subevent ATLAS 3 subevent cumulant can further suppress the nonflow effects. Clear splitting between v 2{2} & v 2{4} [ATLAS Collaboration], Phys. Rev. C 97, no. 2, 024904 (2018). Wenb

W. Zhao and H. Song private notes