Post on 02-Jan-2017
Forefront Questions in Nuclear Science and the Role of High Performance Computing
January 26-28, 2009 · Washington D.C.
Major Issues in Nuclear Physics Aided by Massive Computation
David B. Kaplan ~ Institute for Nuclear Theory
The challenge of nuclear theory
• Many-body problem of interaction nucleons• Quantum mechanical• Strongly interacting
pre- 1970s ~ phenomenological models:
Predictive phenomenology; qualitative theoretical understanding
The challenge of nuclear theory
• Many-body problem of interaction nucleons• Quantum mechanical• Strongly interacting
pre- 1970s ~ phenomenological models:
Predictive phenomenology; qualitative theoretical understanding
• A simply expressed, complete theory of the strong interactions• Nucleons are bound state of quarks & gluons• Quantum mechanical• Strongly interacting• Relativistic
post-1970s ~ Quantum Chromodynamics (QCD):
1980’s, 1990’s: a ferment of ideas
Exchange of ideas with particle theory:• Effective field theory• String theory• Symmetries
Experiments:• Precision symmetry tests• SN 1987A• Neutrino masses• RHIC, JLab• Trapped atoms...
Exchange of ideas withcondensed matter & atomic theory:• Quantum chaos• Density functional theory• Color superconductivity of quark matter
5th ANL/MSU/JINA/INT FRIB Workshop: Bulk Nuclear Properties Nov. 19 - 22, 2008
Workshop on Relativistic Dynamics of Graphene Jan. 8 - 11, 2008
Nuclear Interactions at Ultra-high Energy in Light of Recent Results from Auger Feb. 20 - 22, 2008
From Strings to Things: String Theory Methods in QCD and Hadron Physics March 24 - June 6, 2008
Soft Photons and Light NucleiJune 16-20, 2008
The QCD Critical PointJuly 28 - Aug. 22, 2008
Low Energy Precision Electroweak Physics in the LHC EraSept. 22 - Dec. 5, 2008
Atomic, Chemical, and Nuclear Developments in Coupled Cluster Methods June 23-July 25, 2008
The diversity of nuclear theory: INT programs 2008
2000’s: Computational nuclear physics comes of age
• Lattice QCD• Advances in the shell model• Density functional methods• Coupled cluster techniques...
2000’s: Computational nuclear physics comes of age
• Lattice QCD• Advances in the shell model• Density functional methods• Coupled cluster techniques...
Experience from TFlops-years: a glimpse of a unified nuclear theory grounded in QCD, aided by extreme computing
2000’s: Computational nuclear physics comes of age
Structure: from quarks and gluons to heavy nuclei
• Lattice QCD• Advances in the shell model• Density functional methods• Coupled cluster techniques...
Experience from TFlops-years: a glimpse of a unified nuclear theory grounded in QCD, aided by extreme computing
2000’s: Computational nuclear physics comes of age
Structure: from quarks and gluons to heavy nuclei
Dynamics: evolution of the quark gluon plasma in a heavy ion collision; fusion and fission processes which power stars and create elements; the cataclysmic explosions of stars
• Lattice QCD• Advances in the shell model• Density functional methods• Coupled cluster techniques...
Experience from TFlops-years: a glimpse of a unified nuclear theory grounded in QCD, aided by extreme computing
Origins
Fusio
nMeta
ls
Supe
rnova
Collapse
The extreme computing opportunities and challenges for nuclear physics all come together in describing the life of a star
Origins
Nucleons, resonances
Quark-gluon plasma
Matter/anti-matter annihilation
Exascale challenges:
QCD phase diagram
Structure of the nucleon & resonances
Clues about origin of matter/anti-matter asymmetry
Origins
Phase diagrams: Water...
...primarily explored in the laboratory
Origins
Phase diagrams: QCD (an educated guess!)...
...to be explored by accelerator, telescope & computer
Origins
Phase diagrams: QCD (an educated guess!)...
...to be explored by accelerator, telescope & computer
Observation
Origins
Phase diagrams: QCD (an educated guess!)...
...to be explored by accelerator, telescope & computer
Observation
Origins
Experiment &EXASCALE
Computation
Phase diagrams: QCD (an educated guess!)...
...to be explored by accelerator, telescope & computer
Observation
EXASCALE Computation
Origins
Experiment &EXASCALE
Computation
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Expansion and hadronization
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Expansion and hadronization
Relativistic hydrodynamics
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Expansion and hadronization
Relativistic hydrodynamics
Departure from thermal equilibrium
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Expansion and hadronization
Relativistic hydrodynamics
Departure from thermal equilibrium
Predicting observables
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Expansion and hadronization
Relativistic hydrodynamics
Departure from thermal equilibrium
Predicting observables
• Important physics• Ambitious computation
EXASCALE?
Origins
Recreating the Big Bang in heavy ion collisions
Experimental program: RHIC, LHCTheory challenge: a unified description of a heavy ion collisions
Initial conditions: quark/gluon distribution in ions
Thermalization; quark-gluon plasma
Expansion and hadronization
Relativistic hydrodynamics
Departure from thermal equilibrium
Predicting observables
• Important physics• Ambitious computation
Origins
Quark & gluon structure of nucleons and resonancesfrom lattice QCD
Experimental program: JLabTheory challenges:
Origins
Quark & gluon structure of nucleons and resonancesfrom lattice QCD
• What is the momentum distribution of quarks and gluons within the nucleon?
Experimental program: JLabTheory challenges:
Origins
Quark & gluon structure of nucleons and resonancesfrom lattice QCD
• What is the momentum distribution of quarks and gluons within the nucleon?
Structure functions, strange quark content
Experimental program: JLabTheory challenges:
Origins
Quark & gluon structure of nucleons and resonancesfrom lattice QCD
• What is the momentum distribution of quarks and gluons within the nucleon?
Structure functions, strange quark content
• How do currents couple to the nucleon?
Experimental program: JLabTheory challenges:
Origins
Quark & gluon structure of nucleons and resonancesfrom lattice QCD
• What is the momentum distribution of quarks and gluons within the nucleon?
Structure functions, strange quark content
• How do currents couple to the nucleon? High precision form factors (gA to <1%?)
Experimental program: JLabTheory challenges:
Origins
Quark & gluon structure of nucleons and resonancesfrom lattice QCD
• What is the momentum distribution of quarks and gluons within the nucleon?
Structure functions, strange quark content
• How do currents couple to the nucleon? High precision form factors (gA to <1%?)
• Predict exotic states (glueballs, hybrid mesons)
Experimental program: JLabTheory challenges:
Big Bang:Why is there more matter than anti-matter today?
A key ingredient: violation of baryon number or lepton number symmetryOrigins
Big Bang:Why is there more matter than anti-matter today?
A key ingredient: violation of baryon number or lepton number symmetryOrigins
Neutrino-less double beta decay = lepton violation
• Subject of intense experimental search • Requires calculation of nuclear matrix
element
Experimental program: Majorana, Cuore, EXO, NEMO3...Theory challenges:
Big Bang:Why is there more matter than anti-matter today?
A key ingredient: violation of baryon number or lepton number symmetry
The structure of 76Ge, 150Nd and calculation of relevant matrix element is an Exascale problem
EXASCALE
Origins
Neutrino-less double beta decay = lepton violation
• Subject of intense experimental search • Requires calculation of nuclear matrix
element
Experimental program: Majorana, Cuore, EXO, NEMO3...Theory challenges:
Fusio
n
Extreme computing: the era for understanding and predicting nuclear reaction rates
Solar fusion:Exascale era: solar fusion rates can be obtained directly from QCD
Fusion
Technique: “Lattice QCD”• Quarks & gluons interacting on a
spacetime lattice.
• Highly parallelizable.
Fusion
Technique: “Lattice QCD”• Quarks & gluons interacting on a
spacetime lattice.
• Highly parallelizable.
• High precision nucleon form factors (e.g. gA to <1%?)
• High precision 2-body forces• Calculation of 3- and 4-body forces
directly from QCDeg: energy levels of four neutrons in a box: not accessible experimentally
e-
Fusion
Technique: “Lattice QCD”• Quarks & gluons interacting on a
spacetime lattice.
• Highly parallelizable.
EXASCALE
• High precision nucleon form factors (e.g. gA to <1%?)
• High precision 2-body forces• Calculation of 3- and 4-body forces
directly from QCDeg: energy levels of four neutrons in a box: not accessible experimentally
e-
Fusion
Metals
Metals
3- and 4-nucleon interactions from lattice QCD
Ab initio nuclear structure & reaction calculations
• GFMC• No-core shell model...
The output from one class of extreme computations feeds into the next:
EXASCALE
Metals
Metals
Triple alpha process
O16
Alpha capture on p-shell nuclei
Metals
Triple alpha process
O16
Alpha capture on p-shell nuclei
EXASCALE
Metals
Triple alpha process
Supe
rnova
Super-nova
Cataclysmic events in the heavens
Type I SN: thermonuclear explosion; key distance marker for astronomers
Type II SN: core collapse; origin of heavy elements, neutron stars. Neutrinos play a leading role.
X-ray bursters, colliding neutron stars...
Super-nova
EXASCALE nuclear astrophysics:
• Realistic 3D simulations of core-collapse SN; expect to uncover robust explosion mechanism
• 3D simulations of Type I SN including turbulence effects
• Pulsar & magnetar simulations
Super-nova
Type II Supernovae are the site for synthesis of heavy elements (“r-process”)
Super-nova
Type II Supernovae are the site for synthesis of heavy elements (“r-process”)
Density functional theory:• Compute shell structure along r-process path• Compute shell structure around 100Sn• Provide microphysical theory of fission
Super-nova
Type II Supernovae are the site for synthesis of heavy elements (“r-process”)
Density functional theory:• Compute shell structure along r-process path• Compute shell structure around 100Sn• Provide microphysical theory of fission
FRIB:• Experimental studies of
neutron-rich isotopes
Super-nova
Type II Supernovae are the site for synthesis of heavy elements (“r-process”)
Supernova simulations:• Predict element
abundances• Predict terrestrial
neutrino flux
Density functional theory:• Compute shell structure along r-process path• Compute shell structure around 100Sn• Provide microphysical theory of fission
FRIB:• Experimental studies of
neutron-rich isotopes
Super-nova
Type II Supernovae are the site for synthesis of heavy elements (“r-process”)
Supernova simulations:• Predict element
abundances• Predict terrestrial
neutrino flux
Density functional theory:• Compute shell structure along r-process path• Compute shell structure around 100Sn• Provide microphysical theory of fission
EXASCALE
FRIB:• Experimental studies of
neutron-rich isotopes
Super-nova
Dynamics from density functional theory:
• Evolution of fissioning nuclei from a microphysical Hamiltonian
• Prediction for distribution of fission products
Collapse
One of the biggest challenges: extended matter at nuclear densities and higher, from QCDCollapse
One of the biggest challenges: extended matter at nuclear densities and higher, from QCDCollapse
One of the biggest challenges: extended matter at nuclear densities and higher, from QCDCollapse
Present approaches: nuclear matter from QCD is an exponentially hard problem...NOT Exascale-ready
EXASCALE
However: an important piece of the problem can be addressed
Exotic phases at high density all involve the strange quark
Collapse
Lattice QCD: precision studies of nucleon-hyperon interactions
However: an important piece of the problem can be addressed
Exotic phases at high density all involve the strange quark
Collapse
Lattice QCD: precision studies of nucleon-hyperon interactions
EXASCALE!
Will we be ready?
What does it take?
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: hydrodynamics of heavy ion collisions; lattice QCD determination of critical point
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: hydrodynamics of heavy ion collisions; lattice QCD determination of critical pointStructure: time-dependent density functional studies of fission, reactions
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: hydrodynamics of heavy ion collisions; lattice QCD determination of critical pointStructure: time-dependent density functional studies of fission, reactions
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: lattice studies of nucleon properties, nucleon-meson interactions, multi-nucleon states & matrix elements, hot quark-gluon plasma, hadronic excited states, hyperons;
QCD: hydrodynamics of heavy ion collisions; lattice QCD determination of critical pointStructure: time-dependent density functional studies of fission, reactions
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: lattice studies of nucleon properties, nucleon-meson interactions, multi-nucleon states & matrix elements, hot quark-gluon plasma, hadronic excited states, hyperons;Astro: Major advances in supernova simulations
QCD: hydrodynamics of heavy ion collisions; lattice QCD determination of critical pointStructure: time-dependent density functional studies of fission, reactions
QCD: lattice studies of cold matter & nuclei with A>5
• Can the physics problem productively consume a PFlops-year today?
• Is significant progress expected from an ExaScale machine?
• Does significant progress require an ExaScale machine?
QCD: lattice studies of nucleon properties, nucleon-meson interactions, multi-nucleon states & matrix elements, hot quark-gluon plasma, hadronic excited states, hyperons;Astro: Major advances in supernova simulationsNuclear structure: Major advances in nuclear structure calculations using GFMC, no-core shell model, DFT
QCD: hydrodynamics of heavy ion collisions; lattice QCD determination of critical pointStructure: time-dependent density functional studies of fission, reactions
QCD: lattice studies of cold matter & nuclei with A>5
What nuclear theory challenges could efficiently use an Exascale computer today?
What nuclear theory challenges could efficiently use an Exascale computer today?
None.
To realize extreme computing in nuclear physics requires minds as well as machines
What nuclear theory challenges could efficiently use an Exascale computer today?
None.
To realize extreme computing in nuclear physics requires minds as well as machines
What nuclear theory challenges could efficiently use an Exascale computer today?
None.
• Theory: effective field theory techniques, algorithm development, new formulations of old problems, improved Hamiltonians...
• Computer science: algorithm development, exploiting multi-core chips, parallel architecture, stability with imperfect hardware...
GETTING READY FOR EXASCALE
• Programs such as SciDAC (DOE) and CDI (NSF) have a big role to play
GETTING READY FOR EXASCALE
• Programs such as SciDAC (DOE) and CDI (NSF) have a big role to play
•Access to time on machines of varying sizes is critical: local TFlops clusters to PFlops national class platforms
GETTING READY FOR EXASCALE
• Programs such as SciDAC (DOE) and CDI (NSF) have a big role to play
•Access to time on machines of varying sizes is critical: local TFlops clusters to PFlops national class platforms
•Topical collaborations (2007 Long Range Plan recommendation) would be invaluable for computational nuclear theory
GETTING READY FOR EXASCALE
Recap
WITH - AND ONLY WITH - EXASCALE:
Recap
WITH - AND ONLY WITH - EXASCALE:
•A unified nuclear theory, from QCD to heavy nuclei
Recap
WITH - AND ONLY WITH - EXASCALE:
•A unified nuclear theory, from QCD to heavy nuclei
•Nucleon and nuclear structure becomes a precision science, complementing major experimental programs (~1% errors)
Recap
WITH - AND ONLY WITH - EXASCALE:
•A unified nuclear theory, from QCD to heavy nuclei
•Nucleon and nuclear structure becomes a precision science, complementing major experimental programs (~1% errors)
•A new era for the simulation & prediction of nuclear dynamics and reactions from fundamental interactions (fission, fusion, stellar explosions, heavy ion collisions..)