The NASA HPCC ESS Cooperative AgreementThe NASA HPCC ESS Cooperative Agreement

PI: Andrea MalagoliPI: Andrea Malagoli

University of Chicago/ANL University of Colorado University of Minnesota

Co-InvestigatorsNic Brummell

Fausto CattaneoTom Clune

Anshu DubeyBill Gropp

Rusty LuskDavid Porter

Robert RosnerRick StevensJuri Toomre

Paul Woodward

Turbulent Convection and Dynamos in StarsTurbulent Convection and Dynamos in Stars

NASA/GSFC

MHD-PPM Convection (MHD-PPMC): 195.8 GFlops• The Piecewise Parabolic Method coupled with a parabolic solver.• Solves the equations of compressible (magneto)hydrodynamics (MHD) with radiative diffusion.

Hybrid PseudoSpectral (HPS): 167.0 GFlops• Uses hybrid finite-difference and pseudospectral spatial derivatives

and a second order time stepping method.• Solves the equations of compressible (magneto)hydrodynamics (MHD) in a rotating plane.

MHD PseudoSpectral (MPS): 160.0 GFlops• Uses a fully pseudospectral method with Runge-Kutta time stepping.• Solves the equations of incompressible (magneto)hydrodynamics (MHD) in a triply periodic domain.

Turbulent Convection and Dynamos in Stars: Performance Milestones[PI: Andrea Malagoli - University of Chicago]

All our Scientific Application codes have achieved the 100 GigaFlops performance milestone on the Cray

T3E-1200 testbed at NASA/GSFC

Observations data obtained by current NASA missionsreveal that complex, highly time dependent phenomena are taking place on the surface of the Sun.

These phenomena are related to the vigorous interaction between turbulence, rotation and magnetic fields in the Sun’s Interior.

Simulation-based models on NASA’s High Performance Testbeds are used to interpret observations data and to build detailed models of the Sun’s interior.

These models play a fundamental role in both the interpretation of current missions data and in driving the design of all future missions

The 11-years Cycle

Magnetograms

H images

Model of the Sun’s Interior

Prominences

HelioSeismology

Granulation and Sunspots

X-ray images

The NASA HPCC ESS CAN Turbulent Convection and Dynamos in Stars

PI: Andrea MalagoliFrom Missions Data to Modeling the SunFrom Missions Data to Modeling the Sun

YOHKOH SOLAR BSOLAR PROBETRACESOHO

Turbulent Convection and Dynamos in Stars: Simulations[PI: Andrea Malagoli - University of Chicago]

Convectively driven solar dynamo simulations have been carried out using our Milestones codes on NASA’s T3E testbed.

AUTHORS: Fausto Cattaneo and Anshu Dubey - UofC

MACHINE: Cray T3E-1200 at NASA/GSFC

CODE: MHD PseudoSpectral

RESOLUTION: 512x512x97

Temperature

Temperature fluctuations near the upper boundary in a numerical simulation of convectively driven dynamo action. Dark tones correspond to cold (downflowing) material

Vertical magnetic field

Vertical component of the magnetic field near the upper boundary. Dark and light tones correspond to fields of opposite polarity

High resolution simulations are used to study the origin of magnetic fields in the quiet photosphere of the Sun

Images from the simulations can be compared with observational images

The model suggests that small flux elements are generated locally by dynamo action associated with the granular and supergranular flows

Turbulent Convection and Dynamos in Stars: Simulations[PI: Andrea Malagoli - University of Chicago]

AUTHORS: Fausto Cattaneo and Anshu Dubey

MACHINE: Cray T3E-1200 at NASA/GSFC

CODE: MHD PseudoSpectral

RESOLUTION: 512x512x97

Original Coarse grained

Detail of bipolar magnetic element. The figure shows an area of 200x200 pixels.

Same image convoluted with a Gaussian filter with 12 pixels FWHM.

Simulation data have higher resolution than observational data. Artificial blurring can be used to compare and interpret observational data

Simulations by F. Cattaneo & A. Dubey

Numerical simulations of convectively driven dynamos

orig

inal

coar

se g

rain

ed

• Magnetic elements generated by dynamo action display considerable structural complexity• Limited resolution can lead misleading interpretation of true structure of magnetic elements in terms of oversimplified flux tube models• Numerical simulation can help the development of more realistic models and the design of high resolution instruments for the detection of small-scale magnetic elements

Detail of bipolar magnetic element. The figure shows an area of 200x200 pixels.

Same image convoluted with a Gaussian filter with 12 pixels FWHM. Even a modest amount of coarse graining leads to substantial loss of information.