AMS 691Special Topics in Applied

Mathematics

James GlimmDepartment of Applied Mathematics

and Statistics,Stony Brook University

Brookhaven National Laboratory

0-3 credits• For 2-3 credits, a term paper is required.

– Pick any ongoing area of CAM research, determine what the research directions are, and describe current activities.

– Or pick any result unproven in this course, referred to some reference

• The course will survey ongoing CAM research– Guest lectures from other CAM faculty

• Introduction/survey of all CAM research areas– Some emphasis on turbulent combustion– Requires significant background material, which will

be surveyed and developed as we progress• Some details will be omitted, some will be summarized

CAM Research– Flows with complex geometry

• Mulitphase flows; interface between phases– Very complex if flow is turbulent– Professors Xaiolin Li, Xiangmin Jiao– Fluid structure interactions. Prof. Li, Jiao

– Flows with complex physics• Magnetohydrodunamics (MHD)

– Professor Roman Samulyak• ;Chemistry, combustion, chemical reactions

– James Glimm• Turbulent transport

– James Glimm• Radiation hydrodynamics – James Glimm• Phase transitions, material strength and fracture

– Professor Roman Samulyak• Coupling multiple physical models

– Climate studies Xiangmin Jiao, James Glimm, Roman Samulyak

• Porous media, data analysis for complex geometries– Brent Lindquist

CAM Research, continuedQuantum level modeling; atoms and electrons, Density functional theory

James GlimmMolecular dynamics, biological modeling Yuefan Deng, Biological modeling, Li, LindquistUncertainty quantification and QMU

Analysis of errors; assurance of accuracyVerification: is a numerical solution a valid approximation to the mathematical equationsValidation: are the mathematical equations a valid approximation to the physical problemUncertainty Quantification: estimate of errors from any and all sourcesQuantification of Margins and Uncertainties: numerically designed engineering safety margins for a numerically determined designJames Glimm

Computer Science IssuesJiao, Deng, Glimm, Li, Samulyak

Computational issues in FinanceJames Glimm, Xaiolin Li, Andrew Mullhaupt

CAM Research:Application Areas

• Design of laser fusion (JG); magnetically confined fusion (RS)• Design of new high energy accelerators (RS)• Turbulence, turbulent mixing, turbulent combustion (JG)• Modeling of Scramjet with uncertainty quantification, quantified

margins of uncertainty, verification and validation (JG)• Solar cell design (JG)• Modeling of windmills, parachutes (XL,XJ)• Brittle fracture (RS)• Chemical processing and nuclear power rod fuel separation (JG,XJ)• Flow in porous media; pollution control (XL,BL)• Short term weather forecasting for estimation/optimization of

solar/wind energy (JG)• Porous Media (BL)• Coupling atmosphere and oceans in climate studies (XJ)• Atmospheric modeling (RS,JG)• Compressible/incompressible flows with complex geometry and

physics (XJ)

AssignmentDue next week

Learn the main themes and ideas of the research of each of theCAM faculty.

Write a summary of this.

• Central Themes– Mathematical theory, physics modeling and high

performance computing– Computer science tools to enable effective computing– Problem specific subject matter– Required knowledge goes well beyond what is

possible to learn (over the course of your graduate studies), so as a student, you will learn the parts of these subjects that you need, for each specific problem/application. • Knowledge will be shared among graduate students, to

accelerate the learning process

First Unit: Equations of Fluid Dynamics

• In some sense, this lecture is an overview of your main courses for the next two years

• References: author = "A. Chorin and J. Marsden", title = "A Mathematical Introduction to Fluid Mechnics", publisher = "Springer Verlag", address = "New York--Heidelberg--Berlin", year = "2000",

author = "L. D. Landau and E. M. Lifshitz", title = "Fluid Mechanics", publisher = "Reed Educational and Professional Publishing Ltd", address = "London, England", year = "1987"

Nonlinear Hyperbolic Conservation Laws

1 1( )... ; ( ) ...

( )

( ) 0n n

t

U F UU F U

U F U

U F U

Total Quantity U is conserved

( , ) ( ( , )) 0D D

tR R

U x t dx F U x t dx

(assuming that U vanishes at infinity). Each component of U is conserved. Fundamental laws of classical physics are often of this form.For fluids, mass, momentum and energy arethe conserved quantities.

Simple case: Burgers’ Equationn = 1, D = 1

2

( , )( ) 0

1( )2

0

t

t x

u u x tf uux

f u u

u uu

Simpler case: f(u) = aulinear equation (a = const)

0

0

0

0( , ) ( )( ) ( , 0) is given data

t xu a uu x t u x atu x u x tu

Linear transport equation• Ut + aUx = 0• Solution is constant on lines x = x0 + at.• These lines are called characteristic

curves.• Each characteristic line meets initial line, t

= 0 at a unique point .• Thus solution is defined for all space time:

U(x,t) = U(x-at,0)• Initial discontinuities in U are preserved in

time, moving with velocity a.

Moving discontinuity for linear transport equation

Space time plot ofcharacteristic curves

Moving discontinuity,plotted u vs. x, moving intime

Simple Equation: Burgers’ Equation• Ut +(1/2) (U2)x = 0• Ut + U Ux = 0• U is a speed, the speed of propagation of information. • Characteristic curves: x = Ut +x0• U = constant on characteristic curve, thus determined by value at t =

0. Characteristic curves are straight lines in 1D space, and time. Thus solution can be written in closed form by a formula. – U(x,t) = U0(x-U0t)– U0(x) = initial data

• Increasing regions of U: characteristic curves spread out, solution becomes smoother.

• Decreasing regions of U: characteristic curves converge, solution develops steep gradients, discontinuity, and solution becomes multivalued.

Moving rarefaction wave for Burgers equation

Space time plot ofcharacteristic curves

Burgers equation and shock waves

• [q] = jump in q at discontinuity• s = speed of moving discontinuity• Burgers equation interpreted as a

distribution (weak form of equation) at a discontinuity– s[u] = [(1/2) u2]– Solve for s and get formula for solution, with

moving discontinuity (shock wave)– Extends solution after formation of

discontinuity

2

2 2

[ ] jump in quantity across discontinuity1[ ] = [ ] at moving discontinuity2

1[ ( )] ( ) [ ]2 2 2

2

a a a

s u u

u u u uu u u u u

u us

Weak Solution

2

2

2

2 2

1 all smooth 21=2

Choose = ( ) "pillbox"1= '2

1 1' ; ' [ ]; ' [ ]2 2

[ ]

t x

t x

u u dxdt

u u dxdt

x st

su u dxdt

u u u u

s u

Compression wave breaking into a shock wave for Burgers equation

Space time plot ofcharacteristic curves.curves meet at the line of discontinuity (a shockwave)

Compressible Fluid Dynamics Euler Equation (1D)

1 1

3 3

2

( )... ; ( ) ...

( )

( ) 0mass density; momentum density, = pressure;

1 + = total energy density; = internal energy2

t

U F UU m F U

U E F U

U F Um P

E mv e e

vF vv P

Ev vP

Equation of State (EOS)• System does not close. P = pressure is an extra unknown; e =

internal energy is defined in terms of E = total energy.• The equation of state takes any 2 thermodymanic variables and

writes all others as a function of these 2.• Rho, P, e, s = entropy, Gibbs free energy, Helmholtz free energy are

thermodynamic variables. For example we write P = P(rho,e) to define the equation of state.

• A simple EOS is the gamma-law EOS.

• Reference:• author = "R. Courant and K. Friedrichs",• title = "Supersonic Flow and Shock Waves",• publisher = "Springer-Verlag",• address = "New York",• year = "1967

Entropy

• Entropy = s(rho,e) is a thermodynamic variable. A fundamental principle of physics is the decrease of entropy with time.– Mathematicians and physicists use opposite

signs here. Confusing!

Analysis of Compressible Euler Equations

(2 ) (2 ) matrix

acoustic matrixGoverns small amplitude (linear) disturbancesEigenvalues and eigenvectors of known by exact formulae (for simpleequations of state), and these are used in s

F A D DUA

A

omemodern numerical schemes

Compressible Fluid Dynamics Euler Equation

• Three kinds of waves (1D)• Nonlinear acoustic (sound) type waves: Left or

right moving– Compressive (shocks); Expansive (rarefactions)– As in Burgers equation

• Linear contact waves (temperature, and, for fluid concentrations, for multi-species problems)– As in linear transport equation

Nonlinear Analysis of the Euler Equations

• Simplest problem is the Riemann problem in 1D• Assume piecewise constant initial state, constant for x <

0 and x > 0 with a jump discontinuity at x = 0.• The solution will have exactly three kinds of waves

(some may have zero strength): left and right moving “nonlinear acoustic” or “pressure” waves and a contact discontinuity (across which the temperature can be discontinuous)

• Exercise: prove this statement for small amplitude waves (linear waves), starting from the eigenvectors and eigenvalues for the acoustic matrix A

• Reference: Chorin Marsden