Hamilton description of plasmas and other models of matter: structure and applications I P. J. Morrison Department of Physics and Institute for Fusion Studies The University of Texas at Austin [email protected] http://www.ph.utexas.edu/ morrison/ ∼ MSRI August 20, 2018 Survey Hamiltonian systems that describe matter: particles, fluids, plasmas, e.g., magnetofluids, kinetic theories, . Hamilton description of plasmas and other models of matter: structure and applications I P. J. Morrison Department of Physics and Institute for Fusion Studies The University of Texas at Austin [email protected] http://www.ph.utexas.edu/ morrison/ ∼ MSRI August 20, 2018 Survey Hamiltonian systems that describe matter: particles, fluids, plasmas, e.g., magnetofluids, kinetic theories, . \Hamiltonian systems .... are the basis of physics." M. Gutzwiller Coarse Outline William Rowan Hamilton (August 4, 1805 - September 2, 1865) I. Today: Finite-dimensional systems. Particles etc. ODEs II. Tomorrow: Infinite-dimensional systems. Hamiltonian field theories. PDEs Why Hamiltonian? Beauty, Teleology, . : Still a good reason! • 20th Century framework for physics: Fluids, Plasmas, etc. too. • Symmetries and Conservation Laws: energy-momentum . • Generality: do one problem do all. • ) Approximation: perturbation theory, averaging, . 1 function. • Stability: built-in principle, Lagrange-Dirichlet, δW ,.... • Beacon: -dim KAM theorem? Krein with Cont. Spec.? • 9 1 Numerical Methods: structure preserving algorithms: • symplectic, conservative, Poisson integrators, . e.g. GEMPIC. Statistical Mechanics: energy, measure . e.g. absolute equil. • Today Natural Hamiltonian systems • \Unnatural" Hamiltonian systems • Noncanonical Hamiltonian systems • Action Principle Hero of Alexandria (60 AD) Fermat (1600's) −! −! Hamilton's Principle (1800's) The Procedure: Configuration Space/Manifold Q: qi(t) ; i = 1; 2;:::;N #DOF • Lagrangian (Kinetic Potential): L = T V L : TQ R • − ! Action Functional: • Z t1 S[q] = L(q; q_) dt ; δq(t0) = δq(t1) = 0 t0 Extremal path Lagrange's equations ) Variation Over Paths S[qpath] = number First Variation (Fr´echetderivative): d δS[q; δq] = DS δq = S[q + δq] 0 δq(t) · d =0 ≡ 8 ) Lagrange's Equations: @L d @L = 0 : @qi − dt@q_i Hamilton's Equations Canonical Momentum: p = @L inverse function theorem i @q_i Legendre Transform: H(q; p) = p q_i L(_q; q) i − @H @H p_ = ; q_i = ; i i −@q @pi Phase Space Coordinates: z = (q; p) ; α, β = 1; 2;::: 2N ! @H 0 I z_α = Jαβ = zα;H ; (Jαβ) = N N ; c @zβ f g c I 0 − N N Jc := Poisson tensor, Hamiltonian bi-vector, cosymplectic form 1 c βγ γ symplectic 2-form = (cosymplectic form)− : !αβJc = δα, Natural Hamiltonian Systems Natural Hamiltonian Systems Natural Hamiltonian: H(q; p) = T (q; p) + V (q) Kinetic Energy: 1 X 1 T (q; p) = mij− (q) pi pj 2 i;j where mij pos. def. mass matrix (metric tensor). Potential energy: V (q) = V (q1; q2; : : : ; qN ) Equations of motion: X 1 @V q_i = mij− pj and_ pi = j −@qi for mij constant. Natural Hamiltonian Examples Mass spring systems, pendula, particle in potential well, etc. • 3 N-Body problem qi = (qxi; qyi; qzi) Q R ; i = 1; 2;:::N • 2 ⊂ N 2 N X p X cij H = jj ijj + 2m q q i=1 i i;j=1 jj i − jjj where depending on sign cij it represents attracting gravitational interaction (satellites, planets, stars, ...), repelling electrostatic interaction (electrons), attracting electrons and ions (protons), collection of both in plasmas. \Unnatural" Hamiltonian Systems \Unnatural" := Natural Hamiltonian Systems : Charged particle in given electromagnetic fields: • e mq¨ = eE(q; t) + q_ B(q; t) c × where E; B electric and magnetic fields, respectively, e charge, m mass, c speed of light. Potentials: 1@A B = AE = φ r × −∇ − c @t Hamiltonian: 2 1 e H(q; p; t) = p A(q; t) + eφ(q; t) 2m − c Other Unnatural Hamiltonian Systems Interaction of point vortices in the plane • N X 2 2 H = c κiκj ln (xi xj) + (yi yj) ij=1 − − Chaotic advection in two dimensions • H = (x; y; t) ; v = 0 v(x; y; t) = (@ =@y; @ =@x) ; r · ! − neutrally buoyant particle or dye moves with fluid. Stream func- tion . Magnetic field line flow (integral curves of B(x)) • Other: preditor-prey, etc. • Chaotic Advection H. Swinney Lab circa 1989. Nontwist! del-Castillo-Negrete et al. Cyclonic (eastward) jet particle streaks dye Particle in B-Field Equation of motion: • e mq¨ = q_ B(q; t) c × Solution for B uniform: • Particle in B-Field Equation of motion: • e mq¨ = q_ B(q; t) c × Solution for B uniform: • Particle in B-Field Equation of motion: • e mq¨ = q_ B(q; t) c × Solution for B uniform: • gyroradius: ρg = mcv =(eB). gyrofrequency:Ω g = eB=(mc) ? B-lines as Hamiltonian system If particles are tied to field lines natural to look at field line flow. ) If interested in confinement, then if field lines escape particle will. Use fields for confinement: Stellerator, Tokamak, etc. Also related are particle accelerators. Field line flow is Hamiltonian: Kruskal, Kerst, I. M. Gelfand, Mo- rozov & Solov'ev, ... Cary and Littlejohn 1983 Tokamak Fields Superposition of dipole + toroid Tokamak ≈ B-line Hamitonian for Straight Torus For simplicity remove metrical components and consider topolog- ical torus with cylindrical coords( r; θ; z) with z `toroidal' angle (long way aorund) θ poloidal angle (short way around): B = BT z^ +z ^ (r; θ; z) BT = const >> B 1 way × r ? ) assures B = 0. r · Integral curves of B(x): dR = B dσ parametrize by z ) dX @ dY @ = Bx = and = By = dz −@Y dz @X (X; Y; z) is the Hamiltonian. General system is 1.5 dof but inte- grable if @ =@z 0 (becomes 1 dof) desired equilibrium state. ≡ Surface of Section Symmetry breaking k ⇔ ↑ Early Symplectic Map Early Symplectic Map (cont) George Miloshevich $250 Universal Symplectic Maps of Dimension Two Standard (Twist) Map: x0 = x + y0 k y0 = y sin(2πx) − 2π Standard Nontwist Map: x = x + a(1 y 2) 0 − 0 y = y b sin(2πx) 0 − Parameters: a measures shear, while b and k measure ripple Drifts: E and B Not Uniform 9 Even for large B, particles don't follow field lines because of drifts. Drift Types: E B, B, curvature, polarization. × r Example B (recall ρg 1=B): r ∼ Reductions base on magnetic moment µ = m v 2=(2B) being an j j adiabatic invariant. ? Drifts: E and B Not Uniform (cont) 9 Drifts: E and B Not Uniform (cont) 9 Drifts: E and B Not Uniform (cont2) 9 Drifts - E and B Not Uniform (cont2) 9 Guiding center equations for transport. Hannes Alfv´en. Hamiltonian system in noncanonical coordinates ! Noncanonical Hamiltonian systems. ⊂ Kinetic theories on guiding centers etc. Drift kinetics and Gyrokinetics. ! Noncanonical Hamiltonian Systems Usual Geometry Dynamics takes place in phase space, (needn't be T Q), a Z ∗ differential manifold endowed with a closed, nondegenerate 2-form !. A patch has canonical coordinates z = (q; p). Hamiltonian dynamics flow on symplectic manifold: i ! = dH , X Poisson tensor( Jc) is bivector inverse of !, defining the Poisson bracket @f αβ @g f; g = df; Jc(dg) = !(X ;Xg) = J ; α, β = 1; 2;::: 2N f g h i f @zα c @zβ Flows generated by Hamiltonian vector fields ZH = JdH, H a 0- form, dH a 1-form. Poisson bracket = commutator of Hamiltonian vector fields etc. Early refs.: Jost, Mackey, Souriau, Arnold, Abraham &Marsden Noncanonical Hamiltonian Definition A phase space diff. manifold with binary bracket operation P on C ( ) functions f; g : R, s.t. ; : C ( ) C ( ) 1 P P! f · · g 1 P × 1 P ! C ( ) satisfies 1 P Bilinear: f + λg; h = f; h + λ g; h ; f; g; h and λ R • f g f g f g 8 2 Antisymmetric: f; g = g; f ; f; g • f g −{ g 8 Jacobi: f; g; h + g; h; f + h; F; g 0 ; f; g; h • f f gg f f gg f f gg ≡ 8 Leibniz: fg; h = f g; h + f; h g ; f; g; h. • f g f g f g 8 Above is a Lie algebra realization on functions. Take fg to be pointwise multiplication. Eqs. Motion: @Ψ = Ψ;H ,Ψ an observable & H a Hamiltonian. @t f g Example: flows on Poisson manifolds, e.g. Weinstein 1983 .... Noncanonical Hamiltonian Dynamics Sophus Lie (1890) Noncanonical Coordinates: @H @f @g z_α = Jαβ = zα;H ; f; g = Jαβ(z) ; α, β = 1; 2;:::M @zβ f g f g @zα @zβ Poisson Bracket Properties: antisymmetry f; g = g; f ; −!f g −{ g Jacobi identity f; g; h + g; h; f + h; f; g = 0 −!f f gg f f gg f f gg G. Darboux: detJ = 0= J Jc Canonical Coordinates 6 ) ! Sophus Lie: detJ = 0= Canonical Coordinates plus Casimirs ) 0 1 0N IN 0 J J = B I 0 0 C : ! d @ − N N A 0 0 0M 2N − Flow on Poisson Manifold Definition. A Poisson manifold is differentiable manifold with M bracket ; : C ( ) C ( ) C ( ) st C ( ) with ; f g 1 M × 1 M ! 1 M 1 M f g is a Lie algebra realization, i.e., is i) bilinear, ii) antisymmetric, iii) Jacobi, and iv) consider only Leibniz, i.e., acts as a derivation. Flows are integral curves of noncanonical Hamiltonian vector fields, ZH = JdH. Because of degeneracy, functions C st f; C = 0 for all f 9 f g 2 C ( ). Called Casimir invariants (Lie's distinguished functions.) 1 M Poisson Manifold Cartoon M Degeneracy in J Casimirs: ) f; C = 0 f : R f g 8 M! Lie-Darboux Foliation by Casimir (symplectic) leaves: Leaf vector fields, Z = z; f = Jdf are tangent to leaves. f f g Lie-Poisson Brackets Coordinates: αβ αβ γ J = cγ z αβ where cγ are the structure constants for some Lie algebra.