Sum rule and Giant Resonances Sum rule and giant resonances Giant resonances are typical collective modes of excitation at high energy, and exhaust major portion of the sum rule Isoscalar (T=0) Monopole (0+) Quadrupole (2+) Octupole (3−) Isovector (T=1) Monopole (0+) Dipole (1 −) Quadrupole (2+) Gamow-Teller (1+) Spin-monopole (1+) Spin-dipole (0 −, 1 −, 2 −) f λ (r )Yλ µ τ i f λ (r )Yλ µ τ iσ j f 0 ( r ) τ i f 0 ( r )( σ × Y1 ) Iµ Sum rule m1 = ∑ En n F 0 ∑ Enp n F 0 2 ⇒ (Energy - weighted) Sum Rule n mp = 2 n Odd-p moments can be expressed by the ground-state expectation value. m1 = ∑ n mp = ∑ n If Fˆ = ∑ 1 En n F 0 = 0 [ F , [ H , F ]] 0 2 2 1 p En n F 0 = 0 [[ F , H ], [ H , [ H , F ]]] 0 2 2 f (ri ) and the Hamiltonian does not have momentum dependence, i A 1 m1 = 0 ∑ (∇ f ) i2 0 2m i = 1 2 1 2 ∂ 2 m3 = E (η ) 2 2 m ∂η η =0 , E (η ) = 0 e −η G ηG He m 0 , G = − [H , F ] 2 Physical meaning of sum rule Nucles under an impulse external field A V (t ) = Fδ (t ) = δ (t )∑ f ( ri ) i= 1 Nucleons i change its momentum by ∆ pi = − ∇ i f (ri ) The nucleon at t<0 has zero velocity expectation value, then this field creates the velocity field: ∇ f (r ) v (r ) = − Energy transferred to nucleon i is m ∆εi = 2 ∇ i f ( ri ) 2m Thus, the energy weighted sum rule has the following physical meaning. ∑ En n F 0 n 2 A 1 = 0 ∑ (∇ f ) i2 0 2m i = 1 The energy absorbed by nucleus: (Exc. energy) x (probability) The energy transferred to nucleons in the nucleus Meaing of m3 2 1 2 ∂ 2 m3 = E (η ) 2 2 m ∂η η =0 E (η ) = 0 e −η G ηG He m 0 , G = − [H , F ] 2 If f (r ) = r λ Yλ µ (rˆ) and the Hamiltonian does not have momentum dependence, G= − ηG The state e ( ) m 1 [ H , F ] = ∑ ∇ r λ Yλ µ (rˆ) i ⋅ ∇ 2 2 i 0 i λ introduces the velocity field v (r ) ~ ∇ r Yλ µ (rˆ) Its energy curvature with respect to the “deformation” is directly related to m3. E1 sum rule “E1 recoil charge” Isovector dipole operator A Fˆ = e∑ (τ z ) i ( zi − Rz ) = i= 1 N ∑ i= 1 Ze zi − A N+ Z Ne zi ∑ i= N + 1 A Assuming the ground state has I=0, NZ m1 = (1 + κ ) 2 Am The interaction usually contains the isospin-dependent terms. κ > 0 This includes effects of meson exchange current. κ = 0 ⇒ Thomas-Reiche-Kuhn (TRK) sum rule Isoscalar giant resonance Fˆ = ∑ f ( r ) = r λ Yλ 0 (rˆ) f (ri ) i The previous formulae lead to 2 2 A 1 λ (2λ + 1) 1 2 ∂ m1 = 0 ∑ ri 2 λ − 2 0 , m3 = E (η ) 2 2m 4π 2 m ∂η i= 1 η =0 E (η ) = 0 e −η G ηG He ( 1 A 0 , G = ∑ (∇ r λ Yλ 0 ) ⋅ ∇ 2 i= 1 ) i For instance, for the quadrupole operator, 2 1 2 5 m3 = ⋅8 T 2 m 16π <T >: Kinetic energy expectation value Giant quadrupole resonance energy ω 2 2+ 4T m3 ~ = ≈ 2ω 0 , 2 m1 mA r ( T ≈ mω 02 A r 2 2 ) Fermi Liquid Properties 1 ∂2 m3 ∝ E (η ) ∝ T 2 2 ∂η η =0 The density change of the ISGQR is a surface type, however, the restoring force for ISGQR originates from the kinetic energy. pz P( p) pz py px py px The vibration leads to a deformation in the momentum distribution This is different from low-lying surface vibrations and different from the classical (incompressible) liquid model. Four-current sum rule ∑ n Suzuki, Rowe, NPA286 (1977) 307. Suzuki, Prog. Theor. Phys. 64 (1980) 1627. i 0 j (r ) n n ρ (r ' ) 0 = − ρ 0 (r )∇ δ (r − r ' ) 2m ∇ ⋅ j ( r ) = i [ ρ ( r ), H ] Using the continuity equation and a property of the density operator ∫ ρ (r ) f (r )dr = A ∑ i= 1 f (ri ) we can obtain the energy-weighted sum rule for the density operator ∑ n 1 En 0 ρ ( r ) n n F 0 = − ∇ ⋅ ρ 0 (r )∇ f (r ) 2m The normal m1 sum rule can be easily derived from this formula. Taking the photoexcitation operator ∑ n f ( r ) = r λ Yλ 0 (rˆ) λ λ − 1 dρ 0 En 0 ρ ( r ) n n F 0 = − r Yλ µ (rˆ) 2m dr dρ 0 ρ (r ) n ~ r λ − 1 0 Yλ µ (rˆ) Transition density of the Tassie dr model for the giant resonance Time-Dependent Density Functional Theory (TDDFT) Basic ideas of the unified (collective) model • Nucleons are independently moving in a potential that slowly changes. – Collective motion induces oscillation/rotation of the potential. – The fluctuation of the potential changes the nucleonic single-particle motion. Consistent with the idea of Time-Dependent Mean-Field Theory or Time-Dependent Density-Functional Theory Time-dependent density-functional theory (TDDFT) • Basic theorem of DFT (Hohenberg-Kohn) • Basic theorem of TDDFT (Runge-Gross) • Perturbative regime: Linear response and random-phase approximation – Matrix formulation – Green’s function method – Real-time method – Finite amplitude method • Non-perturbative regime – Theories of large-amplitude collective motion Density Functional Theory • Quantum Mechanics – Many-body wave functions; Ψ (r1 , , rN ) • Density Functional Theory – Density clouds; F [ ρ (r )] The many-particle system can be described by a functional of density distribution in the three-dimensional space. Hohenberg-Kohn Theorem (1) Hohenberg & Kohn (1964) The first theorem Density ρ(r) determines v(r) , except for arbitrary choice of zero point. A system with a one-body potential v ( r ) Hv = H + ∑ v( ri ) HvΨ v gs = E gsv Ψ v gs i = ∑ i 2 pi + 2m ∑ i< j w(ri , r j ) + ∑ v( ri ) i gs ( ) v r ↔ Ψ ↔ ρ (r Existence of one-to-one mapping: v v ) Strictly speaking, one-to-one or one-to-none v-representative ① v( r ) ↔ Ψ v gs Here , we assume the non-degenerate g.s. Reductio ad absurdum: Assuming Ψ gsv different and produces v( r ) v' ( r ) the same ground state v v v ( H + V )Ψ gs = E gs Ψ gs V = ∑ v( ri ) i − ) ( H + V ' )Ψ v gs = E Ψ v' gs V '= v gs ∑ v' ( ri ) i (V − V ' )Ψ v gs = ( E gsv − E gsv ' )Ψ v gs V and V’ are identical except for constant. → Contradiction ② Ψ v gs ↔ ρv Again, reductio ad absurdum v assuming different states Ψ gs , Ψ ρv same density E gsv = Ψ v gs < Ψ v' gs H+V Ψ v' gs ( H + V )Ψ v gs = E gsv Ψ v gs ( H + V ' )Ψ v' gs = E gsv ' Ψ v' gs ( ) ( v r , v ' r with ) produces the v gs H + V Ψ gsv ' v v' E gs < E gs + ∫ dr [ v( r ) − v' ( r ) ] ρ v ( r ) H + V = H + V '+ (V − V ' ) v v Ψ gs V Ψ gs = ∫ dr v( r ) ρ v ( r ) Replacing V ↔ V’ E gsv ' < E gsv + ∫ dr [ v' ( r ) − v( r ) ] ρ v ( r ) ∴ E gsv + E gsv ' < E gsv + E gsv ' Contradiction ! ρ v v-representative. Here, we assume that the density is For degenerate case, we can prove one-to-one v( r ) ↔ ρ v (r ) Hohenberg-Kohn Theorem (2) The second theorem There is a energy density functional and the variational principle determines energy and density of the ground state. Any physical quantity must be a functional of density. v( r ) ↔ Ψ ↔ ρv Many-body wave function Ψ [ ρ ( r ) ] is a functional of densityρ(r). From theorem (1) Ev [ ρ ] ≡ Ψ [ ρ ] H + V Ψ [ ρ Energy functional for external potential v (r) Ev [ ρ v ] = Evgs < Ev [ ρ Ev [ ρ ] = FHK [ ρ ] + ∫ v gs ] ρ ( r ) v ( r ) dr ] Variational principle holds for vrepresentative density FHK [ ρ ] : v-independent universal functional The following variation leads to all the ground-state properties. { δ F[ ρ ] + ∫ ρ ( r ) v ( r ) dr − µ (∫ )} ρ ( r ) dr − N = 0 In principle, any physical quantity of the ground state should be a functional of density. Variation with respect to many-body wave functions ↓ Variation with respect to one-body density ↓ Physical quantity Ψ (r1 , , rN ) ρ (r ) A[ ρ (r )] = Ψ [ ρ ] Aˆ Ψ [ ρ ] v-representative→ N-representative Levy (1979, 1982) The “N-representative density” means that it has a corresponding many-body wave function. Ritz’ Variational Principle Min Ψ ( r1 ,.., rN ) H Ψ ( r1 ,.., rN ) ⇒ Ψ HΨ gs ( r1 ,.., rN ) = E gs Ψ gs ( r1 ,.., rN ) gs Decomposed into two steps Min Ψ H Ψ Min Ψ H Ψ = Min ρ ( r ) Ψ → ρ ( r ) F [ ρ ( r ) ] ≡ Min Ψ H Ψ Ψ → ρ (r) ( r1 ,.., rN ) One-to-one Correspondence External potential Minimum-energy state Ψ Ground state V (r ) Ψ V Density ρ (r ) v-representative density ρ V (r ) Time-dependent “HK” theorem Runge & Gross (1984) One-to-one mapping between time-dependent density ρ(r,t) and time-dependent potential v(r,t) except for a constant shift of the potential Condition for the external potential: Possibility of the Taylor expansion around finite time t0 ∞ 1 k v(r, t ) = ∑ vk (r )( t − t0 ) k = 0 k! The initial state is arbitrary. This condition allows an impulse potential, but forbids adiabatic switch-on. ∂ i Ψ (t ) = H (t ) Ψ (t ) ∂t Schrödinger equation: Current density follows the equation [ ] ∂ i j(r, t ) = Ψ (t ) ˆj(r ), H (t ) Ψ (t ) ∂t (1) Different potentials, v(r,t) , v’(r,t), make time evolution from the same vk (r ) − vk' (r ) ≠ c for ∃ k initial state into Ψ(t)、Ψ’(t) ∂ ∂t k+ 1 { j(r, t ) − j' (r, t )} t = t = − ρ (r, t0 )∇ wk (r ) ∂ wk (r ) = ∂t ∴ 0 k { v(r, t ) − v' (r, t )} t = t j(r, t ) ≠ j' (r, t ) 0 Continuity eq. = vk (r ) − v'k (r ) ≠ c ρ (r, t ) ≠ ρ ' (r, t ) at t > t0 Problem 1: Two external potentials are different, when their expansion ∞ 1 k v(r, t ) = ∑ vk (r )( t − t0 ) k = 0 k! has different coefficients at the zero-th order v0 (r ) − v0' (r ) ≠ c Using eq. (1), show ∂ { j(r, t ) − j' (r, t )} t = t0 = − ρ (r, t0 )∇ w0 (r ) ∂t w0 (r ) = { v(r, t ) − v' (r, t )} t = t = v0 (r ) − v'0 (r ) ≠ c 0 Next, if v0 (r ) − v0' (r ) = c then, show ∂ ∂t 2 { , but v1 (r ) − v1' (r ) ≠ c , j(r, t ) − j' (r, t )} = − ρ (r, t0 )∇ w1 (r ) t = t0 Problem 2: Using the continuity equation and the following equation ∂ ∂t k+ 1 { j(r, t ) − j' (r, t )} t = t = − ρ (r, t0 )∇ wk (r ) 0 ∂ wk (r ) = ∂t prove that ∂ ∂t k+ 2 k { v(r, t ) − v' (r, t )} t = t { ρ (r, t ) − ρ ' (r, t )} 0 = vk (r ) − v'k (r ) ≠ c = ∇ ⋅ { ρ (r, t0 )∇ wk (r )} t = t0 Then, show that the right-hand side cannot vanish identically, with ∇ wk (r ) ≠ 0 One-to-one Correspondence External potential V (r , t ) Time-dependent state starting from the initial state Ψ (t0 ) Time-dependent density TD state v-representative density Ψ (t ) ρ V (r,t) V (t ) The universal density functional exists, and the variational principle determines the time evolution. From the first theorem, we have ρ(r,t) ↔Ψ(t). Thus, the variation of the following function determines ρ(r,t) . ∂ S [ ρ ] = ∫ dt Ψ [ ρ ](t ) i − H (t ) Ψ [ ρ ](t ) t0 ∂t t1 ~ S [ ρ ] = S [ ρ ] − ∫ dt ∫ drρ (r, t )v(r, t ) t1 t0 The universal functional ~ S [ρ ] is determined. v-representative density is assumed. TD Kohn-Sham Scheme Real interacting system TD state V (r , t ) TD density Ψ (t) V ρ (r,t) Virtual non-interacting system Vs (r , t ) TD state Ψ (t) TD density S ρ (r,t) Time-dependent KS theory Assuming non-interacting v-representability ρ (r,t) = Time-dependent Kohn-Sham (TDKS) equation N ∑ i= 1 2 φ i (r ,t) 2 2 − ∇ + vs [ ρ ](r, t ) φ i (r, t ) 2m δ S [ρ ] vs [ ρ ](r, t ) = δ ρ (r, t ) t1 ∂ S [ ρ ] ≡ S [ ρ ] − ∫ Φ D [ ρ ](t ) i − T Φ D [ ρ ](t ) t0 ∂t ∂ i φ i (r, t ) = ∂t Solving the TDKS equation, in principle, we can obtain the exact time evolution of many-body systems. The functional depends on ρ(r,t) and the initial state Ψ0 . Time-dependent quantities → Information on excited states Ψ (0) = ∑ cn Φ ⇒ n Ψ (t ) = n ∑ cn e − iEnt Φ n n Energy projection 1 2π ∫ ∞ −∞ Ψ (t ) e iEt dt = ∑ cn Φ n δ ( E − En ) n Finite time period → Finite energy resolution T ~1 Γ 1 2π ∫ ∞ −∞ iEt − Γ t 2 Ψ (t ) e e dt = ∑ n cn Γ 2 Φ 2 2 π ( E − E n ) + ( Γ 2) n TDHF(TDDFT) calculation in 3D real space H. Flocard, S.E. Koonin, M.S. Weiss, Phys. Rev. 17(1978)1682. Small-amplitude limit (Random-phase approximation) One-body operator under a TD external potential i ∂ ρ (t ) = [ hKS [ ρ (t )] + Vext (t ), ρ (t )] ∂t Assuming that the external potential is weak, ρ (t ) = ρ 0 + δ ρ (t ) h(t ) = h0 + δ h(t ) = h0 + ∂ i δ ρ (t ) = [ h0 , δ ρ (t )] + [δ h(t ) + Vext (t ), ρ 0 ] ∂t δh δρ ⋅ δ ρ (t ) ρ0 Let us take the external field with a fixed frequency ω, Vext (t ) = Vext (ω )e − iω t + Vext+ (ω )e + iω t The density and residual field also oscillate with ω, δ ρ (t ) = δ ρ (ω )e − iω t + δ ρ + (ω )e + iω t δ h(t ) = δ h(ω )e − iω t + δ h + (ω )e + iω t The linear response (RPA) equation ω δ ρ (ω ) = [ h0 , δ ρ (ω )] + [δ h(ω ) + Vext (ω ), ρ 0 ] Note that all the quantities, except for ρ0 and h0, are non-hermitian. A ∑ ( δ ψ (t ) δ ρ (ω ) = ∑ ( X (ω ) δ ρ (t ) = i i= 1 A i= 1 i φ i + φ i δ ψ i (t ) ) φ i + φ i Yi (ω ) ) This leads to the following equations for X and Y: ω X i (ω ) = ( h0 − ε i ) X i (ω ) + Qˆ {δ h(ω ) + Vext (ω )} φ i ω Yi (ω ) = − Yi (ω ) ( h0 − ε i ) − φ i {δ h(ω ) + Vext (ω )} Qˆ Qˆ = A ∑ (1 − i= 1 φi φi ) These are often called “RPA equations” in nuclear physics. X and Y are called “forward” and “backward” amplitudes. If we start from the TDHF with a “density-independent” Hamiltonian (not from the energy functional), then, there is other ways to formulate the RPA. (see TextBooks) Matrix formulation ω X i (ω ) = ( h0 − ε i ) X i (ω ) + Qˆ {δ h(ω ) + Vext (ω )} φ i ω Yi (ω ) = − Yi (ω ) ( h0 − ε i ) − φ i {δ h(ω ) + Vext (ω )} Qˆ (1) Qˆ = A ∑ (1 − i= 1 φi φi If we expand the X and Y in particle orbitals: X i (ω ) = ∑ m> A φ m X mi (ω ) , Yi (ω ) = ∑ m> A φ m Ymi* (ω ) Taking overlaps of Eq.(1) with particle orbitals A * B B (Vext ) mi 1 0 X mi (ω ) − ω = − * A 0 − 1 Ymi (ω ) (Vext ) im Ami ,nj = (ε m − ε )δ mnδ ij + φ m Bmi ,nj = φ m ∂h ∂ ρ jn ∂h ∂ ρ nj φi ρ0 φi ρ0 In many cases, setting Vext=0 and solve the normal modes of excitations: → Diagonalization of the matrix ) Lowest negative-parity states (SGII functional) cal exp Green’s function method ω δ ρ (ω ) = [ h0 , δ ρ (ω )] + [δ h(ω ) + Vext (ω ), ρ 0 ] (2) Multiply Eq.(2) with φ i φ i from the right and from the left: δ ρ (ω ) φ i φ i = ( ω + ε i − h0 ) [Vscf , ρ 0 ] φ i φ i (3-1) φ i φ i δ ρ (ω ) = φ i φ i [Vscf , ρ 0 ]( ω − ε i + h0 ) (3-2) −1 −1 Vscf (ω ) ≡ Vext (ω ) + δ h(ω ) Sum up with respect to occupied orbitals i, then, add (3-1) and (3-2), using δ ρ (ω ) = { ρ 0 , δ ρ (ω )} the orthonormalization condition for KS orbitals (ρ2=ρ): δ ρ (ω ) = ∑ {G (ε 0 i + ω )Vscf φ i φ i + φ i φ i Vscf G0 (ε i − ω )} i G0 ( E ) ≡ ( E − h0 ) −1 If the Vscf is local, we can rewrite this as follows: δ ρ (r; ω ) = ∑ ∫ dr{G (r, r ' ; ε 0 i } + ω )Vscf (r ' )φ i (r ' )φ i* (r ) + φ i (r )φ i* (r ' )Vscf (r ' )G0 (r ' , r : ε i − ω ) i = ∫ drΠ 0 (r, r ' ; ω )Vscf (r ' ; ω ) where the independent-particle response function is defined by Π 0 (r , r ' ; ω ) ≡ ∑ ∫ dr{φ i i } (r )G0* (r, r ': ε i − ω * )φ i* (r ' ) + φ i* (r )G0 (r, r ' ; ε i + ω )φ i (r ' ) Green’s function method (cont.) An advantage of the Green’s function method is that we can treat the continuum exactly. Shlomo and Bertsch, NPA243 (1975) 507. ω → ω + iη Π 0 (r , r ' ; ω + iη ) = ∑ ∫ dr{ φ i } (r )G0( + )* (r, r ': ε i − ω )φ i* (r ' ) + φ i* (r )G0( + ) (r , r ' ; ε i + ω )φ i (r ' ) i G0( ± ) ( E ) ≡ ( E ± iη − h0 ) −1 In case h0 is spherical, the Green’s function can be easily obtained by the partial-wave expansion: 1 ul (r< )vl( + ) (r> ) * ˆ G ( E ) = 2m ∑ Y ( r ) Y (rˆ' ) lm lm (+ ) rr ' lm W [ul , vl ] (+ ) 0 In case h0 is deformed, we can construct the Green’s function by using the following identity: T.N. and Yabana, JCP114 (2001) 2550; PRC71 (2005) 024301. (± ) (± ) (± ) (± ) Gdef ( E ) = Gsph ( E ) + Gsph ( E )( hdef − hsph )Gdef (E) Real-time method In the RPA calculations (matrix formulation & Green’s function method), the most tedious part is the calculation of the residual induced fields: δ h(ω ) = δh δρ ⋅ δ ρ (ω ) ρ0 In the original time-dependent equations, this effect is included in the self-consistent potential: h[ ρ (t )] = h0 + δ h(t ) , δ h(t ) = δh δρ ⋅ δ ρ (t ) ρ0 Therefore, in principle, the RPA can be achieved by solving the TD Kohn-Sham equations, starting from the ground state with a weak perturbation. ∂ i φ i (r , t ) = ∂t 2 2 − ∇ + vs [ ρ ](r, t ) + Vext (t ) φ i (r, t ) 2m Skyrme TDDFT in real space Time-dependent Hartree-Fock equation ( − iη~ ( r ) ) ∂ i ψ i (rσ τ , t ) = hSk [ ρ ,τ , j, s, J ](t ) + Vext (t ) ψ i (rσ τ , t ) ∂t 3D space is discretized in lattice n = 1,Mt Single-particle orbital: ϕ i (r, t ) = {ϕ i (rk , t n )}k = 1,Mr , i = 1, , N N: Number of particles y [ fm ] Mr: Number of mesh points Mt: Number of time slices Spatial mesh size is about 1 fm. Time step is about 0.2 fm/c Nakatsukasa, Yabana, Phys. Rev. C71 (2005) 024301 X [ fm ] Calculation of time evolution Time evolution is calculated by the finite-order Taylor expansion t+ ∆ t ψ i (t + ∆ t ) = exp − i ∫ h(t ' )dt ' ψ i (t ) t ≈ ∑ ( − i∆ t h(t + n n! ∆ t 2) ) n ψ i (t ) Violation of the unitarity is negligible if the time step is small enough: ∆ t ε max < < 1 ε max The maximum (single-particle) eigenenergy in the model space Real-time calculation of response functions 1. Weak instantaneous external perturbation Ψ (t ) Fˆ Ψ (t ) Vext (t ) = η Fˆδ (t ) 3. Calculate time evolution of Ψ (t ) Fˆ Ψ (t ) 5. Fourier transform to energy domain dB (ω ; Fˆ ) 1 = − Im ∫ Ψ (t ) Fˆ Ψ (t ) e iω t dt dω πη dB (ω ; Fˆ ) dω ω [ MeV ] Real-time dynamics of electrons in photoabsorption of molecules 1. External perturbation t=0 Vext ( r, t ) = − ε riδ (t ), i = x, y , z 2. Time evolution of dipole moment di ( t ) ∝ ∫ r ρ (r, t ) i E at t=0 Ethylene molecule Comparison with measurement (linear optical absorption) TDDFT accurately describe optical absorption Dynamical screening effect is significant PZ+LB94 i ∂ ψ i (r , t ) = h[n(r , t )]ψ i (r , t ) ∂t with Dynamical screening without i ∂ ψ i (r , t ) = h[n0 (r )]ψ i (r , t ) ∂t TDDFT Exp Without dynamical screening (frozen Hamiltonian) T. Nakatsukasa, K. Yabana, J. Chem. Phys. 114(2001)2550. -- - - - - - Eext ( t ) ++ + + + ++ Eind ( t ) Photoabsorption cross section in C3H6 isomer molecules Nakatsukasa & Yabana, Chem. Phys. Lett. 374 (2003) 613. TDLDA cal with LB94 in 3D real space • 33401 lattice points (r < 6 Å) • Isomer effects can be understood in terms of symmetry and antiscreening effects on bound-to-continuum excitations. Cross section [ Mb ] • Photon energy [ eV ] Neutrons 16 O δ ρ n (t ) = ρ n (t ) − ( ρ 0 ) n Time-dep. transition density δρ> 0 δρ< 0 δ ρ p (t ) = ρ p (t ) − ( ρ 0 ) p Protons 16 O 18 O Prolate 10 20 30 Ex [ MeV ] 40 26 Mg 24 Triaxial Prolate 10 20 30 Ex [ MeV ] Mg 40 10 20 30 Ex [ MeV ] 40 Si Si 28 30 Oblate Oblate 10 20 30 Ex [ MeV ] 10 20 30 Ex [ MeV ] 40 40 Ca 44 Prolate 40 48 Ca 10 20 Ca 30 Ex [ MeV ] 10 10 20 30 Ex [ MeV ] 40 20 30 Ex [ MeV ] 40 Finite Amplitude Method T.N., Inakura, Yabana, PRC76 (2007) 024318. A method to avoid the explicit calculation of the residual fields (interactions) ω X i (ω ) = ( h0 − ε i ) X i (ω ) + Qˆ {δ h(ω ) + Vext (ω )} φ i ω Yi (ω ) = − Yi (ω ) ( h0 − ε i ) − φ i {δ h(ω ) + Vext (ω )} Qˆ (1) Residual fields can be estimated by the finite difference method: δ h(ω ) = ψ i 1 ( h[ ψ ' , ψ η ]− h ) = φ i + η X i (ω ) , 0 ψ ' i = φ i + η Yi (ω ) Starting from initial amplitudes X(0)and Y(0), one can use an iterative method to solve eq. (1). Programming of the RPA code becomes very much trivial, because we only need calculation of the single-particle potential, with different bras and kets. Fully self-consistent calculation of E1 strength distribution Inakura, T.N., Yabana, in preparation Z SkM* Rbox= 15 fm Γ = 1 MeV N Large Amplitude Collective Motion Beyond the small-amplitude approximation • In the small-amplitude limit, the normal modes are obtained by diagonalizing the RPA matrix. → “Quantization” is on hand. • Large amplitude collective motion – Real-time approach to non-linear response – Adiabatic TDHF – Self-consistent collective coordinate method Real-time approach to non-linear response • In principle, non-linear response can be studied with the real-time method. – Accuracy – Applicability TDHF(TDDFT) calculation in 3D real space H. Flocard, S.E. Koonin, M.S. Weiss, Phys. Rev. 17(1978)1682. Ionization by Laser Electrons in a strong electric field Laser field, E ~ Electric field of ions binding electrons Laser frequency ω ~ HOMO-LUMO gap Re-scattering process: A new probe for electronic structure Wave packet split by the laser field Re-accelerated toward the origin Scattered by the remaining part of itself eE(t)z γ«1 z Keldysh parameter γ = τ tunnelω laser Higher-order harmonic generation ψ (t) = α φ + β e ik ⋅ r − iE k t / d (t ) = ψ ( t ) z ψ ( t ) ≈ α * β φ z e I (ω ) ∝ ∫ dte iω t d A ( t ) ik ⋅ r − iE k t / + cc eE(t)z 2 Ne z HOMO orbital in N2 (Molecular tomography) Ab inito calculation N2 molecule 2x1014W/cm2, 800nm laser Calculated by Yabana 50fs ψ (t) = α φ + β e ik ⋅ r − iEk t / d (t ) = ψ ( t ) z ψ ( t ) ≈ α β φ z e * I (ω ) ∝ ∫ dte iω t dA(t) 2 ik ⋅ r − iEk t / + cc |dA(ω ) | 2 Comparison with experiments Calculation 10 Experiment 8 10 7 10 6 10 5 10 4 10 3 15 N2 0 deg. N2 45 deg. N2 90 deg. Ar 20 25 30 35 Harmonic order 40 45 θ Large amplitude collective motion (LACM) in nuclei • Fission • Decay of superdeformed band • Shape-coexistence phenomena Φ = c1 + c2 Adiabatic theories of LACM • Baranger-Veneroni, 1972-1978 ρ (t ) = e iχ (t ) ρ 0 e − iχ (t ) •Expansion with respect to χ • Villars, 1975-1977 • Eq. for the collective subspace (zero-th and first-order w.r.t. momenta) ∂V Q (q ) Φ (q ) = 0 ∂q ∂ δ Φ (q ) [ H , Q (q )] + iM (q ) − 1 Φ (q ) = 0 ∂q δ Φ (q ) H − •Non-uniqueness problem “Validity condition” (Goeke-Reinhard, 1978-) Goeke, Reinhard, Rowe, NPA359 (1981) 408 Approaches to Non-uniqueness Problem (1) Yamamura-Kuriyama-Iida, 1984 Requirement of “analyticity” (ex) Moya de Guerra-Villars, 1978) (2) Rowe, Mukhejee-Pal, 1981 Therefore, in principle, we can determine a unique collective path Requirement of “Point transf.” and in the ATDHF. The higher-order in p equations up to O(p2) can be systematically treated. In practice, it is only applicable to simple models. There is no systematic way to go beyond the second order in p. In practice, the method is applicable to realistic models as well. Non-adiabatic theories of LACM • Rowe-Bassermann, Marumori, Holzwarth-Yukawa, 1974• Local Harmonic Approach (LHA) • Curvature problem • Correspondence between, Q,P ↔ Infinitesimal generator, is not guaranteed. δ Φ (q ) H − ∂V Q (q ) Φ (q) = 0 ∂q δ Φ (q ) [ H , Q (q )] + iM (q ) − 1 P (q ) Φ (q ) = 0 δ Φ (q ) [ H , P (q )] − iC (q )Q (q ) Φ (q ) = 0 • Marumori et al, 1980• Self-consistent collective coordinate (SCC) method • The problems of LHA are solved. • The SCC equation is solved by the expansion with respect to (q,p). δ Φ ( q, p ) H − ∂H ∂H Q− P Φ (q) = 0 ∂q ∂p H ≡ Φ ( q, p ) H Φ ( q, p ) • “Adiabatic” approx. → LACM (Matsuo, TN, Matsuyanagi, 2000) TDHF(B) → Classical Hamilton’s form Blaizot, Ripka, “Quantum Theory of Finite Systems” (1986) Yamamura, Kuriyama, Prog. Theor. Phys. Suppl. 93 (1987) The TDHF(B) equation can be described by the classical form. For instance, using the Thouless form 1 z = exp z µ ν aµ+ aν+ Φ 2 0 The TDHF(B) equation becomes in a form ∂H + ( 1 + z z) ∂ z+ ∂H iz + = − 2(1 + z + z ) (1 + zz + ) ∂z iz = 2(1 + zz + ) The Holstein-Primakoff-type mapping leads to ∂H ∂β + ∂H iβ + = − 2 ∂β iβ = 2 β µ ν = ( ξ + iπ + H ( z, z ) = [ zH z z z β µ ν = z (1 + z + z )1/ 2 )µν 2 ] µν ∂H ξµ ν = ∂ π µν iπ µ ν = − ∂H ∂ ξ µν TDHF(B) →Small amplitude limit Small fluctuation around the HF(B) state ξ ,π H ξ ,π = Φ 0 HΦ 2qp index α = ( µ ν ) A 1 1 + + − TrA + a a , aa * 2 2 B ( 0 Aα β = Φ 0 [(aa )α , [ H , (a + a + ) β ]] Φ 0 B aa * + + A a a ) , Bα β = Φ 0 [(aa )α , [ H , (aa ) β ]] Φ 0 α Rewriting the last term in terms of variables (ξ , π α ) A B κ 1 + ξ , π H ξ , π = ERPA + (κ * , κ ) * , κ ≡ β ( 1 − β β )≈ β * * 2 B A κ 1 1 ( A + B )α β ξ α ξ β + ( A − B )α β π α π β 2 2 α µ Linear point transformation (ξ , π α ) → (q , pµ ) leads to 1 = ERPA + ∑ pµ2 + ω n2 (q µ ) 2 2 n = ERPA + [ ] qµ = δ µν 1 ωµ ∑ (X α µ + Yµ ) α ξ α , pµ = ν ∂ qµ αβ ∂q 2 = ( A − B ) , ω δ µ α β ∂ξ ∂ξ µν ω µ ∑ (X α µ − Yµ ) α πα ∂ξ α ∂ξ β = ( A + B )α β µ ∂q ∂ qν Decoupled classical motion within the point transformation Klein, Walet, DoDang, Ann. Phys. 208 (1991) 90 Expanding the classical Hamiltonian w.r.t. momentum up to 2nd order 1 H (ξ , π ) = Bα β π α π 2 β + V (ξ ) , B ∂ 2H ≡ ∂π α∂π αβ (ξ , π ) → (q, p) Point transformation q µ = f µ (ξ ) , ξ α = g α (q) β π =0 ∂ gα ∂ ξ α g ≡ = , ∂ qµ ∂ qµ α ,µ pµ = g ,αµ π α , π α = f ,αµ pµ ∂f µ ∂ qµ f ≡ = ∂ξ α ∂ξ α µ ,α Point transformation conserves the quadratic form in momenta. 1 H ( q, p ) = 2 B µ ν pµ pν + V (q) , B µ ν ≡ f ,αµ Bα β f ,νβ γβ β Metric tensor: Bα β : defined by Bα γ B = δ α Shift-up and down of α µ β indexes: Chain rules: g ,µ f , β = δ α , V ,α ≡ Bα β V, β f ,αµ g ,αν = δ ν µ (=canonical variable cond.) Assuming that there is a decoupled path (1-dim. collective submanifold) q1 : Collective coordinate, q n : Non - collective coord. Decoupling condition: q = pn = 0 ⇒ q = p n = 0 n n qn q1 = f 1 (ξ ) (1) V,n = 0 (1) V,α = V,1 f ,α1 (2) B n1 = 0 (2) Bα β f ,1β = B 11 g ,α1 (3) B,11 n = 0 11 1 (3) B,11 α = B,1 f ,α 1 1 1 1 Decoupling condition (1) ↔ HF(B) with the constraint q = Φ (q ) Qˆ (q ) Φ (q ) ∂ V 1 δ V (ξ ) − 1 q = δ { H (ξ , π = 0) − λ q1 (ξ )} = δ Φ (q1 ) H − λ Qˆ (q1 ) Φ (q1 ) = 0 ∂q Using the decoupling conditions (2) and (3), we may construct the constraint operator Q(q). More precisely speaking, we can determine the 2qp parts of Q(q). µ Differentiating the chain relation V,α = V, µ f ,α V,α β = V, µ ν f ,αµ f ,νβ + V, µ f ,αµ β The last term indicates that the second derivative of the potential is not covariant. This can be rewritten in a covariant derivative V;α β ≡ V,α β − Γ αγ β V,γ , V;µ ν ≡ V, µ ν − Γ µρν V, ρ V;α β = V;µ ν f ,αµ f ,νβ Here, two different definitions of the metric tensor are possible: (i) Riemannian type Mass tensor as the metric tensor Γ αγ β ≡ 1 γδ B ( Bδ α , β + Bδ β ,α − Bα β ,δ 2 ) Affine connection (ii) Symplectic type Metric tensor Γ αγ β ≡ Kα β = ∑ µ f ,αµ f , βµ , K α β = 1 γδ K ( Kδ α , β + K δ β ,α − Kα β ,δ ) = g ,γµ f ,αµ β 2 ∑ µ g ,αµ g ,βµ Kα β = Bα β With this metric, the decoupled space is assumed to be “flat”. A certain combination of the decoupling conditions (1-3) leads to the following Local Harmonic Equation (LHE) (with metric tensor Kij): ,β ;α f =ω f 1 ,β V 2 1 ,α V;α, β ≡ B β γ V;α γ , V;α β ≡ V,α β − Γ αγ β V,γ (i) Riemannian LHE The condition (3) is equivalent to that the decoupled collective path is geodesic with metric tensor of Bα β δ ∫ B11 (q1 )dq1 = 0 ⇒ f ,α1β − Γ αγ β f ,γ1 + Γ 111 f ,α1 f ,1β = 0 Then, using the condition (2), we can derive the LHE above. ω 2 = V;11 = B 11 (V,11 − Γ 111 V,1 ) (ii) Symplectic LHE Without the condition (3), we can derive the LHE. ω 2 = V;11 = B 11V,11 1 Either neglect, or determine by a certain condition, the curvature f ,α β Riemannian LHE vs Symplectic LHE Symplectic LHE is (almost) identical to the “adiabatic” approximation of the Self-consistent Collective Coordinate (SCC) Method Original formulation: Matsuo, TN, Matsuyanagi, Prog. Theor. Phys. 103 (2000) 959 Gauge-invariant formulation: Hinohara et al, PTP 117 (2007) 451 We believe that the Symplectic LHE (ASCC) is superior to the Riemannian LHE in the following reasons: • Extension to lift the restriction to the point transformation can be consistently achieved. • Both formalisms coincide with the RPA at equilibrium. However, in case of superconducting nuclei, the “extended” symplectic LHE naturally becomes identical to the QRPA. • Nambu-Goldstone modes are automatically separated from the decoupled collective variables, as zero-energy solutions. Separation of Nambu-Goldstone modes Extended “point” transformation 1 q = f (ξ ) + f (1) µ α β π α π 2 pµ = g ,αµ π α + O (π 3 ) µ µ β + O (π ) 4 ξ α πα 1 (1)α µ ν g pµ pν + O ( p 4 ) 2 = f ,αµ pµ + O ( p 3 ) = g α (q ) + This extension leads to the modification of mass parameter, but the other ~ formulation is kept invariant. B α β ≡ Bα β − V f (1) µ α β ,µ (1) Symmetry operator S = momentum ps = g ,αs π α + O (π 3 ) { ps , H }PB = 0 TN, Walet, DoDang, PRC61 (1999) 014302 g ,αsV,α = 0 ⇒ V;α β g ,αs = 0 (2) Symmetry operator S = coordinate q s = f s (ξ ) + 1 (1) sα β f π α π β + O (π 4 ) 2 ~ B α β f , βs = Bα β f , βs − V,µ f (1) µ α β f ,βs = Bα β f , βs − V, µ f (1) sα β f , βµ = Bα β f , βs − V, β f (1) sα β = 0 ({q µ , qν }PB = 0) ({q s , H }PB = 0) Collective path and re-quantization Solve the constrained MF eq. and LHE to obtain self-consistent solutions Symplectic LHE Adiabatic SCC (CMF) V,α = V,1 f ,α1 (CMF) δ φ (q) Hˆ − (∂ V ∂ q)Qˆ (q) φ (q) = 0 (LHE) V;α, β f ,1β = ω 2 f ,α1 (LHE) δ φ (q) Hˆ (q), iQˆ (q) − B(q) Pˆ (q) φ (q ) = 0 ( V;α, β ≡ B β γ V,α λ − V,1 f ,α1 γ ) δ [ ] φ (q) [ Hˆ − (∂ V ∂ q)Qˆ (q), Pˆ (q) i ] − C (q )Qˆ (q) 1 [ [ − Hˆ , (∂ V ∂ q)Qˆ (q)] , Qˆ (q)] φ (q) = 0 2 B(q ) We obtain a series of “Slater determinants”, as the solutions. φ (q1 ) , φ (q2 ) , φ (q3 ) , GCM B (q1 ) , B (q2 ) , B (q3 ) , V (q1 ) ,V (q2 ) ,V (q3 ) , H ( q, p ) = q1 = f 1 (ξ ) “Collective Hamiltonian” ∂ ∂ 1 1 B (q) + V (q) B (q) p 2 + V (q) ⇒ B (q) 2 2 i∂ q i∂ q Applications to simple models Applications to O(4) models Model Hamiltonian Monopole+ “Quadrupole” pairing + “Quadrupole” int. P0 ≡ σ jm ∑ ( ) ( ) 1 1 1 G0 P0+ P0 + P0 P0+ − G2 P2+ P2 + P2 P2+ − χ Q 2 2 2 2 ∑ c j − mc jm , P2 ≡ ∑ ∑ σ jm c j − mc jm , Q ≡ ∑ d j ∑ σ H = h0 − j m> 0 j m> 0 1 m < Ω j /2 = − 1 m > Ω j /2 Parameters ε 1 = 0, ε 2 = 1.0, ε 3 = 3.5 d1 = 2.0, d 2 = 1.0, d 3 = 1.0 Ω 1 = 14, Ω 2 = 10, Ω 3 = 4 j jm c +jm c jm m ε3 ε2 ε1 2Ω j = 2 j + 1 Hinohara, TN, Matsuo, Matsuyanagi, PTP115 (2006) 567 Potential Mass: M=1/B Larger G0 Effects of time-odd MF D (q ) ≡ φ (q ) Q φ (q ) Cranking Mass Exact Adiabatic SCC CHB with Mcranking Time-odd effects are neglected in the cranking mass ! Curvature effects Qˆ ( q) = ∑ µν QµAν ( aµ+ aν+ + h.c.) + ∑ QµBν aµ+ aν µν f ,α1 β In this model, requiring the gauge invariance, we can determine them. The curvature effects are weak. Neglected calc Full calc Model of protons and neutrons T.N. & Walet, PRC58 (1998) 3397 Gp=0.3 Neutrons Gp=10 Protons Upper orbital has a larger quadrupole moment CHB+Mcr LHE Exact Adiabatic vs Diabatic Dynamics Review: Nazarewicz, NPA557 (1993) 489c The problem has been discussed since the paper by Hill and Wheeler (1953) The pairing interaction plays a key role for configuration changes at level crossings. ε ε t ic a b ia d ε β β ε adia b β “Specialization energy” atic β Spontaneous fission life time is much larger for odd nuclei. Applications to more realistic models: Separable-force model Calculations carried out by Dr Nobuo Hinohara (YITP, Japan) Shape coexist ence in N~Z~40 r egion Se 68 neut r on single par t icle ener gy Kr 72 50 42 40 36 34 34 38 28 20 Z,N = 34,36 (oblat e magic number s) Z,N = 38 (pr olat e magic number ) Fischer et al . Phys.Rev.C67 (2003)064318. Bouchez et al. Phys.Rev.Lett.90(2003) 082502. oblat e-pr olat e shape coexist ence oblat e gr ound st at e shape coexist ence/mixing Skyr me-HFB: Yamagami et al . Nucl.Phys.A693 (2001)579. Mi cr os copi c t heor y t o des cr i be s hape coexi s t ence Lar ge-Scale Shell Model Calculat ion Dimension becomes too large for medium-heavy nuclei (1013 dim for 80Zr, 40Ca core) Too har d t o per for m ! 68Se: Kaneko et al. Phys.Rev.C70 (2004)051301. Model Space: 56Ni core, fpg-shell 1.6 x 108 dim GCM 72Kr: Bender et al. Phys. Rev. C74 (2006) 024312. Skyrme interaction Generator Coordinate: axial symmetric deformation The t r iaxial defor mat ion is ignor ed. Adiabat ic TDHF Adiabatic Self-consistent Collective Coordinate Method 68Se, 72Kr: Kobayasi et al. (Prog.Theor.Phys.112(2004), 113(2005)) Almehed et al. (Phys. Lett. B604 (2004)163.) Impor t ance of t r iaxial defor mat ion is discussed Pair ing + Quadr upole Model (68Se, 72Kr ) Microscopic Hamiltonian SP energy + Pairing (Monopole, Quadrupole) + Quadrupole interaction Model Space two major shells (Nsh=3,4) (40Ca core) Parameters sp energy: Modified Oscillator interaction strength monopole pairing and quadrupole int. strength: adjusted to the pairing gaps and deformations of Skyrme-HFB (Yamagami et al. NPA693(2001) ) quadrupole pairing strength G2: G2 = 0 G2 = (G2)self (self-consistent value) Sakamoto and Kishimoto PLB245 (1990) 321. (G2)self restores the Galilean invariance in RPA order, which was broken by the monopole pairing. Collect ive pat h in 68Se G2=0: Kobayasi et al., PTP113(2005), 129. Collective potential Moving-frame QRPA frequency Collective mass Moment of Inertia β γ collective path Triaxial deformation connects two local minima Enhancement of the collective mass and MoI by the quadrupole pairing Due to the contribution from the time-odd component Prog.Theor.Phys.115(2006)567. Collect ive pat h in 72Kr G2=0: Kobayasi et al., PTP113(2005), 129. Collective potential Collective mass q=0 Moment of Inertia prolate oblate Moving-frame QRPA frequency γ β β β γ γ γ β Bifurcation of the path Triaxial degrees of freedom: important Enhancement of the collective mass and MoI by the quadrupole pairing Basic Scheme of ASCC Met hod (2) 3rd Step: Requantize the collective Hamiltonian. Collective wave function Collective Hamiltonian vibrational wavefunctions K: 3-axis component of angular momentum 3-axis: quantization axis, symmetry axis(γ=0o) boundary conditions for collective wave functions periodic boundary condition at γ=0°and 60°for 68Se Kumar and Baranger Nucl. Phys. A92 (1967) 608. box boundary condition for 72Kr 4th Step: Calculate EM transitions E2 transitions, spectroscopic quadrupole moments … Ener gy spect r a of 68Se two rotational bands ( ) …B(E2) e2 fm4 effective charge: epol = 0.904 02+ state quadrupole pairing reduces ex.energy EXP:Fischer et al., Phys.Rev.C67 (2003) 064318. Collect ive Wavefunct ions of 68Se G2 = 0 G2 = G2self localization I = 0: oblate and prolate shapes are strongly mixed via a triaxial degree of freedom ground band: mixing of different K states、excited band: K=0 dominant oblate-prolate mixing: strong in 0+ states, reduced as angular momentum increases Ener gy Spect r a of 72Kr effective charge is adjusted to this value two rotational bands 2 4 ( ) …B(E2) e fm small inter-band B(E2): shape mixing rather weak EXP:Fischer et al., Phys.Rev.C67 (2003) 064318, Bouchez, et al., Phys.Rev.Lett.90 (2003) 082502. Gade, et al., Phys.Rev.Lett.95 (2005) 022502, 96 (2006) 189901 Collect ive wavefunct ions of 72Kr G2 = 0 G2 = G2self oblate 01+ state: well localized around oblate shape 02+ state: weak oblate-prolate shape mixing other states: well defined shape character prolate oblate prolate Spect r os copi c quadr upol e moment Fission Staszczak et al. • Optimal path to fission • Diabatic vs Adiabatic dynamics • Collective mass parameters Self-consistent, self-determined, microscopic description of nuclear fission Summary • Liquid-drop, shell, unified models, cranking model • Nuclear structure at high spin and large deformation • Sum-rule approaches to giant resonances • Basic theorem for the Time-dependent densityfunctional theory (TDDFT) • Linearized TDDFT (RPA) and elementary modes of nuclear excitation • Theories of large-amplitude collective motion • Anharmonic vibrations, shape coexistence phenomena