Computational methods for Coulomb four-body systems Zong-Chao Yan University of New Brunswick Canada Collaborators: G. W. F. Drake Liming Wang Chun Li August 24-27, 2015, Trento NSERC, SHARCnet, ACEnet, CAS/SAFEA The rms nuclar charge radius : 1 rc c ( r ) r 2d 3r Z where c is the nuclear charge density, which requires 2 the nuclear wa ve function. Atomic physics method: Proposed by Drake in 90s for isotope shifts Shiner et al 3,4He Riis,..,Drake, 6,7Li+ Extended to radioactive isotopes: in the past 10 years 6He, 8He, 11Li, 11Be (Argonne, GSI, Drake, Pachucki et al.) Currently: 8B, one-proton halo (Argonne, GSI) Etheory Enr 2 Erel 3 EQED Enuc 2 Z rc Enuc 3 2 r i i a b a b a b a b Etheory Enra b 2Erel 3EQED Enuc for a transiti on a b A,A' IS A,A' MS M A M A' , mass shift M A M A' A,A' MS A,A' FS , for two isotopes A and A' A,A' FS Z (0) ( rc2 )( A, A' ), field shift 2 b b a a A No experiment can separate MS and FS so that we have to reply on theory to determine MS accurately. A’ Isotope shift MS 1 FS Z Why isotope shifts? Mass independen t terms 1, 2 , 3 , 4 are canceled exactly for IS 2 rc Since Enuc 3 4 M 2 r i Mass dependent terms ( M ), M , 2 ( M ), 2 ( M )2 , 3 ( M ) 2 are known to high precision 1 MHz i 10 kHz 1 MHz ignored as a theoreti cal uncertaint y In particular 4 ~ 1 MHz, but cancelled for IS Finally, nuclear polarizability: Several nuclides have a halo in the excited state not in the ground state (Pachucki et al) Absolute measurement Theoretical background For low-Z systems, we use perturbation theory: H H H rel H QED H E Etot E H rel H QED Variational principle: tr H 0 tr Etr tr tr H E then Etr E0 Rayleigh-Ritz Method: Choose a basis set { p , p , , N } Then Now Letting tr N c p p p Etr Etr (c,c , ,cN ) Etr 0 c p we have a generalized eigenvalue equation Hc Oc H ij i H 0 j Oij i j 1 2 Z 2 i r i 1 i 3 H0 1 i j M i j rij 3 Hylleraas basis set: j1 j2 j3 j12 j23 j31 r1 r2 r3 r1 r2 r3 r12 r23 r31 e The basis is generated according to j j j j j j The nonlinear parameters are optimized by E | H | E | LM l1l2 l12 ,l3 Y rˆ1, rˆ2 , rˆ3 I dr1dr2 dr3 r1 r2 r3 r12 r23 r31 e j1 j2 j3 j1 2 j3 1 r1 r2 r3 j2 3 Perkins expansion: M L q k r Pq cos C jqk rq k rj q k j j , L j q for even j j for odd j , L M M If all I j , j , j are odd, then the integral becomes an infinite series: T q q In terms of W integrals: W l , m , n; , , l! l dxx e x x m dyy e y y dzz n e z l m n p ! p l p !l m p l m n F , l m n p ; l m p ; p Ground state of lithium Li 2s-2p oscillator strength Ω 10 11 12 13 14 15 length 0.7469568293363 0.7469568131213 0.7469568104873 0.7469568101163 0.7469568100337 0.74695681002 velocity 0.7469573407 0.7469568605 0.7469568221 0.7469568120 0.7469568102 0.7469568101 accelera. 0.7472526 0.7469435 0.7469664 0.7469573 0.7469581 0.7469572 Relativistic and QED corrections H rel i j rij (rij i ) j i rij i j rij ri ln nLS EQED Z Q i lim r a i j nLS a n ij pn ln rij i j ln a rij En E ln En E pn E n E n The Bethe logarithm nLS is very difficult to calculate. Q Drake-Goldman Method: Can. J. Phys. 77, 835 (1999) nLS n 0pn 2 En E0 ln En E0 0pn 2 E n E0 n j1 j2 j3 j1 2 j2 3 rˆ1 , rˆ2 , rˆ3 YlLM 1l2 l1 2 ,l3 j3 1 [ Z r1 ( Z 1) g K r2 ( Z 2 ) r3 / n ] r1 r2 r3 r12 r23 r31 e j1 j2 j3 j12 j23 j31 K 2 K 0,, g 10 Works for atoms: H, He, Li Molecule: H2+ (converged to 9-10 digits) M MP ( 0) MP ln( Z 2 / M ) M ( M( 0 ) ( 0 ) ) /( / M ) 1 (N1, N2) (0) Rati o MP (4172, 875) 2.980036890 0.113984800 (4172, 1452) 2.980822654 0.113843054 (4172, 2445) 2.980912900 8.7 0.113802133 (4172, 4109) 2.980930379 5.6 0.113797473 (4172, 6809) 2.980937153 2.5 0.113802599 Extrapolation 2.980941(4) 0.11383(3) P-K-P (2013) 2.980944(4) 0.11381(3) Puchalski, Kedziera, Pachucki PRA 87, 032503 (2013) l! l m n p 2 ! W ( l , m , n; , , ) l mn 3 p 0 l 1 p ! l m 2 p 2 F1 1, l m n p 3; l m p 3; Slow convergence when: ~1 Li, Wang, Yan, Int. J. Quantum Chem. 113,1307(2013) p Singular integral: type I I dr1dr2dr3r1 r2 r3 r12 r23 r31 e j1 j2 j3 2 j23 j31 r1 r2 r3 Our approach: 2 • Expand r etc. into infinite series • Perform multiple summation with convergence accelerators • Absolutely numerically stable rj Pachucki’s approach: • Recursion relations with quadrature Singular integral: type II I dr1dr2dr3r1 r2 r3 r12 r j1 j2 j3 2 2 j3 1 r1 r2 r3 23 31 r e This integral appears in using global operator 4 method. We currently calculate this using the asymptotic - expansion technique twice. King et al : direct expansion method Pachucki et al : resursive methods Other methods a)Explicitly correlated Gaussian Extensively used by Adamowicz et al and Pachucki et al (sometimes mixed use with Hylleraas) up to Be b) Hylleraas-CI Sims and Hagstrom, He, Li, Be, but for nonrelativistic case