39
Effective free dynamics of a tracer-particle coupled to a Fermi gas in the high-density limit David Mitrouskas Department Mathematik der LMU M¨ unchen joint work with Maximilian Jeblick, S¨oren Petrat and Peter Pickl Basque Center for Applied Mathematics, September 29

E ective free dynamics of a tracer-particle coupled to a Fermi gas … · 2016. 1. 27. · E ective free dynamics of a tracer-particle coupled to a Fermi gas in the high-density limit

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

  • Effective free dynamics of a tracer-particlecoupled to a Fermi gas in the high-density limit

    David Mitrouskas

    Department Mathematik der LMU München

    joint work with Maximilian Jeblick, Sören Petrat and Peter Pickl

    Basque Center for Applied Mathematics, September 29

  • motivation

    Phys. Rev. Letters 32, 23 (1974)

    1 E. Fermi and E. Teller, Phys. Rev. 72, 399 (1947)

    2 M. Gryzinski, Phys. Rev. 111, 900 (1958)

  • table of contents

    1 Microscopic Model

    2 Main result and effective model

    3 Sketch of the Proof

    4 Explanation as a mean field type result

    5 Conclusion

  • table of contents

    1 Microscopic Model

    2 Main result and effective model

    3 Sketch of the Proof

    4 Explanation as a mean field type result

    5 Conclusion

  • microscopic model

    • N+1 particle wave function on a d-dim. torus of length Λ

    Ψ ∈ L2(Td(N+1), ddx1...ddxN ddy)

    • non-relativistic Schrödinger equation

    i∂tΨt = HNΨt ⇒ Ψt = e−iHN tΨ0

    • microscopic Hamiltonian

    HN = −∆y −N∑k=1

    ∆xk + g

    N∑k=1

    v(xk − y)

    - v ∈ C∞0 (Td)- g = 1: not in the weak-coupling regime

  • microscopic model

    • N+1 particle wave function on a d-dim. torus of length Λ

    Ψ ∈ L2(Td(N+1), ddx1...ddxN ddy)

    • non-relativistic Schrödinger equation

    i∂tΨt = HNΨt ⇒ Ψt = e−iHN tΨ0

    • microscopic Hamiltonian

    HN = −∆y −N∑k=1

    ∆xk + g

    N∑k=1

    v(xk − y)

    - v ∈ C∞0 (Td)- g = 1: not in the weak-coupling regime

  • microscopic model

    • N+1 particle wave function on a d-dim. torus of length Λ

    Ψ ∈ L2(Td(N+1), ddx1...ddxN ddy)

    • non-relativistic Schrödinger equation

    i∂tΨt = HNΨt ⇒ Ψt = e−iHN tΨ0

    • microscopic Hamiltonian

    HN = −∆y −N∑k=1

    ∆xk + g

    N∑k=1

    v(xk − y)

    - v ∈ C∞0 (Td)- g = 1: not in the weak-coupling regime

  • microscopic model

    • N+1 particle wave function on a d-dim. torus of length Λ

    Ψ ∈ L2(Td(N+1), ddx1...ddxN ddy)

    • non-relativistic Schrödinger equation

    i∂tΨt = HNΨt ⇒ Ψt = e−iHN tΨ0

    • microscopic Hamiltonian

    HN = −∆y −N∑k=1

    ∆xk + g

    N∑k=1

    v(xk − y)

    - v ∈ C∞0 (Td)- g = 1: not in the weak-coupling regime

  • microscopic model

    • initial condition Ψ0 = φ0 ⊗ Ω0

    - φ0 ∈ H2(Td, ddy),∣∣∣∣φ0∣∣∣∣L2(Td) = 1

    - ground state of the ideal Fermi gas

    Ω0(x1, ..., xN ) =1√N !

    ∑σ∈SN

    (−1)σN∏i=1

    ϕσ(i)(xi)

    ϕk(x) = Λ− d

    2 exp(ipkx), pk = 2πΛ−1zk with N smallest zk ∈ Zd

    • Fermi momentum and average density : kF = Cdρ1d

  • table of contents

    1 Microscopic Model

    2 Main result and effective model

    3 Sketch of the Proof

    4 Explanation as a mean field type result

    5 Conclusion

  • main result and effective model

    • main result (similar for d = 1)

    Let d = 2 and � > 0. Denote average dens. by ρ = N/Λ2.Then there exists a positive constant C� s.t. for all t > 0

    limN,Λ→∞ρ=const.

    ∣∣∣∣e−iHN tΨ0 − e−iHmf tΨ0∣∣∣∣L2(Td(N+1)) ≤ C�t 32 ρ− 18 +�• effective Hamiltonian

    Hmf = −∆y −N∑k=1

    ∆xk + ρv̂(0)

    - V → E|Ω0|2 [V ](y) = 〈Ω0, V Ω0〉L2(TdN )(y) = ρv̂(0)

    - physically equivalent to free dynamics

  • main result and effective model

    • main result (similar for d = 1)

    Let d = 2 and � > 0. Denote average dens. by ρ = N/Λ2.Then there exists a positive constant C� s.t. for all t > 0

    limN,Λ→∞ρ=const.

    ∣∣∣∣e−iHN tΨ0 − e−iHmf tΨ0∣∣∣∣L2(Td(N+1)) ≤ C�t 32 ρ− 18 +�• effective Hamiltonian

    Hmf = −∆y −N∑k=1

    ∆xk + ρv̂(0)

    - V → E|Ω0|2 [V ](y) = 〈Ω0, V Ω0〉L2(TdN )(y) = ρv̂(0)

    - physically equivalent to free dynamics

  • main result and effective model

    • main result (similar for d = 1)

    Let d = 2 and � > 0. Denote average dens. by ρ = N/Λ2.Then there exists a positive constant C� s.t. for all t > 0

    limN,Λ→∞ρ=const.

    ∣∣∣∣e−iHN tΨ0 − e−iHmf tΨ0∣∣∣∣L2(Td(N+1)) ≤ C�t 32 ρ− 18 +�• effective Hamiltonian

    Hmf = −∆y −N∑k=1

    ∆xk + ρv̂(0)

    - V → E|Ω0|2 [V ](y) = 〈Ω0, V Ω0〉L2(TdN )(y) = ρv̂(0)

    - physically equivalent to free dynamics

  • main result and effective model

    • main result (similar for d = 1)

    Let d = 2 and � > 0. Denote average dens. by ρ = N/Λ2.Then there exists a positive constant C� s.t. for all t > 0

    limN,Λ→∞ρ=const.

    ∣∣∣∣e−iHN tΨ0 − e−iHmf tΨ0∣∣∣∣L2(Td(N+1)) ≤ C�t 32 ρ− 18 +�• effective Hamiltonian

    Hmf = −∆y −N∑k=1

    ∆xk + ρv̂(0)

    - V → E|Ω0|2 [V ](y) = 〈Ω0, V Ω0〉L2(TdN )(y) = ρv̂(0)

    - physically equivalent to free dynamics

  • table of contents

    1 Microscopic Model

    2 Main result and effective model

    3 Sketch of the Proof

    4 Explanation as a mean field type result

    5 Conclusion

  • sketch of the proof

    • general strategy

    - appropriate expansion of e−iHN tΨ0 around e−iHmf tΨ0

    - control of fluctuations by means of oscillatory integrals

    - proof relies highly on the anti-symmtry of Ω0

    - explicit propagation estimates: rate of convergence

  • sketch of the proof

    • repetitive application of Duhamel’s formula

    e−iHN tΨ0 =e−iHmf tΨ0 − i

    ∫ t0

    e−iHN (t−s)(V − ρv̂(0))e−iHmf sΨ0 ds

    • ∣∣∣∣e−iHN tΨ0 − e−iHmf tΨ0∣∣∣∣ ≤√∣∣∣∣ΨA(t)∣∣∣∣+ ∣∣∣∣ΨB(t)∣∣∣∣- first order contribution

    ΨA(t) =

    ∫ t0

    eiHmf s(V − ρv̂(0))e−iHmf sΨ0 ds

    - higher order contributions ΨB(t)

  • sketch of the proof

    • repetitive application of Duhamel’s formula

    e−iHN tΨ0 =e−iHmf tΨ0 − i

    ∫ t0

    e−iHN (t−s)(V − ρv̂(0))e−iHmf sΨ0 ds

    • ∣∣∣∣e−iHN tΨ0 − e−iHmf tΨ0∣∣∣∣ ≤√∣∣∣∣ΨA(t)∣∣∣∣+ ∣∣∣∣ΨB(t)∣∣∣∣- first order contribution

    ΨA(t) =

    ∫ t0

    eiHmf s(V − ρv̂(0))e−iHmf sΨ0 ds

    - higher order contributions ΨB(t)

  • sketch of the proof

    • controlling the norm of ΨA- too rough:

    ∣∣∣∣ΨA(t)∣∣∣∣2 ≤ t2Var[V ] = Ct2ρ 12- better:

    ∣∣∣∣ΨA(t)∣∣∣∣2 = N∑k=1

    ∞∑l=N+1

    |v̂(pk − pl)|2

    Λ2︸ ︷︷ ︸=Var[V ]

    ∣∣∣∣ ∫ t0

    ei(Ek−El)shkl(s)φ0︸ ︷︷ ︸oscillating integrand

    ds∣∣∣∣2

    with hkl(s) = e−i∆sei(pk−pl)yei∆s

    • fast decay of v̂ (Paley-Wiener ⇒ no large mom. transfer)• oscillation of ei(Ek−El)s for |pk| ≈ kF

    (Ek − El) = p2k − p2l = (|pk|+ |pl|)(|pk| − |pl|) ≥ kF · �(kF )

  • sketch of the proof

    • controlling the norm of ΨA- too rough:

    ∣∣∣∣ΨA(t)∣∣∣∣2 ≤ t2Var[V ] = Ct2ρ 12- better:

    ∣∣∣∣ΨA(t)∣∣∣∣2 = N∑k=1

    ∞∑l=N+1

    |v̂(pk − pl)|2

    Λ2︸ ︷︷ ︸=Var[V ]

    ∣∣∣∣ ∫ t0

    ei(Ek−El)shkl(s)φ0︸ ︷︷ ︸oscillating integrand

    ds∣∣∣∣2

    with hkl(s) = e−i∆sei(pk−pl)yei∆s

    • fast decay of v̂ (Paley-Wiener ⇒ no large mom. transfer)• oscillation of ei(Ek−El)s for |pk| ≈ kF

    (Ek − El) = p2k − p2l = (|pk|+ |pl|)(|pk| − |pl|) ≥ kF · �(kF )

  • sketch of the proof

    • controlling the norm of ΨA- too rough:

    ∣∣∣∣ΨA(t)∣∣∣∣2 ≤ t2Var[V ] = Ct2ρ 12- better:

    ∣∣∣∣ΨA(t)∣∣∣∣2 = N∑k=1

    ∞∑l=N+1

    |v̂(pk − pl)|2

    Λ2︸ ︷︷ ︸=Var[V ]

    ∣∣∣∣ ∫ t0

    ei(Ek−El)shkl(s)φ0︸ ︷︷ ︸oscillating integrand

    ds∣∣∣∣2

    with hkl(s) = e−i∆sei(pk−pl)yei∆s

    • fast decay of v̂ (Paley-Wiener ⇒ no large mom. transfer)• oscillation of ei(Ek−El)s for |pk| ≈ kF

    (Ek − El) = p2k − p2l = (|pk|+ |pl|)(|pk| − |pl|) ≥ kF · �(kF )

  • sketch of the proof

    • denote v̂a(pk − pl) = θ(ρa − 2πΛ−1|pk − pl|)v̂(pk − pl)

    ∣∣∣∣ΨA∣∣∣∣2 ≈ Ct2 [ N∑k=1

    ∞∑l=N+1

    +

    {stationary points}

    1

    k2F · �(kF )2N∑k=1

    ∞∑l=N+1

    ]|v̂a(pk − pl)|2

    Λ2

    • sufficient control

    limN,Λ→∞ρ=const.

    ∣∣∣∣ΨA(t)∣∣∣∣2 ≤ t2(Cρ− 14 + 3a2 + Caρ− 1a)

  • sketch of the proof

    • denote v̂a(pk − pl) = θ(ρa − 2πΛ−1|pk − pl|)v̂(pk − pl),

    ∣∣∣∣ΨA∣∣∣∣2 ≈ Ct2 [ N∑k=1

    ∞∑l=N+1

    +

    {stationary points}

    1

    k2F · �(kF )2N∑k=1

    ∞∑l=N+1

    ]|v̂a(pk − pl)|2

    Λ2

    • sufficient control

    limN,Λ→∞ρ=const.

    ∣∣∣∣ΨA(t)∣∣∣∣2 ≤ t2(Cρ− 14 + 3a2 + Caρ− 1a)

  • sketch of the proof

    • denote v̂a(pk − pl) = θ(ρa − 2πΛ−1|pk − pl|)v̂(pk − pl),

    ∣∣∣∣ΨA∣∣∣∣2 ≈ Ct2 [ N∑k=1

    ∞∑l=N+1

    +

    {stationary points}

    1

    k2F · �(kF )2N∑k=1

    ∞∑l=N+1

    ]|v̂a(pk − pl)|2

    Λ2

    • sufficient control

    limN,Λ→∞ρ=const.

    ∣∣∣∣ΨA(t)∣∣∣∣2 ≤ t2(Cρ− 14 + 3a2 + Caρ− 1a)

  • sketch of the proof

    ΨB =

    ∫ t0

    ∫ s10

    eiHNs2(V -ρv̂(0))eiHmf (s1-s2)(V -ρv̂(0))e-iHmf s1Ψ0 ds2 ds1

    • controlling the norm of ΨB- full time-evolution e−iHN t (need to reexpand)

    - immediate recollisions require energy renormalization

    Hmf = −∆y −N∑k=1

    ∆xk + ρv̂(0)− Ea(ρ)

    with Ea(ρ)/ρ→ 0 for ρ→∞

    • sufficient control at third order

    limN,Λ→∞ρ=const.

    ∣∣∣∣ΨB(t)∣∣∣∣2 ≤ t3(Cρ− 14 + 7a4 + Caρ− 1a)

  • sketch of the proof

    ΨB =

    ∫ t0

    ∫ s10

    eiHNs2(V -ρv̂(0))eiHmf (s1-s2)(V -ρv̂(0))e-iHmf s1Ψ0 ds2 ds1

    • controlling the norm of ΨB- full time-evolution e−iHN t (need to reexpand)

    - immediate recollisions require energy renormalization

    Hmf = −∆y −N∑k=1

    ∆xk + ρv̂(0)− Ea(ρ)

    with Ea(ρ)/ρ→ 0 for ρ→∞

    • sufficient control at third order

    limN,Λ→∞ρ=const.

    ∣∣∣∣ΨB(t)∣∣∣∣2 ≤ t3(Cρ− 14 + 7a4 + Caρ− 1a)

  • table of contents

    1 Microscopic Model

    2 Main result and effective model

    3 Sketch of the Proof

    4 Explanation as a mean field type result

    5 Conclusion

  • explanation as a mean field type result

    • fluctuations of V around ρv̂(0) in d dimensions

    Var[V ](y) = 〈Ω0, (V − ρv̂(0))2Ω0〉L2(TdN )(y) = Cdρd−1d

    - P|Ω0|2(∣∣∑N

    k=1 v(xk − y)− ρv̂(0)∣∣) ≈ 1 does not hold

    - random forces in the gas are large (strong coupling g = 1)

    • Comparison with bosons (symmetric initial Ω0)- fluctuations: VarB [V ] = Cdρ (due to

    √N law)

    - lifetime of φ0 decreases for increasing density ρ1

    1 Emission of Cerenkov radiation as a mechanism for Hamiltonianfriction, J. Fröhlich and Z. Gang, Advances in Mathematics 264(2014) 183–235

  • explanation as a mean field type result

    • fluctuations of V around ρv̂(0) in d dimensions

    Var[V ](y) = 〈Ω0, (V − ρv̂(0))2Ω0〉L2(TdN )(y) = Cdρd−1d

    - P|Ω0|2(∣∣∑N

    k=1 v(xk − y)− ρv̂(0)∣∣) ≈ 1 does not hold

    - random forces in the gas are large (strong coupling g = 1)

    • Comparison with bosons (symmetric initial Ω0)- fluctuations: VarB [V ] = Cdρ (due to

    √N law)

    - lifetime of φ0 decreases for increasing density ρ1

    1 Emission of Cerenkov radiation as a mechanism for Hamiltonianfriction, J. Fröhlich and Z. Gang, Advances in Mathematics 264(2014) 183–235

  • explanation as a mean field type result

    • first glance: mean field theory seems not to apply

    • closer look at fluctuations:

    - (V − ρv̂(0))Ψ0 = Λ−1∑Nk=1 V (pk)Ψ0 with

    V (pk)Ψ0 =

    ∞∑l=N+1

    v̂(pk − pl)Λ

    ei(pk−pl)ya∗(pk)a(pl)Ψ0

    - {V (pk)}Nk=1 family of uncorrelated random variables

    Var[V ] = Λ−2N∑k=1

    Var[V (pk)]

  • explanation as a mean field type result

  • explanation as a mean field type result

  • explanation as a mean field type result

  • table of contents

    1 Microscopic Model

    2 Main result and effective model

    3 Sketch of the Proof

    4 Explanation as a mean field type result

    5 Conclusion

  • conclusion

    • long lived resonance of φ0 emerges under e−iHN t (ρ→∞)- previously known in the physics literature (ion stopping)

    • open question for 3d model- applied method does not yield sufficient estimates

    • fermions: mean field regime � weak-coupling regime- compared to bosons: mean field regime = weak-coupling regime

    • further research- thermal initial states, ground states in periodic external potential,

    long time behaviour of φ0, similar result in the fully interacting case

  • conclusion

    • long lived resonance of φ0 emerges under e−iHN t (ρ→∞)- previously known in the physics literature (ion stopping)

    • open question for 3d model- applied method does not yield sufficient estimates

    • fermions: mean field regime � weak-coupling regime- compared to bosons: mean field regime = weak-coupling regime

    • further research- thermal initial states, ground states in periodic external potential,

    long time behaviour of φ0, similar result in the fully interacting case

  • conclusion

    • long lived resonance of φ0 emerges under e−iHN t (ρ→∞)- previously known in the physics literature (ion stopping)

    • open question for 3d model- applied method does not yield sufficient estimates

    • fermions: mean field regime � weak-coupling regime- compared to bosons: mean field regime = weak-coupling regime

    • further research- thermal initial states, ground states in periodic external potential,

    long time behaviour of φ0, similar result in the fully interacting case

  • conclusion

    • long lived resonance of φ0 emerges under e−iHN t (ρ→∞)- previously known in the physics literature (ion stopping)

    • open question for 3d model- applied method does not yield sufficient estimates

    • fermions: mean field regime � weak-coupling regime- compared to bosons: mean field regime = weak-coupling regime

    • further research- thermal initial states, ground states in periodic external potential,

    long time behaviour of φ0, similar result in the fully interacting case

  • Thank you for your attention

    Microscopic ModelMain result and effective modelSketch of the ProofExplanation as a mean field type resultConclusion