TheoremBase

At a bounded position the Hamiltonian is read off from the operator tuple of the position. For the momentum alpha(X-Y) both commutators equal alpha times the commutator of the two position operators, so only the running cost differs; the commutator bound gives a Lipschitz estimate in the momentum.

Proof

Each result cited below is universally quantified over the data in its own statement.

Claim 1 (Evaluation). Since sΩ=Xs\Omega=X, the pair (X,P)(X,P) of Square-Integrable Tuples in a Tracial W*-Probability Space: Their Norm, Affine Images, Pairs, Embedded Images and Laws §operations equals (sΩ,P)(s\Omega,P), so π=law(X,P)\pi=\mathrm{law}(X,P) satisfies π=law(sΩ,P)\pi=\mathrm{law}(s\Omega,P) and the first case of The Controlled von Neumann Hamiltonian on Phase-Space Noncommutative Laws §hamiltonian applies with the space (H,M,Ω)(H,M,\Omega), the tuple ss and the momentum PP. By Plan Jets and Hamiltonians on Square-Integrable Noncommutative Laws: Standing Notation §lifts, HMvN(X,P)=HvN(π)\mathcal{H}^{\mathrm{vN}}_{M}(X,P)=\mathcal{H}^{\mathrm{vN}}(\pi), hence

HMvN(X,P)=12∑j=1d∥[Pj,sj]∥2−f(pr#1π).\mathcal{H}^{\mathrm{vN}}_{M}(X,P)=\frac{1}{2}\sum_{j=1}^{d}\lVert[P_{j},s_{j}]\rVert^{2}-f(\mathrm{pr}^{1}_{\#}\pi).

By Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §push-forward, pr#1π=law(pr1(X,P))\mathrm{pr}^{1}_{\#}\pi=\mathrm{law}(\mathrm{pr}^{1}(X,P)). The coordinate datum pr1\mathrm{pr}^{1} of Affine Data and Affine Substitutions of Noncommutative Polynomials §coordinate has zero shift and coefficient matrix with entries 11 at the positions (i,i)(i,i), i∈[d]i\in[d], and 00 elsewhere, so the affine image of Square-Integrable Tuples in a Tracial W*-Probability Space: Their Norm, Affine Images, Pairs, Embedded Images and Laws §operations gives (pr1(X,P))i=Xi(\mathrm{pr}^{1}(X,P))_{i}=X_{i} for i∈[d]i\in[d], that is, pr1(X,P)=X\mathrm{pr}^{1}(X,P)=X. Hence f(pr#1π)=f(law(X))f(\mathrm{pr}^{1}_{\#}\pi)=f(\mathrm{law}(X)), which is the claimed formula.

Claim 2 (Structure condition). Assume ff is uniformly continuous. Let R>0R>0 and η>0\eta>0 be real. By Uniformly Continuous Map Between Metric Spaces, with the metrics of Plan Jets and Hamiltonians on Square-Integrable Noncommutative Laws: Standing Notation §metrics, there is a real r>0r>0 such that ∣f(μ)−f(ν)∣<η|f(\mu)-f(\nu)|<\eta for all μ,ν∈Σd2\mu,\nu\in\Sigma^{2}_{d} with W^2(μ,ν)<r\widehat{W}_{2}(\mu,\nu)<r; we show that this rr has the property required in The Structure Condition at Bounded Positions for a Hamiltonian on Phase-Space Noncommutative Laws §structure. Let (H,M,Ω)(H,M,\Omega) be a tracial W*-probability space, let X,YX,Y be L2L^{2} dd-tuples with law(X),law(Y)∈κd(Σd,R)\mathrm{law}(X),\mathrm{law}(Y)\in\kappa_{d}(\Sigma_{d,R}), and let α>0\alpha>0 be real with α∥X−Y∥22+∥X−Y∥2<r\alpha\lVert X-Y\rVert_{2}^{2}+\lVert X-Y\rVert_{2}<r. By Square-Integrable Tuples with a Bounded Law are Vacuum Tuples of Bounded Self-Adjoint Operators §operator there are self-adjoint dd-tuples ss and tt in MM with sΩ=Xs\Omega=X and tΩ=Yt\Omega=Y. Put P=α(X−Y)P=\alpha(X-Y), the real multiple of Plan Jets and Hamiltonians on Square-Integrable Noncommutative Laws: Standing Notation §pairing; it is an L2L^{2} dd-tuple with Pj=α(Xj−Yj)P_{j}=\alpha(X_{j}-Y_{j}). For j∈[d]j\in[d], Pj=α(sjΩ−tjΩ)=bjΩP_{j}=\alpha(s_{j}\Omega-t_{j}\Omega)=b_{j}\Omega with bj=α(sj−tj)∈Mb_{j}=\alpha(s_{j}-t_{j})\in M. By The Commutator of a Square-Integrable Vector with a Bounded Self-Adjoint Operator: Linearity, the Norm Bound, Vacuum Vectors and Dependence on the Law Only §vacuum,

[Pj,tj]=(bjtj−tjbj)Ω=α(sjtj−tjsj)Ω,[Pj,sj]=(bjsj−sjbj)Ω=α(sjtj−tjsj)Ω,[P_{j},t_{j}]=(b_{j}t_{j}-t_{j}b_{j})\Omega=\alpha(s_{j}t_{j}-t_{j}s_{j})\Omega,\qquad[P_{j},s_{j}]=(b_{j}s_{j}-s_{j}b_{j})\Omega=\alpha(s_{j}t_{j}-t_{j}s_{j})\Omega,

since tjtjt_{j}t_{j} and sjsjs_{j}s_{j} cancel in the respective differences. Thus [Pj,tj]=[Pj,sj][P_{j},t_{j}]=[P_{j},s_{j}] for every j∈[d]j\in[d], and Claim 1, applied at XX with ss and at YY with tt, gives

HMvN(Y,P)−HMvN(X,P)=f(law(X))−f(law(Y)).\mathcal{H}^{\mathrm{vN}}_{M}(Y,P)-\mathcal{H}^{\mathrm{vN}}_{M}(X,P)=f(\mathrm{law}(X))-f(\mathrm{law}(Y)).

Since α∥X−Y∥22≥0\alpha\lVert X-Y\rVert_{2}^{2}\ge0, we have ∥X−Y∥2<r\lVert X-Y\rVert_{2}<r, and W^2(law(X),law(Y))≤∥X−Y∥2\widehat{W}_{2}(\mathrm{law}(X),\mathrm{law}(Y))\le\lVert X-Y\rVert_{2} by Calculus of Laws of Square-Integrable Tuples: Bounded Tuples, the Lipschitz Bound, Moments, Affine Push-Forwards, Couplings and Embeddings §lipschitz. Hence f(law(X))−f(law(Y))<ηf(\mathrm{law}(X))-f(\mathrm{law}(Y))<\eta, which is the required inequality.

Claim 3 (Momentum continuity). Let R>0R>0 and η>0\eta>0 be real and put r=η/(4dR3)r=\eta/(4dR^{3}). Let (H,M,Ω)(H,M,\Omega) be a tracial W*-probability space, let XX be an L2L^{2} dd-tuple with law(X)∈κd(Σd,R)\mathrm{law}(X)\in\kappa_{d}(\Sigma_{d,R}), with ss as in Claim 1, so that ∥sj∥op≤R\lVert s_{j}\rVert_{\mathrm{op}}\le R by Square-Integrable Tuples with a Bounded Law are Vacuum Tuples of Bounded Self-Adjoint Operators §bound, and let P,P′P,P' be L2L^{2} dd-tuples with ∥P∥2≤R\lVert P\rVert_{2}\le R, ∥P′∥2≤R\lVert P'\rVert_{2}\le R and ∥P−P′∥2<r\lVert P-P'\rVert_{2}<r. By Claim 1 at XX for PP and for P′P',

HMvN(X,P)−HMvN(X,P′)=12∑j=1d(∥aj∥2−∥aj′∥2),aj=[Pj,sj],aj′=[Pj′,sj].\mathcal{H}^{\mathrm{vN}}_{M}(X,P)-\mathcal{H}^{\mathrm{vN}}_{M}(X,P')=\frac{1}{2}\sum_{j=1}^{d}\Bigl(\lVert a_{j}\rVert^{2}-\lVert a'_{j}\rVert^{2}\Bigr),\qquad a_{j}=[P_{j},s_{j}],\quad a'_{j}=[P'_{j},s_{j}].

By The Commutator of a Square-Integrable Vector with a Bounded Self-Adjoint Operator: Linearity, the Norm Bound, Vacuum Vectors and Dependence on the Law Only §linear, aj−aj′=[Pj−Pj′,sj]a_{j}-a'_{j}=[P_{j}-P'_{j},s_{j}], so The Commutator of a Square-Integrable Vector with a Bounded Self-Adjoint Operator: Linearity, the Norm Bound, Vacuum Vectors and Dependence on the Law Only §bound gives ∥aj−aj′∥≤2R∥Pj−Pj′∥\lVert a_{j}-a'_{j}\rVert\le2R\lVert P_{j}-P'_{j}\rVert, ∥aj∥≤2R∥Pj∥\lVert a_{j}\rVert\le2R\lVert P_{j}\rVert and ∥aj′∥≤2R∥Pj′∥\lVert a'_{j}\rVert\le2R\lVert P'_{j}\rVert. By the definition of the L2L^{2} norm in Square-Integrable Tuples in a Tracial W*-Probability Space: Their Norm, Affine Images, Pairs, Embedded Images and Laws §tuples, ∥Pj∥≤∥P∥2≤R\lVert P_{j}\rVert\le\lVert P\rVert_{2}\le R, ∥Pj′∥≤R\lVert P'_{j}\rVert\le R and ∥Pj−Pj′∥≤∥P−P′∥2\lVert P_{j}-P'_{j}\rVert\le\lVert P-P'\rVert_{2}. The triangle inequality of The Induced Norm is a Norm, and Induces a Metric gives ∣∥aj∥−∥aj′∥∣≤∥aj−aj′∥\bigl|\lVert a_{j}\rVert-\lVert a'_{j}\rVert\bigr|\le\lVert a_{j}-a'_{j}\rVert, hence

∣∥aj∥2−∥aj′∥2∣=∣∥aj∥−∥aj′∥∣(∥aj∥+∥aj′∥)≤2R∥P−P′∥2⋅4R2=8R3∥P−P′∥2.\bigl|\lVert a_{j}\rVert^{2}-\lVert a'_{j}\rVert^{2}\bigr|=\bigl|\lVert a_{j}\rVert-\lVert a'_{j}\rVert\bigr|\bigl(\lVert a_{j}\rVert+\lVert a'_{j}\rVert\bigr)\le2R\lVert P-P'\rVert_{2}\cdot4R^{2}=8R^{3}\lVert P-P'\rVert_{2}.

Summing over j∈[d]j\in[d] and halving, ∣HMvN(X,P)−HMvN(X,P′)∣≤4dR3∥P−P′∥2<4dR3r=η\bigl|\mathcal{H}^{\mathrm{vN}}_{M}(X,P)-\mathcal{H}^{\mathrm{vN}}_{M}(X,P')\bigr|\le4dR^{3}\lVert P-P'\rVert_{2}<4dR^{3}r=\eta. This is the property of Hamiltonians on Phase-Space Noncommutative Laws that are Uniformly Continuous in the Momentum at Bounded Positions §momentum.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…