TheoremBase

The Galerkin Wick-Ordered Phi4Phi^4 Potential on the Torus is an Admissible Cylindrical Potential, with Semiconvexity Constant the Positive Part of Three Times the Coupling Times the Wick Constant Minus the Mass

At a cube cutoff, the Wick-ordered quartic plus Wick-ordered mass term of the Galerkin free field on the torus is an admissible cylindrical potential in the white-noise geometry; its Hessian is bounded below by the mass minus three times the coupling times the Wick constant, measured in the noise norm, which here is exactly the square-integrable norm of the field. The Wick constant is the explicit sum of the inverse Fourier weights over the cube, and it diverges in dimension at least two.

Statement

In the setting of The Free Field on the Torus as the Gaussian Reference Measure, with Square-Integrable White Noise: Standing Notation, so that X=H−m(Tn)X=H^{-m}(\mathbb{T}^{n}) with orthonormal basis (ej)j∈N(e_{j})_{j\in\mathbb{N}}, noise weights aja_{j}, variances cjc_{j} and reference measure γc\gamma_{c}, let xjx_{j} be the jj-th coordinate of x∈Xx\in X and pd:X→Rdp_{d}:X\to\mathbb{R}^{d} (d∈Nd\in\mathbb{N}) the coordinate map, as in A Diagonal Gaussian Reference Measure on the Noise Wasserstein Space, Rescaled Heads and Gaussian Tails: Standing Notation §background; XaX^{a} and ∣⋅∣a|\cdot|_{a} are the noise space and noise norm, described by White Noise in the Square-Integrable Space and the Free Field on the Torus as Noise Weights and a Variance Sequence on a Negative Sobolev Space §noise-space. Let N∈NN\in\mathbb{N}, and let ϰ,m∈R\varkappa,\mathfrak{m}\in\mathbb{R} satisfy 0<ϰ0<\varkappa. The notation of The Galerkin Wick-Ordered Phi^4 Potential and Its Wick Constant on the Torus is used for this cutoff, coupling and mass: the maps ψk\psi_{k} and numbers μk−1\mu_{k}^{-1} (k∈Znk\in\mathbb{Z}^{n}), the measure λQ\lambda_{Q} on the cell QQ, the numbers aj−1/2a_{j}^{-1/2} and the value x(k)x(k) of the coefficient family x∈Xx\in X at kk, as fixed in its preamble; the cube ΓN\Gamma_{N}; the Galerkin head dimension dNd_{N}, with the indicator χN\chi_{N}, the set JNJ_{N} of active indices (those j∈[dN]j\in[d_{N}] with κ(j)∈ΓN\kappa(j)\in\Gamma_{N}) and the convention ∑j∈JNtj=∑j=1dNχN(j) tj\sum_{j\in J_{N}}t_{j}=\sum_{j=1}^{d_{N}}\chi_{N}(j)\,t_{j} of that clause; the Galerkin fields ΦN[x]\Phi_{N}[x] (x∈Xx\in X) and φN[u]\varphi_{N}[u] (u∈RdNu\in\mathbb{R}^{d_{N}}); the Wick constant CNC_{N}; the Wick polynomial fNf_{N}; and the Galerkin Wick-ordered ϕ4\phi^{4} potential VN:X→RV_{N}:X\to\mathbb{R} with its profile vN:RdN→Rv_{N}:\mathbb{R}^{d_{N}}\to\mathbb{R}. For N′∈NN'\in\mathbb{N}, CN′C_{N'} is the Wick constant at cutoff N′N'. Sums over the finite set ΓN\Gamma_{N} are sums over a finite index set. Functions of class C2C^{2} on RdN\mathbb{R}^{d_{N}} are those of C^k Maps on a Euclidean Open Set, with the partial derivatives ∂jv\partial_{j}v and ∂j∂kv\partial_{j}\partial_{k}v; max⁡{s,s′}\max\{s,s'\} is the maximum of s,s′∈Rs,s'\in\mathbb{R}. Put

KN=max⁡{3ϰCN−m,0}.K_{N}=\max\{3\varkappa C_{N}-\mathfrak{m},0\}.

1. (The Wick constant) 0<CN0<C_{N}, and

CN=∑k∈ΓNμk−1=∑j∈JNcjaj.C_{N}=\sum_{k\in\Gamma_{N}}\mu_{k}^{-1}=\sum_{j\in J_{N}}\frac{c_{j}}{a_{j}} .

For every z∈Rnz\in\mathbb{R}^{n},

∑j∈JNμκ(j)−1 ψκ(j)(z)2=CN;\sum_{j\in J_{N}}\mu_{\kappa(j)}^{-1}\,\psi_{\kappa(j)}(z)^{2}=C_{N};

moreover the functions x↦ΦN[x](z)x\mapsto\Phi_{N}[x](z) and x↦ΦN[x](z)2x\mapsto\Phi_{N}[x](z)^{2} on XX are integrable with respect to γc\gamma_{c}, with

∫XΦN[x](z) γc(dx)=0,∫XΦN[x](z)2 γc(dx)=CN.\int_{X}\Phi_{N}[x](z)\,\gamma_{c}(dx)=0,\qquad\int_{X}\Phi_{N}[x](z)^{2}\,\gamma_{c}(dx)=C_{N}.

2. (Coordinates and the noise norm) For every x∈Xx\in X and j∈[dN]j\in[d_{N}], x(κ(j))=aj−1/2xjx(\kappa(j))=a_{j}^{-1/2}x_{j}; hence ΦN[x]=φN[pdN(x)]\Phi_{N}[x]=\varphi_{N}[p_{d_{N}}(x)] for every x∈Xx\in X, and VN=vN∘pdNV_{N}=v_{N}\circ p_{d_{N}}. The profile depends only on the active coordinates: if u,u′∈RdNu,u'\in\mathbb{R}^{d_{N}} satisfy uj=uj′u_{j}=u'_{j} for every j∈JNj\in J_{N}, then φN[u]=φN[u′]\varphi_{N}[u]=\varphi_{N}[u'] and vN(u)=vN(u′)v_{N}(u)=v_{N}(u'). For every h∈RdNh\in\mathbb{R}^{d_{N}} the class U=[ φN[h]∣Q ]∈L2(Tn)U=[\,\varphi_{N}[h]|_{Q}\,]\in L^{2}(\mathbb{T}^{n}) satisfies U^=∑j∈JNhjej\hat{U}=\sum_{j\in J_{N}}h_{j}e_{j}, and

∫QφN[h](z)2 λQ(dz)=∥U∥L22=∣∑j∈JNhjej∣a2=∑j∈JNhj2aj ≤ ∑k=1dNhk2ak.\int_{Q}\varphi_{N}[h](z)^{2}\,\lambda_{Q}(dz)=\lVert U\rVert_{L^{2}}^{2}=\Bigl|\sum_{j\in J_{N}}h_{j}e_{j}\Bigr|_{a}^{2}=\sum_{j\in J_{N}}\frac{h_{j}^{2}}{a_{j}}\ \le\ \sum_{k=1}^{d_{N}}\frac{h_{k}^{2}}{a_{k}} .

3. (The Hessian) vNv_{N} is of class C2C^{2} on RdN\mathbb{R}^{d_{N}}. For every j∈[dN]j\in[d_{N}] with κ(j)∉ΓN\kappa(j)\notin\Gamma_{N} and every k∈[dN]k\in[d_{N}], the partial derivatives ∂jvN\partial_{j}v_{N}, ∂j∂kvN\partial_{j}\partial_{k}v_{N} and ∂k∂jvN\partial_{k}\partial_{j}v_{N} vanish identically on RdN\mathbb{R}^{d_{N}}. For all u,h∈RdNu,h\in\mathbb{R}^{d_{N}},

∑j=1dN∑k=1dN∂j∂kvN(u) hjhk=∫Q(3ϰ φN[u](z)2+m−3ϰCN) φN[h](z)2 λQ(dz) ≥ −KN∑j∈JNhj2aj ≥ −KN∑k=1dNhk2ak.\sum_{j=1}^{d_{N}}\sum_{k=1}^{d_{N}}\partial_{j}\partial_{k}v_{N}(u)\,h_{j}h_{k}=\int_{Q}\bigl(3\varkappa\,\varphi_{N}[u](z)^{2}+\mathfrak{m}-3\varkappa C_{N}\bigr)\,\varphi_{N}[h](z)^{2}\,\lambda_{Q}(dz)\ \ge\ -K_{N}\sum_{j\in J_{N}}\frac{h_{j}^{2}}{a_{j}}\ \ge\ -K_{N}\sum_{k=1}^{d_{N}}\frac{h_{k}^{2}}{a_{k}} .

4. (Admissibility) 0≤KN0\le K_{N}, and VNV_{N} is an admissible cylindrical potential with head dimension dNd_{N}, profile vNv_{N} and semiconvexity constant KNK_{N} in the noise norm.

5. (The noise gradient) For every x∈Xx\in X, the noise gradient of VNV_{N} is

∇aVN(x)=∑j∈JN(∫Q(ϰ(ΦN[x](z)3−3CNΦN[x](z))+m ΦN[x](z))ψκ(j)(z) λQ(dz))E^κ(j).\nabla_{a}V_{N}(x)=\sum_{j\in J_{N}}\Bigl(\int_{Q}\Bigl(\varkappa\bigl(\Phi_{N}[x](z)^{3}-3C_{N}\Phi_{N}[x](z)\bigr)+\mathfrak{m}\,\Phi_{N}[x](z)\Bigr)\psi_{\kappa(j)}(z)\,\lambda_{Q}(dz)\Bigr)\hat{E}_{\kappa(j)} .

6. (The Wick constant diverges) Suppose 2≤n2\le n. Then for every M∈RM\in\mathbb{R} there is N0∈NN_{0}\in\mathbb{N} with M<CN′M<C_{N'} for every N′∈NN'\in\mathbb{N} with N0≤N′N_{0}\le N'.

Proofs

Log in to submit a proof.

Loading...

Citations

Loading…

Dependencies

Loading…

Related

0 relations

Curated associations between results. These are editable and subjective — they do not replace the dependency graph, which is derived from the references in the text.

No relations recorded yet.

Comments

Log in to comment.

Loading…