We present a proof of skeleton inequalities for ferromagnetic lattice spin systems with potential V2)=(a/2)φ2+∑Mn=22n/ (2n)!}φ2n (a real, λ2n≥0) generalizing the Brydges–Fröhlich–Sokal and Bovier–Felder methods. As an application of the inequalities, we prove that, for sufficiently soft systems in d>4 dimensions, critical exponents γ, α , and Δ4 take their mean‐field values (i.e., γ=1, α=0, and Δ4= (3)/(2) ).

1.
D.
Brydges
,
J.
Fröhlich
, and
A. D.
Sokal
,
Commun. Math. Phys.
91
,
117
(
1983
).
2.
D.
Brydges
,
J.
Fröhlich
, and
A. D.
Sokal
,
Commun. Math. Phys.
91
,
141
(
1983
).
3.
A. D.
Sokal
,
Ann. Inst. H. Poincaré A
37
,
317
(
1982
).
4.
M.
Aizenman
,
Phys. Rev. Lett.
47
,
1
(
1981
).
5.
M.
Aizenman
,
Commun. Math. Phys.
86
,
1
(
1982
).
6.
M.
Aizenman
and
R.
Graham
,
Nucl. Phys. B
225
[FS9],
261
(
1983
).
7.
C.
Aragão De Carvalho
,
S.
Caracciolo
, and
J.
Fröhlich
,
Nucl. Phys. B
215
[FS7],
209
(
1983
).
8.
J.
Fröhlich
,
Nucl. Phys. B
200
[FS4],
281
(
1982
).
9.
T.
Hattori
,
J. Math. Phys.
24
,
2200
(
1983
).
10.
A.
Bovier
and
G.
Felder
,
Commun. Math. Phys.
93
,
259
(
1984
).
11.
The inequalities are also useful for the construction of continuum field theories in three dimensions with bare Lagrangian containing terms like ρ2nφ2n. Here the bare couplings ρ2n are adjusted to order εn−2 where ε is the lattice spacing. (For the details, see T. Hara, “Construction of a Nontrivial Field Theory in 3 dimensions Starting from a Lagrangian of φ6 type” University of Tokyo‐Komaba 84‐15).
12.
D. Ruelle, Statistical Mechanics—Rigorous Results (Benjamin, New York, 1969).
13.
D. Brydges, “Field theories and Symanzik’s polymer representation,” in Gauge Theories—Fundamental Interactions and Rigorous Results (Proceedings of the 1981 Poiana Brasov Summer School) edited by G. Dita, V. Georgescu, and R. Purice (Birkhäuser, Boston, 1982).
14.
D.
Brydges
,
J.
Fröhlich
, and
T.
Spencer
,
Commun. Math. Phys.
83
,
123
(
1982
).
15.
F.
Dunlop
and
C. M.
Newman
,
Commun. Math. Phys.
44
,
223
(
1975
).
16.
J.
Ginibre
,
Commun. Math. Phys.
16
,
310
(
1970
).
17.
R. B.
Griffiths
,
J. Math. Phys.
8
,
478
,
484
(
1967
).
18.
G. S.
Sylvester
,
J. Stat. Phys.
15
,
327
(
1976
).
19.
Here, and in each following step, we construct not only the N th‐order inequality, but (formally) all the inequalities of order M⩽N−1, so that these (N+1) inequalities together form a set of alternating bounds …⩽σ(−λ)k These inequalities for M⩽N−1 actually turn out to be identical to those used in the inductive proof. See Sec. IIC.
20.
Of course this choice must be done without breaking the ferromagnetic condition Jxy⩾0 (x≠y). This will be satisfied if (for example) we first set Jxy = J>0 for all x≠y and ax = a sufficiently large, and vary the Cxy ’s in some bounded region including the initial values.
21.
J.
Fröhlich
,
B.
Simon
, and
T.
Spencer
,
Commun. Math. Phys.
50
,
79
(
1976
).
22.
J.
Fröhlich
,
R.
Israel
,
E. H.
Lieb
, and
B.
Simon
,
Commun. Math. Phys.
62
,
1
(
1978
).
23.
J.
Fröhlich
,
R.
Israel
,
E. H.
Lieb
, and
B.
Simon
,
J. Stat. Phys.
22
,
297
(
1980
).
24.
J.
Rosen
,
Adv. Appl. Math.
1
,
37
(
1980
).
25.
As for the present definition26 of α. which reflects the behavior of full specific heat rather than its singular part, the inequality α>0 is valid for arbitrary systems in Zd(d⩾3) with Griffiths‐class a priori measure. To prove the inequality, assume the converse, i.e., C(J)→0 as J→Jc. Then from the Griffiths II inequalities, we have 〈φ0φ1crit = 〈φ02crit. (Hereafter 〈…〉crit denotes limJ→Jc〈…〉J.) Inserting this into the Gell‐Mann‐Low representation3〈φ0φ(n,0,0,…)crit = (Φ0Ω,TΦ0Ω) (T is a Hermitian operator with |T|⩽1), we have 0Ω = Φ0Ω and thus 〈φ0φ(n,0,0,…)crit = 〈φ02crit for all n. This contradicts with the infrared bounds3,21,22〈φ0φxcritconst |x|2−d. We are grateful to the referee for suggesting to us the present proof.
26.
A. D.
Sokal
,
Phys. Lett.
71
,
451
(
1979
).
27.
One can write down this smallness condition in the form
28.
Actually, Eq. (3.12) [and all the other similar relations such as Eq. (3.14)] must be proved in a finite lattice (with periodic boundary condition) for the corresponding quantities such as χΛ≡Σx∈Λ〈φ0φxΛ. If one “works hard2” using some correlation inequalities (e.g., the Simon‐Lieb inequality14,29,30 and the first‐order skeleton inequality), one can prove that these finite volume quantities converge to the infinite volume quantities (3.2)–(3.4) in the high‐temperature region (characterized by χ<∞), and the relations carry over to the infinite systems.
29.
E. H.
Lieb
,
Commun. Math. Phys.
77
,
127
(
1980
).
30.
B.
Simon
,
Commun. Math. Phys.
77
,
111
(
1980
).
31.
To prove this bound note that J<a/(2d), a Gaussian system defined by letting 2n = 0(n⩾2) is well defined. Then the Griffiths II inequality implies χ⩽χGussian<∞ and the system is in its high‐temperature region.
32.
J. L.
Lebowitz
,
Commun. Math. Phys.
35
,
87
(
1974
).
33.
J.
Glimm
and
A.
Jaffe
,
Phys. Rev. D
10
,
536
(
1974
).
This content is only available via PDF.
You do not currently have access to this content.