JOURNAL OF MATHEMATICAL PHYSICS 51, 053306 共2010兲 Power series representations for complex bosonic effective actions. II. A small field renormalization group flow Tadeusz Balaban,1,a兲 Joel Feldman,2,b兲 Horst Knörrer,3,c兲 and Eugene Trubowitz3,d兲 1 Department of Mathematics Rutgers, The State University of New Jersey, 110 Frelinghuysen Rd., Piscataway, New Jersey 08854-8019, USA 2 Department of Mathematics, University of British Columbia, Vancouver, British Columbia V6T 1Z2, Canada 3 Mathematik, ETH-Zentrum, CH-8092 Zurich, Switzerland 共Received 27 October 2009; accepted 28 January 2010; published online 21 May 2010兲 In a previous paper, we developed a power series representation and estimates for ⴱ ⴱ an effective action of the form ln关兰e f共␣1,. . .,␣s;z ,z兲d共zⴱ , z兲 / 兰e f共0,. . .,0;z ,z兲d共zⴱ , z兲兴. Here, f共␣1 , . . . , ␣s ; zⴱ , z兲 is an analytic function of the complex fields ␣1共x兲 , . . . , ␣s共x兲 , zⴱ共x兲 , z共x兲 indexed by x in a finite set X and d共zⴱ , z兲 is a compactly supported product measure. Such effective actions occur in the small field region for a renormalization group analysis. We illustrate the technique by a model renormalization group flow motivated by the ultraviolet regime in many boson systems. © 2010 American Institute of Physics. 关doi:10.1063/1.3329938兴 I. INTRODUCTION Consider the grand canonical partition function, at temperature T and chemical potential , for a many boson system moving in a metric space X with a finite number of points and metric d. Suppose that the Hamiltonian H is the sum of a single particle operator 共for example, the discrete Laplacian兲 with kernel h共x , y兲 and a two body operator given by a real, symmetric, repulsive pair potential 2v共x , y兲. In Ref. 2 共Theorem 2.2兲, we proved the functional integral representation Tr e−共1/kT兲共H−N兲 = lim →0 冕 兿 关d˜ R 共␣ⴱ, ␣t兲共␣−, ␣兲e具␣ 苸Z艚共0,1/kT兴 ⴱ ⴱ ⴱ −,j共兲␣典−具␣−␣v␣−␣典 兴 共1.1兲 for the partition function, under the convention that ␣0 = ␣1/kT and the limit → 0 is restricted to ’s dividing 1 / kT 关that is, 1 / 苸 共kT兲N兴. Here, N is the number operator and, for any r ⬎ 0, ˜ r共 ␣ ⴱ, ␣ 兲 = d 兿 x苸X d␣ⴱ共x兲 ∧ d␣共x兲 −␣ⴱ共x兲␣共x兲 e 共兩␣共x兲兩 ⬍ r兲 2ı denotes the unnormalized Gaussian measure, cutoff at radius r, and 共␣ , 兲 is the characteristic function of a兲 Electronic mail: tbalaban@math.rutgers.edu. Electronic mail: feldman@math.ubc.ca. URL: http://www.math.ubc.ca/~feldman/. c兲 Electronic mail: knoerrer@math.ethz.ch. URL: http://www.math.ethz.ch/~knoerrer/. d兲 Electronic mail: trub@math.ethz.ch. b兲 0022-2488/2010/51共5兲/053306/20/$30.00 51, 053306-1 © 2010 American Institute of Physics Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-2 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. 兵␣, :X → C兩 储␣ − 储⬁ ⬍ p0共兲其. The cutoffs R and p0共兲 grow at an appropriate rate as → 0. 关One can think of R as growing a bit faster than 1 / 冑4 and of p0共兲 as a power of ln共1 / 兲 or a very small power of 1 / .兴 Furthermore, for any ⬎ 0, the operator j共兲 = e−共h−兲. We write the 共R-style兲 scalar product, 具f , g典 = 兺x苸X f共x兲g共x兲 for any two fields f , g : X → C. 共Thus, the usual scalar product over C兩X兩 is 具f ⴱ , g典.兲 The representation 共1.1兲 is the first part of a program to resolve the mathematical difficulties inherent in the time-ultraviolet limit of the formal, coherent state functional integral representation for the partition function and correlation functions of many boson systems 关see Ref. 5, 共2.66兲 and the discussion in the introduction to Ref. 1兴. In Ref. 4 this program is completed using techniques of renormalization group analysis. This paper is a description of the “small field part” of that construction. In Ref. 4 we obtain a representation of the functional integral of 共1.1兲 which can be used to analyze infrared problems. We do so by applying a simple version of a renormalization group procedure, namely, “decimation.” In each decimation step we integrate out every “second” variable. In the first step, we integrate out ␣⬘ with ⬘ = , 3 , 5 , . . .. The integral with respect to these variables factorizes into the product, over = 2 , 4 , 6 , . . ., of the independent integrals 冕 ⴱ ⴱ ⴱ ˜ R 共␣ⴱ−, ␣−兲共␣−2, ␣−兲e具␣−2,j共兲␣−典−具␣−2␣−v␣−2␣−典 d ⴱ ⴱ ⴱ ⫻e具␣−,j共兲␣典−具␣−␣,v␣−␣典共␣−, ␣兲. That is, assuming that 1 / kT 苸 2N, 冕 兿 关d˜ R 共␣ⴱ, ␣兲共␣−, ␣兲e具␣ 苸Z艚共0,1/kT兴 = 冕 兿 苸2Z艚共0,1/kT兴 ⴱ ⴱ ⴱ −,j共兲␣典−具␣−␣v␣−␣典 兴 ˜ R 共␣ⴱ, ␣兲I1共; ␣ⴱ−2, ␣兲, d where I1共; ␣ⴱ, 兲 = 冕 ˜ R 共ⴱ, 兲共␣, 兲e具␣ d ⴱ,j共兲典+具ⴱ,j共兲典 −共具␣ⴱ, ␣ⴱ典+具ⴱ, ⴱ典兲 v v e 共, 兲. 共1.2兲 After n − 1 additional decimation steps we will have integrated out those ␣’s with 苸 共Z \ 共2n兲Z兲 艚 共0 , 1 / kT兴 leaving an integrand which is a function of the ␣’s with 1 兴. Thus, for kT1 苸 2nN, we write 苸 共2n兲Z 艚 关0 , kT 冕 ˜ R 共␣ⴱ, ␣兲共␣−, ␣兲e具␣ 关d 兿 苸Z艚共0,1/kT兴 = 冕 兿 ⴱ ⴱ ⴱ −,j共兲␣典−具␣−␣v␣−␣典 ⴱ 苸共2n兲Z艚共0,1/kT兴 ˜ R 共␣ⴱ, ␣兲In共, ␣−2n, ␣兲, d 兴 共1.3兲 where the functions In共 ; ␣ⴱ , 兲 are recursively defined by 共1.2兲 and In+1共; ␣ⴱ, 兲 = 冕 ˜ R 共ⴱ, 兲In共; ␣ⴱ, 兲In共; ⴱ, 兲. d 共1.4兲 The main result of Ref. 4 is the construction and description of a functional I共␣ⴱ , 兲, defined for 苸 共0 , ⌰兴 关where ⌰ = O共1兲, independent of v兴 such that Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-3 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II I共␣ⴱ, 兲 = lim Im共2−m ; ␣ⴱ, 兲. m→⬁ Then, for any p such that 1 / pkT 苸 共0 , ⌰兴, 冕 兿冋兿 p Tr e−共1/kT兲共H−N兲 = n=1 x苸X 册 dn共x兲ⴱn共x兲 − 共x兲ⴱ 共x兲 ⴱ n e n I共1/pkT兲共n−1 , n兲 2ı 共1.5兲 with the convention 0 = p. 共1.5兲 can be the starting point for an infrared analysis. In Ref. 4 we describe the functions In共 ; ␣ⴱ , 兲 and I共␣ⴱ , 兲 as sums over “large/small field decompositions” of X. The dominant part in the large/small field decomposition is called the “pure small field part” and is obtained by replacing the full integrals 共1.4兲 with integrals over appropriate neighborhoods of stationary points 共see Sec. II兲. In this note, we discuss a toy model, which we call the “stationary phase 共SP兲 approximation,” in which all domains of integration are restricted, simply by fiat, to neighborhoods of stationary points. These neighborhoods will be measured by radii r共␦兲, where ␦ 哫 r共␦兲 is a positive and monotonically decreasing function. We now give a description of the SP approximation and derive estimates for it. To do this, introduce the notation n = 2n and V␦共; ␣ⴱ, 兲 = − 具关j共兲␣ⴱ兴关j共␦ − − 兲兴, v关j共兲␣ⴱ兴关j共␦ − − 兲兴典. 兺 苸Z艚关0,␦兲 共1.6兲 It will turn out that there is a function E␦共 ; ␣ⴱ , 兲 of the fields ␣ⴱ ,  with the property ⴱ 兩X兩 具␣ I共SP兲 n 共; ␣ , 兲 = Zn共兲 e ⴱ,j共 兲典+V 共;␣ⴱ,兲+E 共;␣ⴱ,兲 n n n 共1.7兲 . The function E␦共 ; ␣ⴱ , 兲 is defined for real numbers 0 ⬍ ⱕ ␦ ⱕ ⌰ such that ␦ = 2n for some integer n ⱖ 0. The normalization constant Z␦共兲 is defined in Appendix C. It is chosen so that E␦共 ; 0 , 0兲 = 0. It is extremely close to 1. The “irrelevant” contributions E␦共 ; ␣ⴱ , 兲 to the effective action are characterized by the recursion relation E共; ␣ⴱ, 兲 = 0, ⴱ ⴱ ⴱ E2␦共; ␣ , 兲 = E␦共; ␣ , j共␦兲兲 + E␦共; j共␦兲␣ , 兲 + log ˜ r共␦兲共zⴱ,z兲eA␦共;␣ 兰d ˜ r共␦兲共zⴱ,z兲 兰d ⴱ,;zⴱ,z兲 , 共1.8兲 where A␦共; ␣ⴱ,  ;zⴱ,z兲 = 关V␦共; ␣ⴱ, j共␦兲 + z兲 − V␦共; ␣ⴱ, j共␦兲兲兴 + 关V␦共; j共␦兲␣ⴱ + zⴱ, 兲 − V␦共; j共␦兲␣ⴱ, 兲兴 + 关E␦共; ␣ⴱ, j共␦兲 + z兲 − E␦共; ␣ⴱ, j共␦兲兲兴 + 关E␦共; j共␦兲␣ⴱ + zⴱ, 兲 − E␦共; j共␦兲␣ⴱ, 兲兴. 共1.9兲 The motivation for this recursion relation comes from a SP construction and is given in Sec. II. We estimate E␦共 ; ␣ⴱ , 兲 in terms of norms as in Ref. 3 共Definition 2.6 and, more specifically, 共A1兲兴. Assume that X is a metric space. Choose a constant m ⱖ 0 as a spatial exponential decay rate and a positive monotonically decreasing function ␦ 哫 共␦兲 to measure the radius of convergence of the expansion of E␦共 ; ␣ⴱ , 兲 in powers of the fields ␣ⴱ and . We define the norm of the power series Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-4 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. f共␣ⴱ, 兲 = 兺 兺 k,ᐉⱖ0 x1,. . .,xk苸X y1,. . .,yᐉ苸X a共x1, . . . ,xk ;y1, . . . ,yᐉ兲␣共x1兲ⴱ ¯ ␣共xk兲ⴱ共y1兲 ¯ 共yᐉ兲 关with the coefficients a共x1 , . . . , xk ; y1 , . . . , yᐉ兲 invariant under permutations of x1 , . . . , xk and of y1 , . . . , yᐉ兴 to be 储f共␣ⴱ, 兲储␦ = 兺 max max 兺 k,ᐉⱖ0 x苸X 1ⱕiⱕk+ᐉ 共xជ ,yជ 兲苸Xk⫻Xᐉ 共xជ ,yជ 兲i=x w␦共xជ ;yជ 兲兩a共xជ ;yជ 兲兩 with the weight system w␦共xជ ;yជ 兲 = 共␦兲k+ᐉem共xជ ,yជ 兲 for 共xជ ,yជ 兲 苸 Xk ⫻ Xᐉ , where 共xជ , yជ 兲 is the minimal length of a tree whose set of vertices contains the points of the set 兵x1 , . . . , xk , y1 , . . . , yᐉ其. In the language of Ref. 3 共Definitions 2.5 and 2.6兲, w␦ is the weight system with metric md that associates the constant weight factor 共␦兲 to the fields ␣ⴱ and , and the norm 储f共␣ⴱ , 兲储␦ is denoted 储f共␣ⴱ , 兲储w␦. For any operator A on CX, with kernel A共x , y兲, we define the weighted L1 − L⬁ operator norm 再 冎 兩储A储兩 = max sup 兺 emd共x,y兲兩A共x,y兲兩,sup 兺 emd共x,y兲兩A共x,y兲兩 , x苸X y苸X y苸X x苸X 共1.10兲 as in Ref. 3 共Definition A.1兲. The quantities relevant for the estimates of E␦共 ; ␣ⴱ , 兲, in addition to the radii r共␦兲 and 共␦兲, are the norm 兩储v储兩 of the interaction, a constant K j such 兩储j共兲储兩 ⱕ eK j and 兩储j共兲 − 1储兩 ⱕ K jeK j for ⱖ 0 共1.11兲 共see Corollary B.2兲, and a constant 0 ⬍ ⌰ ⱕ 1 that bounds the range for which the constructions work. On these quantities we make the following. Hypothesis 1.1: We assume that the monotonically decreasing functions r共t兲 and 共t兲 do not decrease too quickly. Precisely, 共i兲 共ii兲 共iii兲 共iv兲 共v兲 共vi兲 1 ⱕ r共t兲 ⱕ 2r共2t兲 and 1 ⱕ 共t兲 ⱕ 2共2t兲, for all 0 ⱕ t ⱕ ⌰ / 2. On the other hand, r共t兲 and 共t兲 must decrease quickly enough and r共t兲 must be sufficiently small compared to 共t兲 that etK j关共2t兲 / 共t兲兴 + 4关r共t兲 / 共t兲兴 ⱕ 1 for all 0 ⱕ t ⱕ ⌰ / 2 and r共t兲关r共t兲 − r共2t兲兴 ⱖ 2 for all 0 ⱕ t ⱕ ⌰ / 2. We also assume that there are constants KE and q ⱖ 1 such that t兩储v储兩r共t兲共t兲3 ⱕ 1 / KE for all 0 ⱕ t ⱕ ⌰; 关1 / C共⌰ , KE兲兴共2 / q兲 ⱕ 共共t兲 / 共2t兲兲4 ⱕ C共⌰ , KE兲4共r共2t兲 / r共t兲兲4 for all 0 ⱕ t ⱕ ⌰ / 2, where C共⌰ , KE兲 = e−4⌰K j关1 − 233e14K j / KE兴; and ⬁ t2兺k=0 共q / 4兲kr共t / 2k兲2共t / 2k兲6 converges uniformly in 0 ⱕ t ⱕ ⌰. Example 1.2: Let v ⬎ 0. 共i兲 Suppose that 共t兲 = 共ii兲 1 冉冊 1 4 冑兩储v储兩 t a and r共t兲 = 1 冉冊 1 4 冑兩储v储兩 t ar for some constants 0 ⬍ ar ⬍ a obeying 3a + ar ⬍ 1. We prove in Appendix D that there are constants KE, ⌰, and q such that Hypothesis 1.1 is fulfilled for all nonzero v with 兩储v储兩 ⱕ v. Suppose that Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-5 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II 1 共t兲 = 冑4 t兩储v储兩 冉 ln 1 t兩储v储兩 冊 冉 b and r共t兲 = ln 1 t兩储v储兩 冊 b for some b ⱖ 1. Again, we prove in Appendix D that there are constants KE, ⌰, and q such that Hypothesis 1.1 is fulfilled for all nonzero v with 兩储v储兩 ⱕ v. We choose p0共兲 of the functional integral representation 共1.1兲 to be r共兲. Theorem 1.3: Under Hypothesis 1.1 储E␦共; ␣ⴱ, 兲储␦ ⱕ KE␦2兩储v储兩2r共␦兲2共␦兲6 for all 0 ⱕ ⱕ ␦ ⱕ ⌰ for which ␦ / is a power of 2. The function E␦共 ; ␣ⴱ , 兲 has degree at least two both in ␣ⴱ and . [By this we mean that every monomial appearing in its power series expansion contains a factor of the form ␣ⴱ共x1兲␣ⴱ共x2兲共x3兲共x4兲.] Theorem 1.3 is proven after Proposition 3.3. The theorem is proven by induction on n, where ␦ = 2n. The induction step is prepared by Proposition 3.3, which is proven using the results in Ref. 3. After the proof of Theorem 1.3, we give the following proof. Theorem 1.4: The limit E共␣ⴱ, 兲 = lim E共2−m ; ␣ⴱ, 兲 m→⬁ exists uniformly in 0 ⱕ ⱕ ⌰. It fulfills the estimate 储E共␣ⴱ, 兲储 ⱕ KE2兩储v储兩2r共兲2共兲6 and has degree of at least 2 in both ␣ⴱ and . To take the limit → 0 in 共1.7兲 observe that, for any fixed ␦, lim V␦共2−n␦ ; ␣ⴱ, 兲 = V␦共␣ⴱ, 兲, n→⬁ where V ␦共 ␣ ⴱ,  兲 = − 冕 ␦ 具关j共t兲␣ⴱ兴关j共␦ − t兲兴, v关j共t兲␣ⴱ兴关j共␦ − t兲兴典dt. 0 The SP approximation to the I共␣ⴱ , 兲 constructed in Ref. 4 then is 共SP兲 −m 共2 ; ␣ⴱ, 兲 = Z兩X兩e具␣ I共SP兲共␣ⴱ, 兲 = lim Im ⴱ,j共兲典+V 共␣ⴱ,兲+E 共␣ⴱ,兲 . m→⬁ The existence of Z = lim Z共 / 2m兲 is proven in Lemma C.1. m→⬁ II. SP AND STOKES’ THEOREM in the recursion To motivate the recursive definition 共1.8兲 of E␦共 ; ␣ⴱ , 兲, we replace In by I共SP兲 n relation 共1.4兲. Inserting 共1.7兲, the resulting integral Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-6 冕 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. 共SP兲 ⴱ ⴱ ˜ R 共ⴱ, 兲I共SP兲 d n 共; ␣ , 兲In 共; , 兲 = Z兩2X兩 n = Z2兩X兩 n 冕 冕冋兿 ˜ R 共ⴱ, 兲e具␣ d x苸X ⴱ,j共 兲典+具ⴱ,j共 兲典 V 共;␣ⴱ,兲+V 共;ⴱ,兲 E 共;␣ⴱ,兲+E 共;ⴱ,兲 n n n n n n e e 册 dⴱ共x兲 ∧ d共x兲 ⴱ ⴱ 共兩共x兲兩 ⬍ R兲 eA共␣ ,; ,兲 2ı 共2.1兲 with Zn = Zn共兲 and A共␣ⴱ,  ; ⴱ, 兲 = − 具ⴱ, 典 + 具␣ⴱ, j共n兲典 + 具ⴱ, j共n兲典 + Vn共; ␣ⴱ, 兲 + Vn共; ⴱ, 兲 + En共; ␣ⴱ, 兲 + En共; ⴱ, 兲. Here we have written A as a function of four independent complex fields ␣ⴱ, , ⴱ, and . The activity in 共2.1兲 is obtained by evaluating A共␣ⴱ ,  ; ⴱ , 兲 with ⴱ = ⴱ, the complex conjugate of . The reason for introducing independent complex fields ⴱ and lies in the fact that the critical point 共with respect to the variables ⴱ and 兲 of the quadratic part − 具ⴱ, 典 + 具j共n兲␣ⴱ, 典 + 具ⴱ, j共n兲典 = − 具ⴱ − j共n兲␣ⴱ, − j共n兲典 + 具␣ⴱ, j共n+1兲典 of A is “not real.” Precisely, the critical point is ⴱ crit ⴱ = j共n兲␣ , crit = j共n兲 ⴱ crit and, in general 共crit ⴱ 兲 ⫽ . To do stationary phase, we make the substitution ⴱ = crit ⴱ + z ⴱ, = crit + z 共2.2兲 with “fluctuation fields” zⴱ and z. With this substitution, the quadratic part of A is equal to −具zⴱ , z典 + 具␣ⴱ , j共n+1兲典. So 共2.1兲 becomes 具␣ Z2兩X兩 n e ⴱ,j共 n+1兲典 冋兿 冕 x苸X M共x兲 册 dzⴱ共x兲 ∧ dz共x兲 −z 共x兲z共x兲 Ã共␣ⴱ,;z ,z兲 ⴱ , e ⴱ e 2ı 共2.3兲 where ⴱ crit + z兲 + En共; crit Ã共␣ⴱ,  ;zⴱ,z兲 = Vn共; ␣ⴱ, crit + z兲 + Vn共; crit ⴱ + zⴱ, 兲, ⴱ + zⴱ, 兲 + En共; ␣ , ⴱ crit M共x兲 = 兵共zⴱ共x兲,z共x兲兲兩共crit ⴱ 共x兲 + zⴱ共x兲兲 = 共x兲 + z共x兲 and 兩crit共x兲 + z共x兲兩 ⬍ R其. 共2.3兲 is the integral over a real 2兩X兩 dimensional subset in the complex 2兩X兩 dimensional space of fields zⴱ and z. The first step in the SP approximation is to replace, for each x 苸 X, the set M共x兲 in 共2.3兲 by the neighborhood ⴱ crit D共x兲 = 兵共zⴱ共x兲,z共x兲兲 苸 C2兩兩zⴱ共x兲兩 ⱕ r共n兲, 兩z共x兲兩 ⱕ r共n兲, 共zⴱ共x兲 + crit ⴱ 共x兲兲 = z共x兲 + 共x兲其 of the critical point. In Ref. 4 共Sec. VI兲 we show that the error introduced by this approximation is extremely small 关even when D共x兲 is empty兴. There, we provide detailed bounds on a “large field–small field” expansion for which the above approximation is the leading term. In Remark 2.1 关part 共a兲, below兴, we illustrate the sources of the smallness for n = 0. ⴱ crit By Stokes’ theorem 关Lemma A.1 with r = r共n兲, = ⴱ = 0, and = 共crit ⴱ 兲 − 兴, one can write Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-7 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II 具␣ Z2兩X兩 n e ⴱ,j共 n+1兲典 冋兿 冕 D共x兲 x苸X 册 dzⴱ共x兲 ∧ dz共x兲 −z 共x兲z共x兲 Ã共␣ⴱ,;z ,z兲 ⴱ e ⴱ e 2ı 共2.4兲 as the sum of 具␣ Z2兩X兩 n e ⴱ,j共 n+1兲典 冋兿 冕 x苸X 具␣ and Z2兩X兩 n e ⴱ ,j共n+1兲典 R⫽0” ⫻ 册 共2.5兲 times 冋冕 兿 冋冕 兿 兺 x苸R R傺X 兩z共x兲兩ⱕr共n兲 dzⴱ共x兲 ∧ dz共x兲 −兩z共x兲兩2 Ã共␣ⴱ,;zⴱ,z兲 e e 2ı C共x兲 x苸X\R dzⴱ共x兲 ∧ dz共x兲 −z 共x兲z共x兲 e ⴱ 2i 兩z共x兲兩ⱕr共n兲 册 册 dz共x兲ⴱ ∧ dz共x兲 −z 共x兲z共x兲 Ã共␣ⴱ,;zⴱ,z兲 z 共x兲=z共x兲ⴱ e 兩ⴱ , e ⴱ 2i for x苸X\R where, for each x 苸 X, C共x兲 is a two real dimensional submanifold of C2 whose boundary is the union of “circles” D共x兲 and 兵共zⴱ共x兲 , z共x兲兲 苸 C2 兩 zⴱⴱ共x兲 = z共x兲 , 兩z共x兲兩 = r共n兲其. In Ref. 4 we argue that −zⴱ共x兲z共x兲 has an extremely large negative real part whenever 共zⴱ共x兲 , z共x兲兲 苸 C共x兲. 关Also see Remark 2.1, part 共b兲, below.兴 The second step in the SP approximation is to ignore these terms. That is, to replace 共2.4兲 with 共2.5兲. Thus, the SP approximation for 冕 共SP兲 ⴱ ⴱ ˜ R 共ⴱ, 兲I共SP兲 d n 共; ␣ , 兲In 共; , 兲 is 具␣ 共2.5兲 = Z2兩X兩 n e ⴱ,j共 n+1兲典 冕 ˜ r共 兲共zⴱ,z兲eÃ共␣ d n ⴱ,;zⴱ,z兲 . By construction ⴱ ⴱ ⴱ Vn共; ␣ⴱ, crit兲 + Vn共; crit ⴱ , 兲 = Vn共; ␣ , j共n兲兲 + Vn共; j共n兲␣ , 兲 = Vn+1共; ␣ , 兲 so that ⴱ ⴱ ⴱ Vn共; ␣ⴱ, crit + z兲 + Vn共; crit ⴱ + z , 兲 = Vn+1共; ␣ , 兲 + 关Vn共; ␣ , j共n兲 + z兲 − Vn共; ␣ⴱ, j共n兲, 兲兴 + 关Vn共; j共n兲␣ⴱ + zⴱ, 兲 − Vn共; j共n兲␣ⴱ, 兲兴. Consequently, the SP approximation for 冕 共SP兲 ⴱ ⴱ ˜ R 共ⴱ, 兲I共SP兲 d n 共; ␣ , 兲In 共; , 兲 can also be written as 具␣ Z2兩X兩 n e ⴱ,j共 ⴱ ⴱ ⴱ n+1兲典+Vn+1共;␣ ,兲 En共;␣ ,j共n兲兲En共;j共n兲␣ ,兲 e 冕 ˜ r共 兲共zⴱ,z兲eAn共;␣ d n ⴱ,;z ,z兲 ⴱ . This is compatible with 共1.7兲 and 共1.8兲 since, by the definition of Appendix C, Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-8 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. Z2n共n兲 = Zn共兲2 冕 兩z兩⬍r共n兲 dzⴱ ∧ dz −兩z兩2 e . 2i Remark 2.1: 共a兲 We now illustrate why the error introduced by the SP approximation is extremely small by considering the case n = 0. The initial functional integral representation 共1.1兲 may be written as Tr e−共1/kT兲共H−N兲 = lim →0 冕 兿 苸Z艚共0,1/kT兴 再冋 兿 x苸X 册 d␣ⴱ共x兲 ∧ d␣共x兲 共兩␣共x兲兩 ⬍ R兲 共␣−, ␣兲 2ı ⴱ ⴱ 冎 ⫻e−共1/2兲具␣−,␣−典I0共; ␣ⴱ−, ␣兲e−共1/2兲具␣ ,␣典 , where I0共; ␣ⴱ, 兲 = e具␣ ⴱ,j共兲典 −共␣ⴱ, ␣ⴱ兲 v e . Dropping one of the “time derivative small field characteristic functions” 共␣− , ␣兲 would introduce only a very small error. This is because writing ␣− = ␣ and ␣ = , the quadratic ⴱ ⴱ part of the exponent of e−共1/2兲具␣ ,␣典I0共 ; ␣ⴱ , 兲e−共1/2兲具 ,典 obeys Re兵− 21 具␣ⴱ, ␣典 + 具␣ⴱ, j共兲典 − 21 具ⴱ, 典其 ⬇ Re兵− 21 具␣ⴱ, ␣典 + 具␣ⴱ, 典 − 21 具ⴱ, 典其 = − 21 储␣ − 储L2 , 2 2 2 which generates a factor on the order of e−共1/2兲p0共兲 = e−共1/2兲r共兲 when 共␣ , 兲 is not in support of 共␣ , 兲. The quartic part −具␣ⴱ , v␣ⴱ典 of the exponent is roughly −具␣ⴱ␣ , v␣ⴱ␣典 ⱕ 0 and so cannot generate a large factor. A similar mechanism generates small factors for the right hand side of 共1.3兲 whenever the difference ␣−n − ␣ between the two arguments of In is larger than roughly r共n兲. Consequently, we apply the SP approximation to the integral I1共; ␣ⴱ, 兲 = 冕 ˜ R 共ⴱ, 兲共␣, 兲I0共; ␣ⴱ, 兲I0共; ⴱ, 兲共共ⴱ兲ⴱ, 兲 d of 共1.2兲 only when the “time derivative small field condition” 储␣ − 储⬁ ⱕ r共2兲 is satisfied. The change in variables 共2.2兲 expresses I1 as I1共; ␣ⴱ, 兲 = e具␣ ⴱ,j共2兲典 冋兿 冕 x苸X M共x兲 册 dzⴱ共x兲 ∧ dz共x兲 −z 共x兲z共x兲 Ã共␣ⴱ,;z ,z兲 ⴱ e ⴱ e 2ı ⫻共␣, j共兲 + z兲共共j共兲␣ⴱ + zⴱ兲ⴱ, 兲. The characteristic function 共␣ , j共兲 + z兲 limits the domain of integration to z’s, obeying 储z + j共兲 − ␣储⬁ ⬍ r共兲. 共b兲 Since 储␣ − 储⬁ ⱕ r共2兲 ⱕ 21 r共兲 and 储j共兲 − 储⬁ ⱕ const R Ⰶ r共兲, this condition is roughly equivalent to 储z储⬁ ⬍ r共兲. On the difference between these two domains of integration, the integrand is extremely small, for reasons like those given above. Similarly, the condition imposed by the second is roughly equivalent to 储zⴱ储⬁ ⬍ r共兲. The two conditions 储z储⬁ ⱕ r共兲 and 储zⴱ储⬁ ⱕ r共兲 are built into the domains of integration D共x兲 in 共2.4兲. The “time derivative small field condition” 储␣ − 储⬁ ⱕ r共2兲 ⱕ 21 r共兲 is also used to ensure that −zⴱ共x兲z共x兲 has an extremely large negative real part whenever 共zⴱ共x兲 , z共x兲兲 lies on C共x兲, the side of the Stokes’ “cylinder.” This may be seen from Remark A.3 with r = r共兲, ⴱ crit = ⴱ = 0, and = 共crit ⴱ 兲 − = j共兲关␣ − 兴. Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-9 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II III. THE INDUCTION STEP In preparation for the proof of Proposition 3.3 below, as well as of the proofs of Theorems 1.3 and 1.4, we collect consequences of Hypotheses 1.1 in the form in which they are actually used in these proofs. Remark 3.1: Hypothesis 1.1 implies that for all 0 ⱕ ␦ ⱕ ⌰ / 2, KE ⱖ 219e9K j , 冋 共3.1a兲 冉 冊 册冉 冊 231e14␦K j k共␦兲 + e 2␦K j KE k共2␦兲 2e2␦K j 冉 冊 k共2␦兲 k共␦兲 4 r共␦兲 r共2␦兲 2 2 ⱕ 2, 共3.1b兲 + 226e9K j␦兩储v储兩r共␦兲共2␦兲3 ⱕ q. 共3.1c兲 Proof: Hypothesis 1.1 关part 共v兲兴 forces C共⌰ , KE兲 ⱖ 0 and hence KE ⱖ 233e14K j, which implies 共3.1a兲. Hypothesis 1.1 关part 共v兲兴 also forces 冉 冊 k共␦兲 共2␦兲 2 ⱕ 2e−2⌰K j 冑 1− 冉 冊 233e14K j r共2␦兲 KE r共␦兲 2 冋 ⱕ 2e−2␦K j 1 − 232e14K j KE 册冉 冊 r共2␦兲 r共␦兲 2 and hence 4 冉 冊冉 冊 231e14␦K j 共␦兲 + e 2␦K j KE 共2␦兲 2 r共␦兲 r共2␦兲 2 ⱕ 2. So 共3.1b兲 now follows from Hypothesis 1.1 关part 共i兲兴, which ensures that 共r共␦兲 / r共2␦兲兲2 ⱕ 4. Finally, by Hypothesis 1.1 关parts 共iv兲 and 共v兲兴, 2e2␦K j 冉 冊 共2␦兲 共␦兲 4 + 226e9K j␦兩储v储兩r共␦兲共2␦兲3 ⱕ e2⌰K jqC共⌰,KE兲 + = qe−2⌰K j − q ⱕq− 226e9K j KE 233e−2⌰K je14K j 226e9K j + KE KE 233e12K j 226e9K j + KE KE ⱕq since q ⱖ 1 and 0 ⬍ ⌰ ⱕ 1. We formulate the recursion relation 共1.8兲 that defines En 共 ; ␣ⴱ , 兲 more abstractly. Definition 3.2: Let 0 ⱕ ⱕ ␦. For an action E共␣ⴱ , 兲, we set ⴱ ⴱ ⴱ R␦,关E兴共␣ , 兲 = E共␣ , j共␦兲兲 + E共j共␦兲␣ , 兲 + log ˜ r共␦兲共zⴱ,z兲eA␦,共E;␣ 兰d 䊏 ⴱ,;zⴱ,z兲 ˜ r共␦兲共zⴱ,z兲 兰d whenever the logarithm is defined. Here, A␦,共E; ␣ⴱ,  ;zⴱ,z兲 = 关V␦共; ␣ⴱ, j共␦兲 + z兲 − V␦共; ␣ⴱ, j共␦兲兲兴 + 关V␦共; j共␦兲␣ⴱ + zⴱ, 兲 − V␦共; j共␦兲␣ⴱ, 兲兴 + 关E共␣ⴱ, j共␦兲 + z兲 − E共␣ⴱ, j共␦兲兲兴 + 关E共j共␦兲␣ⴱ + zⴱ, 兲 − E共j共␦兲␣ⴱ, 兲兴. The recursion relation 共1.8兲 is equivalent to Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-10 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. E共; ␣ⴱ, 兲 = 0, En+1共; ␣ⴱ, 兲 = Rn,关En共; ␣ⴱ, 兲兴. 共3.2兲 To prove Theorem 1.3, we perform induction on n to successively bound En共 ; ·兲 for n = 0 , . . . , log2共⌰ / 兲. For the induction step, we use the following. Proposition 3.3: Assume that r共t兲 and 共t兲 fulfill Hypothesis 1.1. Then, for all 0 ⱕ ⱕ ␦ ⱕ ⌰ / 2, with ␦ an integer multiple of , the following holds. Let E共␣ⴱ , 兲 be an analytic function which has degree of at least 2 both in ␣ⴱ and  and which obeys 储E共␣ⴱ , 兲储␦ ⱕ 29e7␦K j␦兩储v储兩r共␦兲共2␦兲3. Then R␦,关E兴共␣ⴱ , 兲 is well defined, has degree at least two both in ␣ⴱ and , and satisfies the estimate 储R␦,关E兴储2␦ ⱕ 232e14␦K j␦2兩储v储兩2r共␦兲2共2␦兲6 + 2e2␦K j 冉 冊 共2␦兲 4 储E储␦ . 共␦兲 Proof: Observe that the functions V␦共 ; ␣ⴱ , j共␦兲 + z兲 − V␦共 ; ␣ⴱ , j共␦兲兲 and E共␣ⴱ , j共␦兲 + z兲 − E共␣ⴱ , j共␦兲兲 both have degree of at least 2 in ␣ⴱ, degree of at least 1 in z, and do not depend on zⴱ. Similarly, both V␦共 ; j共␦兲␣ⴱ + zⴱ , 兲 − V␦共 ; j共␦兲␣ⴱ , 兲 and E共j共␦兲␣ⴱ + zⴱ , 兲 − E共j共␦兲␣ⴱ , 兲 have degree of at least 2 in , degree of at least 1 in zⴱ, and do not depend on z. Since the integral of ˜ r共␦兲共zⴱ , z兲 is zero unless there are the same number of z’s and zⴱ’s, any monomial against d 冕 ˜ r共␦兲共zⴱ,z兲 A␦,共E; ␣ⴱ,  ;zⴱ,z兲 = 0 d 共3.3兲 and log ˜ r共␦兲共zⴱ,z兲eA␦,共E;␣ 兰d ⴱ,;zⴱ,z兲 ˜ r共␦兲共zⴱ,z兲 兰d has degree of at least 2 both in ␣ⴱ and . This implies that R␦,关E兴共␣ⴱ , 兲 has degree of at least 2 both in ␣ⴱ and . To estimate A␦, we introduce a second auxiliary weight system wfluct. It has a metric md and associates the constant weight factor 共2␦兲 to the fields ␣ⴱ and  and the constant weight factor 4r共␦兲 to the fluctuation fields zⴱ and z. We abbreviate 储f共␣ⴱ ,  ; zⴱ , z兲储fluct = 储f共␣ⴱ ,  ; zⴱ , z兲储wfluct. Clearly, 储f共␣ⴱ , 兲储2␦ = 储f共␣ⴱ , 兲储fluct for functions that are independent of the fluctuation fields. Observe that V␦共; ␣ⴱ, j共␦兲 + z兲 − V␦共; ␣ⴱ, j共␦兲兲 = 兺 苸Z艚共0,␦兴 关具␥ⴱ−g, v␥ⴱ−g典 − 具␥ⴱ−ĝ, v␥ⴱ−ĝ典兴 with ␥ⴱ = j共兲␣ⴱ, g = j共2␦ − 兲, ĝ = j共␦ − 兲共j共␦兲 + z兲 = j共2␦ − 兲 + j共␦ − 兲z. 共3.4兲 We apply Proposition A.3 关part 共ii兲兴 of Ref. 3, with d replaced by md, r = 4, s = 3, h共␥1 , . . . , ␥4兲 = 共␥1␥2 , v␥3␥4兲, ␣1 = ␣ⴱ, ␣2 = , ␣3 = z, weights 1 = ¯ = 4 = 1, and ⌫11 = ⌫13 = j共 − 兲, ˜⌫1 = ˜⌫1 = j共 − 兲, 1 3 ⌫22 = ⌫24 = j共2␦ − 兲, ˜⌫2 = ˜⌫2 = j共2␦ − 兲, 4 2 ⌫32 = ⌫34 = 0, ˜⌫3 = ˜⌫3 = j共␦ − 兲 2 4 共3.5兲 with all other ⌫ij’s and ˜⌫ij’s being zero. Then Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-11 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II 再 = 共2␦兲max 兩储j共 − 兲储兩, 兩储j共2␦ − 兲储兩, 兩储j共2␦ − 兲储兩 + 4 冉 ⱕ 1+4 冊 r共␦兲 兩储j共␦ − 兲储兩 共2␦兲 冎 r共␦兲 2K ␦ e j 共2␦兲, 共2␦兲 ␦ = 4r共␦兲兩储j共␦ − 兲储兩 ⱕ 4eK j␦r共␦兲. 共3.6兲 So, for each 苸 Z 艚 共0 , ␦兴, 储具␥ⴱ−g, v␥ⴱ−g典 − 具␥ⴱ−ĝ, v␥ⴱ−ĝ典储fluct ⱕ 4储h储w␦3 ⱕ 29e7K j␦兩储v储兩r共␦兲共2␦兲3 . Here, we used that r共␦兲 ⱕ 41 共␦兲 ⱕ 21 共2␦兲 by Hypothesis 1.1 关parts 共i兲 and 共ii兲兴. Summing over gives 储V␦共; ␣ⴱ, j共␦兲 + z兲 − V␦共; ␣ⴱ, j共␦兲兲储fluct ⱕ 29e7K j␦␦兩储v储兩r共␦兲共2␦兲3 . Similarly, 储V␦共; j共␦兲␣ⴱ + zⴱ, 兲 − V␦共; j共␦兲␣ⴱ + zⴱ, 兲储fluct ⱕ 29e7K j␦␦兩储v储兩r共␦兲共2␦兲3 . Next, by Corollary A.2 of Ref. 3 for any analytic function f共␣ⴱ , 兲, 储f共␣ⴱ, j共␦兲 + z兲 − f共␣ⴱ, j共␦兲兲储fluct ⱕ 储f共␣ⴱ, j共␦兲 + z兲储fluct ⱕ 储f共␣ⴱ, 兲储␦ 共3.7兲 since 共2␦兲 4r共␦兲 4共r␦兲 共2␦兲 + 兩储j共␦兲储兩 + 兩储1储兩 ⱕ e␦K j ⱕ1 共␦兲 共␦兲 共␦兲 共␦兲 by Hypothesis 1.1 关part 共ii兲兴. In particular 储E共␣ⴱ , j共␦兲 + z兲 − E共␣ⴱ , j共␦兲兲储fluct ⱕ 储E储␦. Similarly 储E共j共␦兲␣ⴱ + zⴱ , 兲 − E共j共␦兲␣ⴱ , 兲储fluct ⱕ 储E储␦. Combining the bounds of the previous two paragraphs with the assumption on 储E储␦, we get 储 A␦,共E; ·兲储fluct ⱕ 210e7␦K j␦兩储v储兩r共␦兲共2␦兲3 + 2储E储␦ ⱕ 211e7␦K j␦兩储v储兩r共␦兲共2␦兲3 ⱕ 1 64 共3.8兲 by Hypothesis 1.1 关part 共iv兲兴 and 共3.1a兲. By 共3.3兲 and Corollary 3.5 共Ref. 3兲 with n = 1, 冐 log ˜ r共␦兲共zⴱ,z兲eA␦,共E;␣ 兰d ˜ r共␦兲共zⴱ,z兲 兰d ⴱ,;z ,z兲 ⴱ 冐 冉 ⱕ 2␦ 2 储 A␦,共E; ·兲储fluct 冊 1 − 储 A␦,共E; ·兲储fluct 20 2 ⱕ 232e14␦K j␦2兩储v储兩2r共␦兲2共2␦兲6 . Combining this estimate and the estimate of Lemma 3.4, below, with f = E, we get the desired bound on 储R␦,关E兴储2␦. 䊏 Lemma 3.4: Let f共␣ⴱ , 兲 be an analytic function that has degree of at least 2 both in ␣ⴱ and . Then, 储f共␣ⴱ, j共␦兲兲储2␦, 储f共j共␦兲␣ⴱ, 兲储2␦ ⱕ e2␦K j 冉 冊 共2␦兲 4 储f储␦ . 共␦兲 Proof: Introduce the auxiliary weight system waux with, in the language of Ref. 3 关Definitions 2.5 and 2.6 and, more specifically, 共A.1兲兴, metric md that associates the constant weight factor to the field ␣ⴱ and the constant weight factor e−␦K j共␦兲 to the field . Since, by 共1.11兲, 关e−␦K j共␦兲 / 共␦兲兴兩储j共␦兲储兩 ⱕ 1, Corollary A.2 of Ref. 3 gives Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-12 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. 储f共␣ⴱ, j共␦兲兲储waux ⱕ 储f储␦ . As f共␣ⴱ , j共␦兲兲 has degree of at least 2 both in ␣ⴱ and  and e−␦K j共␦兲 ⱖ 共2␦兲, by Hypothesis 1.1 关part 共ii兲兴, 储f共␣ⴱ, j共␦兲兲储2␦ ⱕ 冉 冊冉 共2␦兲 共␦兲 2 冊 冉 冊 共2␦兲 2 共2␦兲 4 储f共␣ⴱ, j共␦兲兲储waux ⱕ e2␦K j 储f储␦ . e 共␦兲 共␦兲 −␦K j The estimate on 储f共j共␦兲␣ⴱ , 兲储2␦ is similar. 䊏 Proof of Theorem 1.3: We write ␦ = n = 2n and prove the statement by induction on n. In the case n = 0, there is nothing to prove. For the induction step from n to n + 1, set ␦ = n. The hypothesis of Proposition 3.3, with E = E␦, is satisfied since, 储E␦储␦ ⱕ KE␦2兩储v储兩2r共␦兲2共␦兲6 ⱕ 8␦兩储v储兩r共␦兲共2␦兲3 共3.9兲 by the inductive hypothesis and Hypothesis 1.1 关parts 共i兲 and 共iv兲兴. Using 共3.2兲, Proposition 3.3, and 共3.1a兲 we see that 储En+1储n+1 ⱕ 232e14␦K j␦2兩储v储兩2r共␦兲2共2␦兲6 + 2e2␦K j 冋 冋 ⱕ 232e14␦K j + 2e2␦K j = 冉 冊 共2␦兲 4 KE␦2兩储v储兩2r共␦兲2共␦兲6 共␦兲 冉 冊 册 冉 冊 册冉 冊 共␦兲 2 KE ␦2兩储v储兩2r共␦兲2共2␦兲6 共2␦兲 1 231e14␦K j 共␦兲 + e 2␦K j 2 KE 共2␦兲 2 r共␦兲 r共2␦兲 2 KE共2␦兲2兩储v储兩2r共2␦兲2共2␦兲6 ⱕ KE共2␦兲2兩储v储兩2r共2␦兲2共2␦兲6 2 = KEn+1 兩储v储兩2r共n+1兲2共n+1兲6 . 䊏 In the proof of Theorem 1.4, we shall compare E共2−m−1 ; ␣ⴱ , 兲 and E共2−m ; ␣ⴱ , 兲 to prove that the sequence E共2−m ; ·兲 is Cauchy with respect to our norm. To do so, we shall compare En共 / 2 ; ·兲 and En共 ; ·兲 for each n = 0 , . . . , log2共⌰ / 兲. This is done by induction on n. For the induction step, we use Proposition 3.6, below. To prepare for it, we have as follows. Lemma 3.5: Set W共␣ⴱ, 兲 = V␦共; ␣ⴱ, 兲 − V␦ 冉 冊 ; ␣ ⴱ,  . 2 Then 储W共␣ⴱ,z + j共␦兲兲 − W共␣ⴱ, j共␦兲兲储fluct ⱕ 210e9K j␦兩储v储兩r共␦兲共2␦兲3 , 储W共zⴱ + j共␦兲␣ⴱ, 兲 − W共j共␦兲␣ⴱ, 兲储fluct ⱕ 210e9K j␦兩储v储兩r共␦兲共2␦兲3 . Proof: We prove the first inequality. By definition W共␣ⴱ, 兲 = W1共␣ⴱ, 兲 + W2共␣ⴱ, 兲 with Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-13 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II W 1共 ␣ ⴱ,  兲 = − 兺 关具␥ⴱ−␥, v␥ⴱ−␥典 − 具␥ⴱ−␥−/2, v␥ⴱ−␥−/2典兴, 2 苸Z艚共0,␦兴 W 2共 ␣ ⴱ,  兲 = − 兺 关具␥ⴱ−␥, v␥ⴱ−␥典 − 具␥ⴱ−/2␥, v␥ⴱ−/2␥典兴, 2 苸Z艚共0,␦兴 where ␥ⴱ = j共兲␣ⴱ, ␥ = j共␦ − 兲. Using the notation of 共3.4兲, W1共␣ⴱ, j共␦兲兲 − W1共␣ⴱ,z + j共␦兲兲 = − 兺 关具␥ⴱ−g, v␥ⴱ−g典 − 具␥ⴱ−g−/2, v␥ⴱ−g−/2典 2 苸Z艚共0,␦兴 − 具␥ⴱ−ĝ, v␥ⴱ−ĝ典 + 具␥ⴱ−ĝ−/2, v␥ⴱ−ĝ−/2典兴 =− 冋 兺 具␥ⴱ−g, v␥ⴱ−g典 2 苸Z艚共0,␦兴 冓 冉冊 − ␥ⴱ− j 冉冊 g, v␥ⴱ− j g 2 2 冔 冓 冉冊 − 具␥ⴱ−ĝ, v␥ⴱ−ĝ典 + ␥ⴱ− j 冉 冊 冔册 ĝ, v␥ⴱ− j ĝ 2 2 . This time, we apply Proposition A.3 关part 共iii兲兴 of Ref. 3 using the ⌫ij’s and ˜⌫ij’s of 共3.5兲 and, in addition, A1 = Ã1 = A3 = Ã3 = 1, A2 = A4 = 1, Ã2 = Ã4 = j 冉冊 . 2 The corollary bounds the 储 · 储fluct norm of the term by 42兩储v储兩␦a␦共a兲3 with and ␦ of 共3.6兲 and 再 冏 冐 冉 冊冐冏 冎 a = max 兩储1储兩, a␦ = j 2 冏 冐 冉 冊 冐冏 j −1 2 ⱕ eK j共/2兲 , ⱕ K jeK j共/2兲 2 by 共1.11兲. Inserting and summing over , we get 储W1共␣ⴱ,z + j共␦兲兲 − W1共␣ⴱ, j共␦兲兲储fluct ⱕ ␦ 2 4 兩储v储兩4eK j␦r共␦兲 K jeK j共/2兲 2 2 冉冉 1+4 冊 r共␦兲 2K ␦ e j 共2␦兲eK j共/2兲 共2␦兲 冊 3 ⱕ 29e9K j␦兩储v储兩r共␦兲共2␦兲3 . The same estimate holds for 储W2共␣ⴱ , z + j共␦兲兲 − W2共␣ⴱ , j共␦兲兲储fluct. 䊏 Proposition 3.6: Under the hypotheses of Proposition 3.3, assume that there is a second analytic function Ẽ共␣ⴱ , 兲, which has similar properties to E and is close to E. Precisely, we assume that Ẽ is of degree of at least 2 both in ␣ⴱ and  and obeys 储Ẽ储␦ ⱕ 29e7␦K j␦兩储v储兩r共␦兲共2␦兲3. Then 储R␦,关E兴 − R␦,/2关Ẽ兴储2␦ ⱕ 236e18K j␦2兩储v储兩2r共␦兲2共2␦兲6 + q储E − Ẽ储␦ , Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-14 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. where q is the constant in parts (v) and (vi) of Hypothesis 1.1. Proof: By Definition 3.2, ⴱ R␦,关E兴 − R␦,/2关Ẽ兴 = B共␣ , 兲 + log ˜ r共␦兲共zⴱ,z兲eA␦,共E;␣ 兰d ⴱ,;zⴱ,z兲 ˜ r共␦兲共zⴱ,z兲 兰d − log ˜ r共␦兲共zⴱ,z兲eA␦,/2共Ẽ;␣ 兰d ⴱ,;zⴱ,z兲 , ˜ r共␦兲共zⴱ,z兲 兰d 共3.10兲 where B共␣ⴱ , 兲 = 共E − Ẽ兲共␣ⴱ , j共␦兲兲 + 共E − Ẽ兲共j共␦兲␣ⴱ , 兲. By Lemma 3.4, 储B储2␦ ⱕ 2e2␦K j 冉 冊 共2␦兲 4 储E − Ẽ储␦ 共␦兲 共3.11兲 With the notation of the previous lemma, A␦,共E; ␣ⴱ,  ;zⴱ,z兲 − A␦,/2共Ẽ; ␣ⴱ,  ;zⴱ,z兲 = 关W共␣ⴱ,z + j共␦兲兲 − W共␣ⴱ, j共␦兲兲兴 + 关W共zⴱ + j共␦兲␣ⴱ, 兲 − W共j共␦兲␣ⴱ, 兲兴 + C共␣ⴱ,  ;zⴱ,z兲, where C共␣ⴱ,  ;zⴱ,z兲 = 关共E − Ẽ兲共␣ⴱ, j共␦兲 + z兲 − 共E − Ẽ兲共␣ⴱ, j共␦兲兲兴 + 关共E − Ẽ兲共j共␦兲␣ⴱ + zⴱ, 兲 − 共E − Ẽ兲共j共␦兲␣ⴱ, 兲兴. By 共3.7兲, 储C储fluct ⱕ 2储E − Ẽ储␦. Combining this with Lemma 3.5, we get 储 A␦,共E; ·兲 − A␦,/2共Ẽ; ·兲储fluct ⱕ 211e9K j␦兩储v储兩r共␦兲共2␦兲3 + 2储E − Ẽ储␦ ⱕ 兵211e9K j + 211e7␦K j其␦兩储v储兩r共␦兲共2␦兲3 . 共3.12兲 Consequently, by 共3.8兲, 储 A␦,共E; ·兲储fluct + 储 A␦,共E; ·兲 − A␦,/2共Ẽ; ·兲储fluct ⱕ 兵211e9K j + 212e7␦K j其␦兩储v储兩r共␦兲共2␦兲3 ⱕ 213e9K j␦兩储v储兩r共␦兲共2␦兲3 ⱕ 1 1 − 17 32 by 共3.1a兲 and Hypothesis 1.1 关part 共iv兲兴. Therefore, the hypotheses of Ref. 3 共Corollary 3.6兲 are satisfied and we have, using 共3.3兲 and 共3.12兲, 冐 log ˜ r共␦兲共zⴱ,z兲eA␦,共E;␣ 兰d ⴱ,;z ,z兲 ⴱ − log ˜ r共␦兲共zⴱ,z兲 兰d ˜ r共␦兲共zⴱ,z兲eA␦,/2共Ẽ;␣ 兰d ˜ r共␦兲共zⴱ,z兲 兰d ⴱ,;z ,z兲 ⴱ 冐 2␦ ⱕ 2 兵储 A␦,共E; ·兲储fluct + 储 A␦,共E; ·兲 − A␦,/2共Ẽ; ·兲储fluct其储 A␦,共E; ·兲 − A␦,/2共Ẽ; ·兲储fluct 12 ⱕ 225e9K j␦兩储v储兩r共␦兲共2␦兲3兵211e9K j␦兩储v储兩r共␦兲共2␦兲3 + 2储E − Ẽ储␦其. Using this, 共3.11兲 and 共3.1c兲, we bound 共3.10兲 by 储R␦,关E兴 − R␦,/2关Ẽ兴储2␦ ⱕ 236e18K j␦2兩储v储兩2r共␦兲2共2␦兲6 + q储E − Ẽ储␦ . Corollary 3.7: For all sufficiently small ⬎ 0 and integers 0 ⱕ n ⱕ log2共⌰ / 兲, we have 冐 En共; ␣ⴱ, 兲 − En 冉 ⴱ ;␣ , 2 冊冐 䊏 n n ⱕ KEqn2兩储v储兩2r共兲2共兲6 + 236e18K j兩储v储兩2 兺 qn−k2k r共k兲2共k兲6 . k=1 Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-15 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II Proof: The proof is by induction on n. In the case n = 0, En共 ; ␣ⴱ , 兲 = 0 and 冐 冉 E n ⴱ ;␣ , 2 冊冐 n ⱕ KE2兩储v储兩2r共兲2共兲6 by Theorem 1.3. For the induction step from n to n + 1, apply Proposition 3.6, with ␦ = n, E = En共兲, and Ẽ = En共 / 2兲. This gives, using Hypothesis 1.1 关part 共i兲兴 and the induction hypothesis on 储En共兲 − En共 / 2兲储n, 冐 En+1共兲 − En+1 冉 冊冐 2 n+1 冐 ⱕ 236e18K j兩储v储兩22nr共n兲2共n+1兲6 + q En共兲 − En 冉 冊冐 2 n n+1 ⱕ KEqn+12兩储v储兩2r共兲2共兲6 + 236e18K j兩储v储兩2 兺 qn+1−k2k r共k兲2共k兲6 . k=1 䊏 Proof of Theorem 1.4: By Corollary 3.7, with = 2−m and n = m, we have, for sufficiently large m, 冐冉 E 冊 冉 ⴱ ; ␣ ,  − E m+1 ; ␣ⴱ,  2m 2 冊冐 ⱕ KE2兩储v储兩2 冉冊 冉 冊 冉 冊 兺冉 冊 冉 冊 冉 冊 q 4 2 m 2m 2 m r 6 m−1 + 236e18K j兩储v储兩2 3 q ᐉ 2 r ᐉ ᐉ m 2 ᐉ=0 4 2 2 6 and, consequently, ⬁ 兺 =m 冐冉 E 冊 冉 冊冐 冋兺 冉 冊 冉 冊 冉 冊 冋兺 冉 冊 冉 冊 冉 冊 ⴱ ; ␣ ,  − E +1 ; ␣ⴱ,  2 2 ⬁ ⱕ const 兩储v储兩 2 ⱕ const 兩储v储兩 2 2 =m ⬁ 2 k=m q 4 2 r 2 2 q k r k 4 2 2 k 2 ⬁ 冉 冊 冉 冊 冉 冊册 兺冉 冊 冉 冊 冉 冊 兺 册 −1 6 q ᐉ 2 + 兺 兺 r ᐉ ᐉ 2 2 =m 2 ᐉ=0 4 6 q k 2 + m r k k 2 k=0 4 2 2 ⬁ 6 ⬁ =0 6 1 . 2 Hence, by Hypothesis 1.1 关part 共vi兲兴 and the Cauchy criterion, the sequence E共2−m ; ␣ⴱ , 兲 con䊏 verges uniformly in . This gives Theorem 1.4. APPENDIX A: APPENDIX ON STOKES’ THEOREM Lemma A.1: Let r ⬎ 0 and , ⴱ , 苸 CX obey 兩共x兲 + ⴱ共x兲ⴱ − 共x兲兩 ⬍ 2r for all x 苸 X. Set Dⴱ,,共x兲 = 兵共zⴱ共x兲,z共x兲兲 苸 C2兩兩zⴱ共x兲 − ⴱ共x兲兩 ⱕ r, 兩z共x兲 − 共x兲兩 ⱕ r, z共x兲 − zⴱ共x兲ⴱ = 共x兲其, Dⴱ,, = X Dⴱ,,共x兲. x苸X Let, for each x 苸 X, Cⴱ,,共x兲 be any two real dimensional submanifold of C2 whose boundary is the union of the one real dimensional submanifolds Dⴱ,,共x兲 and the circle 兵共zⴱ共x兲 , z共x兲兲 苸 C2 兩 zⴱⴱ共x兲 = z共x兲 , 兩z共x兲兩 = r其. [By submanifold, we really mean a submanifold with corners. The orientation of Cⴱ,,共x兲 must also be chosen appropriately.] Furthermore, let f共␣1 , . . . , ␣s ; zⴱ , z兲 be a function that is holomorphic in the variables ␣1 , . . . , ␣s in a neighborhood of the origin in CsX and in the variables 共zⴱ , z兲 苸 Xx苸XP共x兲, with, for each x 苸 X, P共x兲 being an open poly disk in C2 Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-16 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. that contains Cⴱ,,共x兲. Then, 冕 D ,, ⴱ = 兿 x苸X 冋 兺兿 R傺X x苸R ⫻ 册 dzⴱ共x兲 ∧ dz共x兲 −z 共x兲z共x兲 f共␣ ,. . .,␣ ;z ,z兲 s ⴱ e 1 e ⴱ 2i 冉冕 C ,,共x兲 ⴱ 兿 冉冕 兩z共x兲兩ⱕr x苸X\R dzⴱ共x兲 ∧ dz共x兲 −z 共x兲z共x兲 e ⴱ 2i 冊 冊 dz共x兲ⴱ ∧ dz共x兲 −z 共x兲z共x兲 f共␣ ,. . .,␣ ;z ,z兲 z 共x兲=z共x兲ⴱ s ⴱ 兩 ⴱ e ⴱ e 1 . 2i for x苸X\R Proof: In the proof we suppress the subscripts ⴱ , , . For each x 苸 X there is a three real dimensional submanifold B共x兲 傺 P共x兲 whose boundary is the union of D共x兲, C共x兲 and DIR共x兲 = 兵共zⴱ共x兲,z共x兲兲 苸 C2兩zⴱⴱ共x兲 = z共x兲,兩z共x兲兩 ⱕ r其. We apply Stokes’ theorem once for each point x 苸 X to the differential form = 兿 x苸X dzⴱ共x兲 ∧ dz共x兲 exp兵− 具zⴱ,z典 + f共␣1, . . . , ␣s ;zⴱ,z兲其. 2i Since is a holomorphic 2兩X兩 form in C2兩X兩, d = 0 and 冕 D = 兺 R傺X 冕 , where M R = MR 兿 DIR共x兲 ⫻ x苸R 兿 C共x兲. x苸R 䊏 Example A.2: In Lemma A.1, C共x兲 = Cⴱ,,共x兲 must be a surface whose boundary coincides with the union of the boundaries of DIR共x兲 and Dⴱ,,共x兲. A possible choice of such a surface is constructed as follows. Interpolate between DIR共x兲 and Dⴱ,,共x兲 by the three dimensional set B共x兲 = 艛0ⱕtⱕ1Dt共x兲, where Dt共x兲 = 兵共zⴱ,z兲 苸 C2兩兩zⴱ − tⴱ共x兲兩 ⱕ r, 兩z − t共x兲兩 ⱕ r, z − zⴱⴱ = t共x兲其. Then C共x兲 = 艛0⬍t⬍1Dt共x兲 has the required boundary Dσ∗ ,σ,ρ (x) C(x) B(x) C(x) DIR (x) Remark A.3: In the above example, Re共zⴱz兲 ⱖ 21 共r2 − 兩共x兲兩2兲 − r共兩共x兲兩 + 兩ⴱ共x兲兩兲 for all 共zⴱ , z兲 苸 C共x兲. Furthermore, the area of C共x兲 is bounded by 8r关兩兩 + 兩ⴱ兩 + 兩兩兴. Proof: Let 共zⴱ , z兲 苸 C共x兲. We suppress the dependence on x. There is a 0 ⱕ t ⱕ 1 such that max兵兩zⴱ − tⴱ兩 , 兩z − t兩其 = r and zⴱ = zⴱ − tⴱ. So zⴱz = 兩z − t兩2 + 2 Re共z − t兲tⴱ + 兩t兩2 − tⴱz, zⴱz = 兩zⴱ − tⴱ兩2 + 2 Re共zⴱ − tⴱ兲tⴱⴱ + 兩tⴱ兩2 + tzⴱ − 兩t兩2 . Adding and taking the real part, Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-17 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II 2 Re共zⴱz兲 = 兩z − t兩2 + 兩zⴱ − tⴱ兩2 + 2 Re共z − t兲tⴱ + 2 Re共zⴱ − tⴱ兲tⴱⴱ + t2共兩兩2 + 兩ⴱ兩2 − 兩兩2兲 ⱖ r2 − 2r共兩兩 + 兩ⴱ兩兲 − 兩兩2 . By construction, C共x兲 is contained in the union of the two cylinders 兵共r + t,rⴱ + t共ⴱ − ⴱ兲兲兩兩兩 = 1, t 苸 关0,1兴其, 兵共rⴱ + t共ⴱⴱ + 兲,r + tⴱ兲兩兩兩 = 1, t 苸 关0,1兴其. The area of the first is bounded by 2冑2r冑兩兩2 + 兩 − 兩2 and the area of the second is bounded by 䊏 2冑2r冑兩ⴱ兩2 + 兩ⴱⴱ − 兩2. APPENDIX B: PROPERTIES OF j„… We discuss the decay properties of the operator j共兲 = e−共h−兲 using the operator norm 共1.10兲. Lemma B.1: (a) For any two operators A , B : L2共X兲 → L2共X兲, (b) For any operator A : L 共X兲 → L 共X兲 and any complex number ␣, 兩储AB储兩 ⱕ 兩储A储兩 兩储B储兩. 2 2 兩储e␣A储兩 ⱕ e兩␣兩兩储A储兩 兩储e␣A − 1储兩 ⱕ 兩␣兩兩储A储兩e兩␣兩兩储A储兩 Proof: 共a兲 By the triangle inequality, for each x 苸 X, 兺 emd共x,y兲兩共AB兲共x,y兲兩 ⱕ y,z苸X 兺 emd共x,z兲兩A共x,z兲兩emd共z,y兲兩B共z,y兲兩 y苸X ⱕ emd共x,z兲兩A共x,z兲兩兩储B储兩 兺 z苸X ⱕ 兩储A储兩 兩储B储兩. 共b兲 The other bound is similar. By part 共a兲, ⬁ 兩储e␣A储兩 ⱕ 兺 n=0 ⬁ 1 1 兩储␣nAn储兩 ⱕ 兺 兩␣兩n兩储A储兩n = e兩␣兩 兩储A储兩 n! n=0 n! and ⬁ 兩储e ␣A − 1储兩 ⱕ 兺 n=1 ⬁ 1 1 兩储␣nAn储兩 ⱕ 兺 兩␣兩n兩储A储兩n ⱕ 兩␣兩 兩储A储兩e兩␣兩 兩储A储兩 . n! n=1 n! 䊏 Corollary B.2: Let ⱖ 0, 兩储j共兲储兩 ⱕ e共兩储h储兩+兲, 兩储j共兲 − 1储兩 ⱕ 共兩储h储兩 + 兩兩兲e共兩储h储兩+兩兩兲 . Proof: Write j共兲 = ee−h and j共兲 − 1 = e共e−h − 1兲 + e − 1. By the previous lemma Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-18 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. 兩储j共兲储兩 = e兩储e−h储兩 ⱕ ee兩储h储兩 and 兩储j共兲 − 1储兩 ⱕ e兩储e−h − 1储兩 + 兩储e − 1储兩 ⱕ 兩储h储兩ee兩储h储兩 + 兩e − 1兩. 䊏 APPENDIX C: THE NORMALIZATION CONSTANT We define the normalization constant Z␦共兲 by the recursion relations Z共兲 = 1, Z2␦共兲 = Z␦共兲2 冕 兩z兩ⱕr共␦兲 dzⴱ ∧ dz −兩z兩2 e . 2i Lemma C.1: For all 0 ⬍ ⬍ ␦ with ␦ / a positive integer power of 2, we have 0 ⬍ Z␦共兲 ⬍ 1 and 兩ln Z␦共兲兩 ⱕ e−r共␦兲 . 2 2 Furthermore, the limit Z␦ = limn→⬁ Z␦共␦ / 2n兲 exists and also obeys 兩ln Z␦兩 ⱕ e−r共␦兲 . Proof: Start by fixing any ⬎ 0 and writing n = 2n. From the inductive definition, ln Zn+1共兲 = 2 ln Zn共兲 + ln Z⬘ , where Z⬘ = n n 冕 兩z兩ⱕr共n兲 dzⴱ ∧ dz −zⴱz e ⬍1 2i so that 0 ⬍ Zn共兲 ⬍ 1 for all n 苸 IN and 2−n−1兩ln Zn+1共兲兩 = 2−n兩ln Zn共兲兩 + 2−n−1兩ln Z⬘ 兩 n which implies that n−1 2−n兩ln Zn共兲兩 = 兺 2−k−1兩ln Z⬘ 兩. 共C1兲 k k=0 Since 1 − Z⬘ = k 冕 兩共x,y兲兩ⱖr共k兲 dxdy −共x2+y2兲 1 e = and 兩ln共1 − x兲兩 ⱕ 兩x兩 / 共1 − 兩x兩兲 ⱕ 2兩x兩 for all 冕 冕 ⬁ 2 dr r共k兲 2 dre−r = 0 冕 ⬁ dse−s = e−r共k兲 2 r共k兲2 兩x兩 ⱕ 21 , n−1 n−1 er共n兲 兩ln Zn共兲兩 = 兺 2n−k−1er共n兲 兩ln共1 − e−r共k兲 兲兩 ⱕ 兺 2n−ke−共r共k兲 2 2 2 k=0 2−r共 兲2兲 n . k=0 By Hypothesis 1.1 关part 共iii兲兴, n−1 n−1 n−1 p=k p=k p=k r共k兲2 − r共n兲2 = 兺 共r共 p兲2 − r共 p+1兲2兲 ⱖ 兺 r共 p兲共r共 p兲 − r共 p+1兲兲 ⱖ 兺 2 = 2共n − k兲 so that n−1 ⬁ er共n兲 兩ln Zn共兲兩 ⱕ 兺 2n−ke−2共n−k兲 ⱕ 兺 2 k=0 ᐉ=1 冉冊 2 e2 ᐉ = 2/e2 ⱕ 1. 1 – 2/e2 For the limit, we rewrite 共C1兲 Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-19 J. Math. Phys. 51, 053306 共2010兲 Complex bosonic effective actions. II n−1 兩ln Zn共兲兩 = 兺 2 n n−k−1 兩ln Z⬘ 兩 = 兺 2ᐉ−1兩ln Z⬘ 兩, k k=0 n−ᐉ ᐉ=1 which implies that 冏 冉 冊冏 ln Z␦ 1 ␦ 2n n = 兺 2ᐉ−1兩ln Z2⬘−ᐉ␦兩 共C2兲 ᐉ=1 and hence 冏 冉 冊冏 lim ln Z␦ n→⬁ 1 ␦ 2n ⬁ = 兺 2ᐉ−1兩ln Z2⬘−ᐉ␦兩 ᐉ=1 ⬁ = 兺 2ᐉ−1兩ln共1 − e−r共2 −ᐉ␦兲2 ᐉ=1 兲兩 ⬁ ⱕe −r共␦兲2 ⬁ 兺2 e ᐉ −共r共2−ᐉ␦兲2−r共␦兲2兲 ᐉ=1 ⱕe −r共␦兲2 兺 2ᐉe−2ᐉ ᐉ=1 ⱕ e−r共␦兲 . 2 APPENDIX D: THE PROOF OF EXAMPLE 1.2 Example 1.2: Let v ⬎ 0. 共i兲 Suppose that 共t兲 = 共ii兲 冉冊 1 1 4 冑兩储v储兩 t a and r共t兲 = 1 冉冊 1 4 冑兩储v储兩 t ar for some constants 0 ⬍ ar ⬍ a obeying 3a + ar ⬍ 1. Then there are constants KE, ⌰, and q such that Hypothesis 1.1 is fulfilled for all nonzero v with 兩储v储兩 ⱕ v. Suppose that 共t兲 = 1 冑t兩储v储兩 4 冉 ln 1 t兩储v储兩 冊 b and 冉 r共t兲 = ln 1 t兩储v储兩 冊 b for some b ⱖ 1. Then there are constants KE, ⌰, and q such that Hypothesis 1.1 is fulfilled for all nonzero v with 兩储v储兩 ⱕ v. Proof: 共i兲 We have 共2t兲 1 = , 共t兲 2a r共2t兲 1 = , r共t兲 2ar t兩储v储兩r共t兲共t兲3 = t1−ar−3a, r共t兲 a −a = t r. 共t兲 So part 共i兲 of the hypothesis is trivially fulfilled if 共1 / 冑4 v兲min兵1 / ⌰ar , 1 / ⌰ak其 ⱖ 1. Part 共ii兲 of the hypothesis, namely, etK j共1 / 2a兲 + 4ta−ar ⱕ 1, is satisfied provided e⌰K j共1 / 2a兲 + 4⌰a−ar ⱕ 1, which is the case if ⌰ is small enough. Part 共iii兲 of the hypothesis, namely, Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp 053306-20 J. Math. Phys. 51, 053306 共2010兲 Balaban et al. r共t兲关r共t兲 − r共2t兲兴 = 1 冑兩储v储兩 t 1 2ar 冋 册 1− 1 ⱖ2 2 ar is satisfied if 共1 / ⌰2ar兲关1 − 1 / 2ar兴 ⱖ 2冑v. Part 共iv兲 of the hypothesis, namely, t1−ar−3a ⱕ 1 / KE, is satisfied provided that ⌰1−ar−3a ⱕ 1 / KE. The uniform convergence 共for each fixed nonzero v兲 of ⬁ t 2 兺 k=0 冉冊冉 冊 冉 冊 q k t r k 4 2 2 t k 2 ⬁ 6 =t 2−2ar−6a 兩储v储兩 −2 兺 k=0 冉 q 2a +6a 2 r 4 冊 k is achieved whenever q ⬍ 22共1−ar−3a兲. Finally, to satisfy part 共v兲, we need 4 2 1 ⱕ 24a ⱕ C共⌰,KE兲 4a C共⌰,KE兲 q 2 r or qⱖ 共ii兲 21–4a C共⌰,KE兲 and 24a+4ar−2 ⱕ C共⌰,KE兲. Since a + ar ⬍ a + 21 a + 21 ar = 21 共3a + ar兲 ⬍ 21 , we have 1 – 4a ⬍ 2共1 − ar − 3a兲 and hence max兵1 , 21–4a其 ⬍ 22共1−ar−3a兲. Fix any q obeying max兵1 , 21–4a其 ⬍ q ⬍ 22共1−ar−3a兲. Then, pick a C ⬍ 1 sufficiently close to 1 that q ⱖ 21–4a / C and 24a+4ar−2 ⱕ C. Then, pick a KE large enough that 1 − 共233e14K j / KE兲 ⬎ C. Finally, choose 0 ⬍ ⌰ ⬍ 1 that is small enough that 共1 / 冑4 v兲min兵1 / ⌰ar , 1 / ⌰ak其 ⱖ 1, e⌰K j共1 / 2a兲 + 4⌰a−ar ⱕ 1, ⌰1−ar−3a ⱕ 1 / KE, 1 / ⌰2ar关1 − 1 / 2ar兴 ⱖ 2冑v, and C共⌰ , KE兲 ⱖ C. Set 冢 C⌰ = 1 − ln 2 2 ln ⌰v 冣 b . Then, for all 0 ⱕ t ⱕ ⌰ / 2 and v with 兩储v储兩 ⱕ v, 冢 r共2t兲 = 1− r共t兲 ln 2 1 ln t兩储v储兩 冣 b 苸 关C⌰,1兲 so that 1 共2t兲 苸 4 关C⌰,1兲, 冑2 共t兲 冉 t兩储v储兩r共t兲共t兲3 = 冑t兩储v储兩 ln The proof now continues as in part 共i兲. 4 1 t兩储v储兩 冊 4b , r共t兲 4 = 冑t兩储v储兩. 共t兲 䊏 1 Balaban, T., Feldman, J., Knörrer, H., and Trubowitz, E., “A functional integral representation for many boson systems. I: The partition function,” Ann. Henri Poincare 9, 1229 共2008兲. 2 Balaban, T., Feldman, J., Knörrer, H., and Trubowitz, E., “A functional integral representation for many boson systems. II: Correlation functions,” Ann. Henri Poincare 9, 1275 共2008兲. 3 Balaban, T., Feldman, J., Knörrer, H., and Trubowitz, E., “Power series representations for complex bosonic effective actions,” J. Math. Phys. 51, 053305 共2010兲. 4 Balaban, T., Feldman, J., Knörrer, H., and Trubowitz, E., “The temporal ultraviolet limit for complex bosonic many-body models,” Ann. Henri Poincare 共to be published兲. 5 Negele, J. W. and Orland, H., Quantum Many-Particle Systems 共Addison-Wesley, Reading, MA, 1988兲. Author complimentary copy. Redistribution subject to AIP license or copyright, see http://jmp.aip.org/jmp/copyright.jsp