Boundary term:

Therefore the second-order derivative of/ appearing in the original form of / is replaced by a term involving first-order derivatives of w and/plus a boundary term. The boundary terms are, normally, cancelled out through the assembly of the elemental stiffness equations over the common nodes on the shared interior element sides and only appear on the outside boundaries of the solution domain. However, as is shown later in this chapter, the appropriate treatment of these integrals along the outside boundaries of the flow domain depends on the prescribed boundary conditions. 

This is the mechanical version of the Reverse Transition Theorem [4]. The first three factors are boundary terms, whereas the final exponent scales with the length of the time interval. 

The physical motivation behind this choice is that S now becomes an antisymmetric observable under time reversal. Albeit 5 p(r) is always antisymmetric, the choice of Eq. (25) is the only one that guarantees that the total dissipation S changes sign upon reversal of the path, iS(T ) = —iS(T). The symmetry property of observables under time reversal and the possibility of considering boundary terms where S is symmetric (rather than antisymmetric) under time reversal has been discussed in Ref. 43. 

It is interesting to observe that this relation is not satisfied by the entropy production because the inclusion of a boundary term, Eq. (24), in the total dissipation is required to respect the fluctuation symmetry. In what follows we discuss some of its consequences in some specific situations. 

In general, it can be very difficult to determine the nature of the boundary terms. A specific result in an exactly solvable case is discussed in Section IV.A.2. Equation (55) is the Gallavotti-Cohen FT derived in the context of deterministic Anosov systems [28]. In that case, Sp stands for the so-called phase space compression factor. It has been experimentally tested by Ciliberto and co-workers in Rayleigh-Bemard convection [52] and turbulent flows [53]. Similar relations have also been tested in athermal systems, for example, in fluidized granular media [54] or the case of two-level systems in fluorescent diamond defects excited by light [55]. 

The FT in Eq. (27) also describes fluctuations in the total dissipation for transitions between steady states, where X varies according to a given protocol. In that case, the system starts at time 0 in a given steady state, F (C), and evolves away from that steady state at subsequent times. The boundary term for steady-state transitions is then given by... 

It is important to stress that Eq. (82) does not satisfy Eqs. (48) and (49) because the last boundary term on the rhs of Eq. (82) (vy A//kT) is not antisymmetric against time reversal. Van Zon and Cohen [61-63] have analyzed in much detail work and heat fluctuations in the NESS. They find that work fluctuations satisfy the exact relation... 

Wx or Wf Let us suppose that/is the control parameter. In this case the JE and GET, Eqs. (40) and (41), are valid for the work, Eq. (96). How large is the error that we make when we apply the JE using Wx instead This question has been experimentally addressed by Ciliberto and co-workers [97, 98], who measured the work in an oscillator system with high precision (within tenths of fesT). As shown in Eq. (99), the difference between both works is mainly a boundary term, A xf). Fluctuations of this term can be a problem if they are on the same order as fluctuations of Wx itself. For a harmonic oscillator of stiffness constant equal to k, the variance of fluctuations mfx are equal to k8(x ), that is, approximately on the order of k T due to the fluctuation-dissipation relation. Therefore, for experimental measurements that do not reach such precision, Wx or Wf is equally good. 

Obviously, the full trace of the kernel in Eq. 3.9 is obtained by imposing appropriate boundary conditions, and integrating with respect to all the variables without the boundary term. Analogously, one can also derive the parallel-transport operator (a generalization of the one just considered) for symmetric tensors (bosonic -particle states) and for forms (fermionic -particle states). 

Taking the variation derivative of the action A 2 with respect to A and A we obtain J2k f k an<4 Sfc fik respectively. Therefore up to the boundary terms the functional (99) should coincide with 2fcTrln[<9r + TLk]. Moreover, the possible boundary terms should vanish because the Green functions of initial t = —00 and final r = 00 coincide. Writing the action (99) in terms of the original functions g, /, /t and using Eqs. (47-49) to bring it into symmetric form for A2 we finally get... 

Recently, there has been interest in considering FR for systems coupled to heat reservoirs at different temperatures." The relevant FR for systems modelled using deterministic dynamics was determined and tested numerically in 2001, however interest in heat flow for simple models is important for studies aimed at determining the nonequilibrium temperatures and studying Fourier s Law, and therefore consideration of FR in these systems has received some attention. Gomez-Marin and Sancho and were unable to verify an FR proposed by Jarz5mski and Wojcik, presumably due to neglect of boundary terms in the theoretical analysis that were relevant for their numerical calculations. Visco" has studied the GC FR in a model heat flow problem. 

Figure 1. The essence of the QM/MM approach is that a small region (e.g. at an enzyme active site) is treated by a quantum chemical method, while the bulk is represented more simply by molecular mechanics. It is often necessary also to apply a boundary term because of the finite size of the simulation system.

The boundary terms vanish by the assumed conditions on / and g. A second integration by parts transforms Eq (4.3) to... 

Each of these terms will be discussed in detail in the following four sections. It is to be noted, however, that, in general, the purely QM and MM parts of the Hamiltonian are essentially the same as for calculations on purely QM or MM systems respectively. The most important differences arise in the form of the interaction Hamiltonian, Hqm/mMi and, to a lesser extent, of the boundary terms. 

Note that to arrive at this result we have carried out an integration by parts in which it was assumed that all boundary terms go to zero. 

The Galerkin-weighted residual equations are obtained by inserting the test function into the SDT, with the boundary term deleted ... 

Frohlich s relations were also used in a slightly modified form of the perturbation theory power series in the dimensionless parameter (R — Ro)/Ro [57], In its current form, derivatives like 9rE(R) are usually derived on the basis of commutation relations as a consequence of some hypervirial relations in R3 (see [44-46]). It is clear that these derivatives appear as the boundary terms in the usual integral relations (see also Sections 3.2 and 3.3). 

The forces so far discussed are those associated with the Poisson equation. It is not surprising that the situation is somewhat more complex when an electrolytic solvent is treated by using the Poisson-Boltzmann equation. However, it can be shown O i that the force density expression in this case is identical to that for the Poisson equation, except for the addition of an ion boundary pressure term. Like the dielectric boundary term, this represents a pure pressure acting at the ion exclusion boundary of the solute. It results from... 

In Fermi s original derivation [48] as well as in the one just given [47] it is clear that the Fermi-contact term arises as a boundary term from an integration by parts. In spite of this one can find rather hazardous explanations of this term in the literature. For a review see Ref. [47]... 

The turn-over rule can be used naively, if the integrand has no singularities at the boundaries. In the present case there are singularities at r = 0, then additional boundary terms have to be considered, or - equivalently - all dilferentiations have to be understood in the distribution sense [47], which leads to 5-function terms. 

Note that in general, boundary terms involving some appropriately defined u9 should be introduced to ensure the nonnegativity of the distribution. For the simultaneous computation of matrix elements for several values of the... 

---