Proving conditional attractivity of a closed set with a family of liapunov functions

Considering a closed set M of some x-space and a solution x(t), y(t) of a differential system x = X(x, y, t), y = Y(x, y, t), we give sufficient conditions in order that x(t) approaches M. We use several auxiliary functions and employ Salvadori's method of a one parameter family of Liapunov functions. An application is given to the two-body problem in the presence of some friction forces and when the reference frame is non-inertial.     

On the use of stretched-exponential functions for both linear viscoelastic creep and stress relaxation

The use of the stretched-exponential function to represent both the relaxation function g(t)=(G(t)-G _∞)/(G ₀-G _∞) and the retardation function r(t) = (J _∞+t/η-J(t))/(J _∞-J ₀) of linear viscoelasticity for a given material is investigated. That is, if g(t) is given by exp (?(t/τ)^β), can r(t) be represented as exp (?(t/λ)^µ) for a linear viscoelastic fluid or solid? Here J(t) is the creep compliance, G(t) is the shear modulus, η is the viscosity (η^?1 is finite for a fluid and zero for a solid), G _∞ is the equilibrium modulus G _e for a solid or zero for a fluid, J _∞ is 1/G _e for a solid or the steady-state recoverable compliance for a fluid, G ₀= 1/J ₀ is the instantaneous modulus, and t is the time. It is concluded that g(t) and r(t) cannot both exactly by stretched-exponential functions for a given material. Nevertheless, it is found that both g(t) and r(t) can be approximately represented by stretched-exponential functions for the special case of a fluid with exponents β=µ in the range 0.5 to 0.6, with the correspondence being very close with β=µ=0.5 and λ=2τ. Otherwise, the functions g(t) and r(t) differ, with the deviation being marked for solids. The possible application of a stretched-exponential to represent r(t) for a critical gel is discussed.     

The Cowin–Mehrabadi theorem for an axis of symmetry

The well-known Cowin–Mehrabadi Theorem deals with necessary and sufficient conditions for a normal n to a symmetry plane. Necessary conditions require that n be a common eigenvector of C_ijkk, C_ikjk and C_ijkln_jn_l. It is shown that a vector parallel to an axis of symmetry must also satisfy these conditions. An axis of rotational symmetry is also a normal to a plane of symmetry except in the case of a trigonal material. Being a common eigenvector of C_ijkk and C_ikjk belonging to a nondegenerate eigenvalue guarantees it to be an axis of symmetry.     

Propagation of Singularities¶for Solutions of Nonlinear First Order¶Partial Differential Equations

Few results are available in the mathematical literature for studying the structure of the singular set of a weak solution u of F(x,u,Du)=0. This paper provides new techniques to analyse such a set when u is semiconcave and F is a nonlinear convex function with respect to p. The main objective achieved here is a classification of the singularities of u that propagate along Lipschitz arcs. Such a propagation phenomenon is also described by means of a generalized characteristics inclusion.     

The unsteady heat transfer due to a heat source in an MHD stretching sheet flow

The time evolution in the temperature field resulting from the sudden introduction of a heat source into the already fully established steady MHD flow of an electrically conducting fluid past a linearly stretching isothermal surface is considered. The problem is shown to be fully described by two dimensionless parameters, a modified magnetic field strength ?? and a heat source strength Q. Numerical solutions of the initial-value problem show that there is a critical value Q _c of the parameter Q, dependent on ??, such that, for Q<Q _c , the solution approaches a steady state at large times and, for Q>Q _c , the solutions grows exponentially large as time increases. This growth rate is determined through an eigenvalue problem which also determines the critical value Q _c . The limits of Q _c for both small and large values of ?? are discussed.     

The comparison model method: A new arithmetic approach to the discrete inverse problem of groundwater hydrology

Let the steady-state pressure z(·) of a fluid in a one-dimensional domain be governed by the equation d _x (a d _x z) = f subject to Dirichlet boundary conditions. We consider the identification of the transmissivity a (·), given f(·), and measured pressure z(·) by the comparison model method, a direct method which has been known and applied for some time but lacked theoretical background. With reference to a domain in one spatial dimension, we examine both the infinite-(‘continuous’) and finite-(discrete) dimensional cases. In the former, the method is based on the solution p(·) of an auxiliary flow equation, where f(·) and the two-point boundary conditions are unchanged with respect to the original or z(·) equation, whereas a tentative constant value b is assigned to the auxiliary transmissivity. The ratio of the first derivatives of p(·) and z(·) multiplied by b yields a solution ã(·) to the inverse problem. We examine in detail the nonuniqueness of ã(·) as a function of b, some cases where existence implies uniqueness, the role of positivity constraints, and a special feature: self-identifiability. We then translate all available results into the discrete case, where the good unknowns for the inverse problem are the internode coefficients. Several algebraic and numerical examples are presented.     

Stress concentration in a uniformly rotating circular plate weakened by an elliptic hole and two rectilinear cuts

We consider a uniformly rotating isotropic plate whose cross-section is a two-connected domain bounded from the outside by a circle of radius R and from inside by an ellipse with semiaxes a and b. The rectilinear cuts coming to the ellipse are located symmetrically on the real axis OX. On the imaginary axis OY, two point masses m are located at the distances ±id from the center of the plate. Such a plate can be considered as a plate under the action of bulk forces (forces of inertia), X, Y, and under the action of two lumped tensile forces P = mdw ² applied at the points ±id.     

Radial Solutions of¶Superlinear Equations on ℝN¶Part II: The Forced Case

We study the effect of a forcing term in the context of the search of multiple nodal solutions u∈h ¹(? ^N ) to a class of elliptic equations of type¶?Δu(x)=f(|x|,u(x))+h(|x|), x∈? ^N ,¶where f(|x|≡0 and f is superlinear but subcritical at infinity.     

Homogenization of a degenerate parabolic problem in a highly heteregeneous mediumHomogénéisation d'un problème parabolique dégénéré dans un milieu fortement hétérogène

We consider the homogenization of a time-dependent heat transfer problem in a highly heteregeneous periodic medium made of two connected components having finite heat capacities c_α(x) and heat conductivities a_α(x), α=1,2, of order one, separated by a third material with thickness of order ε the size of the basic periodicity cell, but with conductivity λa₃(x) where a₃=O(1) and λ tends to zero with ε. Assuming only that c_i(x)?0 a.e., such that the problem can degenerate (parabolic-elliptic), we identify the homogenized problem following the values of δ=lim_ε→0ε²/λ. To cite this article: M. Mabrouk, A. Boughammoura, C. R. Mecanique 331 (2003).     

On the dispersion of a solute in oscillating flow through a channel

The paper presents an exact analysis of the dispersion of a solute in an incompressible viscous fluid flowing slowly in a parallel plate channel under the influence of a periodic pressure gradient. Using a generalised dispersion model which is valid for all times after the solute injection, the diffusion coefficientsK _i (τ)(i=1,2,3,…) are determined as functions of timeτ when the initial distribution of the solute is in the form of a slug of finite extent. The second coefficientK ₂(τ) gives a measure of the longitudinal dispersion of the solute due to the combined influence of molecular diffusion and nonuniform velocity across the channel cross-section. The analysis leads to the novel result thatK ₂(τ) consists of a steady partS and a fluctuating partD ₂(τ) due to the pulsatility of the flow. It is shown thatS increases with increase inλ (the amplitude of pressure pulsation) for small values ofω (the frequency of the pulsation). But for largeω, S decreases with increase inλ. It is also found that for fixedλ, there is very little fluctuation inD ₂(τ) forω=1, butD ₂(τ) shows fluctuation with large amplitude whenω slightly exceeds unity. The amplitude ofD ₂(τ) then decreases with further increase inω. Thus the variation of bothS andD ₂(τ) withω is non-monotonic. Finally,? _m , the average concentration of the solute over the channel cross-section is determined for various values ofλ andω.     

Effect of spanwise-varying local forcing on turbulent separated flow over a backward-facing step

An experimental study was made of turbulent separated flows over a backward-facing step. A local forcing was given to the separated flow by means of a sinusoidally oscillating jet issuing from a thin slit near the separation line. To produce a spanwise-varying local forcing at the separation edge, a banded thin tape covered the slit. Effects of the spanwise-varying local forcings on the separated flow were scrutinized by altering the spatially banded blocking width (w) and the open slit distance (g). An optimal value of w/g was sought, which led to the minimum reattachment length (x _R ). The effect of spanwise-varying local forcing on x _R was found to be slight compared to the case of two-dimensional forcing (w=0). The experiment was made at Re _H =33000 and A ₀=0.018 by changing the forcing frequency (0?St _H ?1.0).     

Radial Solutions of¶Superlinear Equations on ℝN¶Part I: Global Variational Approach

We present a global variational approach to the search for multiple nodal solutions u∈H ¹(? ^N ) to a class of elliptic equations of type¶?Δu(x)=f(|x|,u(x)), x∈? ^N ,¶where N≧ 2, f is superlinear and subcritical, and f(|x|≡0.     

A note on integration of the Emden-Fowler equation

Using a simple change of variables, the Emden-Fowler equation, (x^{v + α}y′)′ + ax^vyⁿ = 0 is shown to be integrable provided that either of the constraints (v + α ? 1)n = 3 ? α + v or (v + α ? 1)n = 3 ? 2α ? v is satisfied. Every integrable case generates a one parameter family of integrable Emden-Fowler equations.     

Turbulent film boiling from a vertical non-isothermal surface

The turbulent film boiling from a vertical non-isothermal surface is formulated with due consideration to thermal radiation from its lateral face. It is observed that the application of Reynolds analogy together with thermal conduction in the test surface has yielded a conjugate solution from which the case of an isothermal condition can be generated as a special case. The analysis has further paved the way in establishing a functional relation between the Nusselt numberNu, radiation parameterN _R , fin parameterM, temperature ratio termT _s /(T _w,0?T _s ), and a product of characteristic modified Grashof, Prandtl and superheating parameter defined as (Gr ² Pr S). In a fully developed turbulent film boiling i.e., modified Grashof number being greater than 10¹⁰, the temperature ratio term accounts for the non-linearities arising due to the inclusion of radiation from the lateral face of the fin. The results are in good agreement with experimental data over a wide range of system conditions.     

Asymptotic expansion of a class of integral transforms via Mellin transforms

An asymptotic expansion for large λ of functions I(λ) defined by definite integrals of the form $$I(\lambda ) = \mathop \smallint \limits_0^\infty h(\lambda t)f(t)dt$$ is obtained in the case where h(t)=O(exp(-βt ^p )) as t→∞ with β, ?>0. To obtain the expansion for such integral transforms, I(λ) is first represented as a contour integral involving M [h; z], the Mellin transform of the kernel h(t) evaluated at z, and M[f; 1-z], the Mellin transform of the function f(t) evaluated at 1-z. By assuming a rather general asymptotic expansion for f(t) near t=0, it is shown that M[f; 1-z] can be continued into the right-half plane as a meromorphic function with poles that can be located and classified. The desired asymptotic expansion of I is then obtained by systematically moving the contour in its integral representation to the right. Each term in the expansion arises as a residue contribution corresponding to a pole of M[f; 1-z]. It is then shown how the expansion, originally found for large positive λ, can be extended to complex λ. Finally several examples are considered which illustrate the scope of our expansion theorems.     

Large strain deformation of polycrystalline metals at low homologous temperatures

Large strain compression data (true strains to about ?3.0) are presented for polycrystalline α U and α Fe at room temperature. The results, together with other published data at low homologous temperatures (≈0.2 T_m), where T_m is the absolute melting temperature, suggest that a steady-state flow stress σ_s is approached after extensive strain-hardening, α U exhibits a very high strain-hardening rate, with σ_s ≈ 2900 MPa (420 ksi) indicating that cold-working is a very potent method of strengthening this metal. All the data evaluated can be fit by the stress-strain relation σ = σ_s? exp (?(Nε)^p)(σ_s? σ_y), where σ_y is the yield stess, p is a constant equal to a for the metals analyzed, N is a constant associated with the strain-hardening characteristics of a material, σ is true stress, and ε is true strain.     

Existence and construction of quadratic invariants for the equation d2dξ2ν = λ(ξ)ddξν + f(ξ,ν)

If the equation ν_ζζ = λ(ξ)ν_ζ + F(ξ,ν) admits a first integral quadratic in the momentum c., the function F(ξ,ν) is defined; a new procedure is introduced to find the first integral. Among the examples is discussed the existence of quadratic invariants for the equation u″ λu′ = Q(u), where λ is a constant, and for the equation ν_ξ,ξ =βξ^σνⁿ. This last equation admits a first integral, quadratic in ν_ξ if σ = 0, or n = −2σ − 3, or n = −σ − 3. Particular eases are exhibited where no invariant quadratic in ν_ξ is found for the equation ν_ξξ = βξ^σνⁿ but it admits a first integral which is a polynomial of degree 4 or 6 in ν_ξ.     

On a vibrating plate with a concentrated mass

We consider the vibrations of an elastic plate that contains a small region whose size depends on a small parameter ε. The density is of order O(ε^−m) in the small region, the concentrated mass, and it is of order O(1) outside; m is a positive parameter. Depending on the value of m (m<2, m=2 and m>2) we describe the asymptotic behaviour, as ϵ→0, of the eigenvalues and eigenfunctions of the corresponding spectral problem. For m>2 the vibrations associated with the low frequencies affect asymptotically only a neighbourhood of the concentrated mass; we also consider the asymptotic behaviour of the eigenfunctions associated with the high frequencies.