Optimal asset allocation for pension funds under mortality risk during the accumulation and decumulation phases

In a financial market with one riskless asset and n risky assets whose prices are lognormal, we solve in a closed form the problem of a pension fund maximizing the expected CRRA utility of its surplus till the (stochastic) death time of a representative agent. We consider a unique asset allocation problem for both accumulation and decumulation phases. The optimal investment in the risky assets must decrease during the first phase and increase during the second one. We accordingly suggest it is not optimal to manage the two phases separately, and outsourcing of allocation decisions should be avoided in both phases. JEL: G23, G11 MSC 2000: 62P05, 91B28, 91B30, 91B70, 93E20     

Convex Sobolev inequalities and spectral gap

This Note is devoted to the proof of convex Sobolev (or generalized Poincaré) inequalities which interpolate between spectral gap (or Poincaré) inequalities and logarithmic Sobolev inequalities. We extend to the whole family of convex Sobolev inequalities results which have recently been obtained by Cattiaux, and Carlen and Loss for logarithmic Sobolev inequalities. Under local conditions on the density of the measure with respect to a reference measure, we prove that spectral gap inequalities imply all convex Sobolev inequalities including in the limit case corresponding to the logarithmic Sobolev inequalities. To cite this article: J.-P. Bartier, J. Dolbeault, C. R. Acad. Sci. Paris, Ser. I 342 (2006).     

The ALTO project at IPN-Orsay

The availability of both a tandem deuteron beam and a linac electron beam, the latter converted into Bremsstrahlung, at the new ALTO facility at IPN-Orsay offers a unique opportunity to compare the performance of a laser ion guide under different regimes. The ALTO accelerator has delivered its first electron beam at the end of 2005 and a design for a gas-cell prototype is being studied.     

Convergence in probability for perturbed stochastic integral equations

Summary In this work, one considers two stochastic integral equations indexed by some parameter and one studies the contiguity of their solutions when the parameter converges to some ₀. Two types of behaviour are described; they lead to the notion of regular and singular perturbations. The method which is used also enables a study of the rate of convergence. Applications to time discretization of equations are given.     

Stabilizing an uncertain production system

Consider a production system that consists ofm machines each of which can produce parts ofn types. When machinek is used, it produces a part of typei with probabilityp _ki . Requests arrive for parts, one at a time. With probability _i an arriving request is for a part of typei. The requests must be served without waiting. Thus, if a requested part is not available, it must be produced. We find necessary and sufficient conditions for the existence of a strategy (a choice of the machines to be used) which makes the inventory of parts stable and we provide such a strategy.Two variations of this model are also considered: the case of batch arrivals of requests, and that of a system where the requests can be queued.     

An improved algorithm for satellite-derived UV radiations

The improved algorithm surface irradiance derived from a range of satellite-based sensors (SIDES) is presented in this article. It calculates various types of surface UV intensities, such as biologically weighted or unweighted UV spectra, integrated doses or irradiance at specific wavelengths, using data from satellite instruments. These surface UV data are mainly useful for environmental impact or process studies where high accuracy or a high temporal resolution is required. In contrast to several previous studies, SIDES has been validated with spectral measurements. By this method an averaging of positive or negative deviations over the complete wavelength range is avoided. This is especially important for UV wavelengths around 300 nm where biological effectiveness is highest. The results of SIDES deviate less than 7% from ground-based observations for wavelengths between 295 and 400 nm. In contrast, the corresponding deviations of the joint research center algorithm escalate for shorter wavelengths, reaching 35% at 295 nm. This large deviation is due to an inaccurate interpolation procedure that has been detected by spectral analysis. Thus, spectral validation is demonstrated to be an appropriate tool to detect weaknesses in such an algorithm and provides information essential for improvement.