Discontinuous Galerkin method with a posterioriL_p(0,t_i) error estimate for second-kind Volterra problems

Summary. We consider a discontinuous Galerkin finite element method applied in time to a model Volterra equation of the second kind. A residual-based computable Galerkin-error estimate is derived for . This estimate does not explicitly contain the time step and therefore the time step control must be based on a heuristic criterion, the estimate can then be used to demonstrate the integrity, or otherwise, of the finite element solution. After performing some numerical experiments we conclude that this approach is at least competetive with classical discretizations since it is computationally simple to implement, but has the added advantage of reliable error feedback. Received June 25, 1995     

The Small Index Property for Free Groups and Relatively Free Groups

1. Some groups with the small index property In this paper we shall use the methods of the paper [6] by Hodges,Hodkinson, Lascar and Shelah to show that the free group ofcountably infinite rank and certain relatively free groups ofcountably infinite rank have the small index property.     

A two-echelon base-stock inventory model with Poisson demand and the sequential processing of orders at the upper echelon

We consider a two-echelon inventory system with a number of non-identical, independent ‘retailers’ at the lower echelon and a single ‘supplier’ at the upper echelon. Each retailer experiences Poisson demand and operates a base stock policy with backorders. The supplier manufactures to order and holds no stock. Orders are produced, in first-come first-served sequence, with a fixed production time. The supplier therefore functions as an M/D/1 queue. We are interested in the performance characteristics (average inventory, average backorder level) at each retailer. By finding the distribution of order lead time and hence the distribution of demand during order lead time, we find the steady state inventory and backorder levels based on the assumption that order lead times are independent of demand during order lead time at a retailer. We also propose two alternative approximation procedures based on assumed forms for the order lead time distribution. Finally we provide a derivation of the steady state inventory and backorder levels which will be exact as long as there is no transportation time on orders between the supplier and retailers. A numerical comparison is made between the exact and approximate measures. We conclude by recommending an approach which is intuitive and computationally straightforward.     

Soluble models for dynamics driven by a super-diffusive noise

We explicitly discuss scalar Langevin type of equations where the deterministic part is linear, but where the integrated noise source is a non-linear diffusion process exhibiting superdiffusive behavior. We calculate transient and stationary probabilities and study the possibility of noise induced transitions from a unimodal to a bimodal probability shape. Illustrations from finance and dynamical systems are given.     

Self-inhibition or the lack of it in the pyrolyses of neopentane and n-hexane

Published data show that in its early stages (up to 3% decomposition), the pyrolysis of n-hexane in the ranges 723–823 K and 10–100 Torr is not inhibited by the olefin products, in contrast with neopentane pyrolysis which is very strongly inhibited in similar conditions. Detailed consideration of the chain mechanisms for the two pyrolyses shows that the reactivity of the chain terminating radical towards hydrogen abstraction from an allylic C? H bond in product olefin is the factor which determines whether or not observable self-inhibition occurs. Thus, n-hexane pyrolysis, whose chain decomposition is terminated by recombination and disproportionation of ethyl, is not significantly self-inhibited, whereas that of neopentane which is terminated by recombination of methyl is very strongly inhibited because methyl is 14× more reactive than ethyl. The implications for other alkanes are briefly discussed.     

Stock Control and the Undershoot of the Re-order Level

Practical stock-control systems are frequently based on a policy of reviewing the stock balance and placing an order if this has fallen to or below a set level R. In order to facilitate the calculation of the control parameters, the assumption is generally made that the stock balance is exactly R when an order is placed, whereas in practice it may have fallen below R. This paper uses a steady-state analysis to evaluate some relatively simple adjustments which may be incorporated to handle this undershoot.