Adaptive eigenvalue computation: complexity estimates

This paper is concerned with the design and analysis of a fully adaptive eigenvalue solver for linear symmetric operators. After transforming the original problem into an equivalent one formulated on ℓ ₂, the space of square summable sequences, the problem becomes sufficiently well conditioned so that a gradient type iteration can be shown to reduce the error by some fixed factor per step. It then remains to realize these (ideal) iterations within suitable dynamically updated error tolerances. It is shown under which circumstances the adaptive scheme exhibits in some sense asymptotically optimal complexity. This research was supported in part by the Leibniz Programme of the DFG, by the SFB 401 funded by DFG, the DFG Priority Program SPP1145 and by the EU NEST project BigDFT.     

Lower complexity estimates for some sequences of functions of multivalued logic

The problem of realization of functions of a multivalued logic by formulas is considered. Some sequences of functions whose realization complexity exceeds exponential one are presented.     

Lower estimates of circuit complexity in the basis of antichain functions

The antichain function is a characteristic function of an antichain in the Boolean cube. The set of antichain functions is an infinite complete basis. We study the computational complexity of Boolean functions over an antichain functional basis. In this paper we prove an asymptotic lower bound of order $\sqrt n $ for the computational complexity of a linear function, a majority function, and almost all Boolean functions of n variables.     

Accurate estimates of the data complexity and success probability for various cryptanalyses

Many attacks on encryption schemes rely on statistical considerations using plaintext/ciphertext pairs to find some information on the key. We provide here simple formulae for estimating the data complexity and the success probability which can be applied to a lot of different scenarios (differential cryptanalysis, linear cryptanalysis, truncated differential cryptanalysis, etc.). Our work does not rely here on Gaussian approximation which is not valid in every setting but use instead a simple and general approximation of the binomial distribution and asymptotic expansions of the beta distribution.     

Improvement in peak stress estimates through post-processing

Three alternatives for calculating the peak stress in the vicinity of a stress raiser have been proposed in previous research: an iterative method, a stiffness derivative method, and a table look-up method. The goal of each is to improve the estimate of the peak stress through post-processing rather than multiple analyses utilizing mesh refinement. In the present paper, the relative effectiveness of the methods is studied by means of test cases. An improved stiffness derivative method is suggested for use in fracture mechanics applications.     

On the complexity of project scheduling to minimize exposed time

We consider project scheduling where the project manager’s objective is to minimize the time from when an adversary discovers the project until the completion of the project. We analyze the complexity of the problem identifying both polynomially solvable and NP-hard versions of the problem. The complexity of the problem is seen to be dependent on the nature of renewable resource constraints, precedence constraints, and the ability to crash activities in the project.     

Gradient methods with regularization for constrained optimization problems and their complexity estimates

We suggest simple implementable modifications of conditional gradient and gradient projection methods for smooth convex optimization problems in Hilbert spaces. Usually, the custom methods attain only weak convergence. We prove strong convergence of the new versions and establish their complexity estimates, which appear similar to the convergence rate of the weakly convergent versions. Preliminary results of computational tests confirm efficiency of the proposed modification.     

A linear Bayesian stochastic approximation to update project duration estimates

By relaxing the unrealistic assumption of probabilistic independence on activity durations in a project, this paper develops a hierarchical linear Bayesian estimation model. Statistical dependence is established between activity duration and the amount of resource, as well as between the amount of resource and the risk factor. Upon observation or assessment of the amount of resource required for an activity in near completion, the posterior expectation and variance of the risk factor can be directly obtained in the Bayesian scheme. Then, the expected amount of resources required for and the expected duration of upcoming activities can be predicted. We simulate an application project in which the proposed model tracks the varying critical path activities on a real time basis, and updates the expected project duration throughout the entire project. In the analysis, the proposed model improves the prediction accuracy by 38.36% compared to the basic PERT approach.     

Optimal strategies for selecting project portfolios using uncertain value estimates

Practically all organizations seek to create value by selecting and executing portfolios of actions that consume resources. Typically, the resulting value is uncertain, and thus organizations must take decisions based on ex ante estimates about what this future value will be. In this paper, we show that the Bayesian modeling of uncertainties in this selection problem serves to (i) increase the expected future value of the selected portfolio, (ii) raise the expected number of selected actions that belong to the optimal portfolio ex post, and (iii) eliminate the expected gap between the realized ex post portfolio value and the estimated ex ante portfolio value. We also propose a new project performance measure, defined as the probability that a given action belongs to the optimal portfolio. Finally, we provide analytic results to determine which actions should be re-evaluated to obtain more accurate value estimates before portfolio selection. In particular, we show that the optimal targeting of such re-evaluations can yield a much higher portfolio value in return for the total resources that are spent on the execution of actions and the acquisition of value estimates.     

Teaching mathematical modelling through project work

The paper presents and analyses experiences from developing and running an inservice course in project work and mathematical modelling for mathematics teachers in the Danish gymnasium, e.g. upper secondary level, grade 10~12. The course objective is to support the teachers to develop, try out in their own classes, evaluate and report a project based problem oriented course in mathematical modelling. The in-service course runs over one semester and includes three seminars of 3, 1 and 2 days. Experiences show that the course objectives in general are fulfilled and that the course projects are reported in manners suitable for internet publication for colleagues. The reports and the related discussions reveal interesting dilemmas concerning the teaching of mathematical modelling and how to cope with these through «setting the scene» for the students modelling projects and through dialogues supporting and challenging the students during their work. This is illustrated and analysed on the basis of two course projects.     

Proof of lower estimates for the complexity of self-correcting circuits by the method of basis changing

A method for obtaining new lower estimates for the implementation complexity of individual Boolean functions is presented in the paper. The method is based on the transition from some considered basis to another one possessing already known good lower estimates for the complexity of those functions. The effective use of this method is illustrated by the example of obtaining the asymptotic value for the implementation complexity of threshold functions by self-correcting circuits composed of multiple-input elements.     

Decay estimates of functions through singular extensions of vector-valued Laplace transforms

Let X be a Banach space and let f∈L^∞(R⁺;X) whose Laplace transform extends analytically to some region containing iR?{0}, possibly having a pole at the origin. In this paper, we give estimates of the decay of certain slight suitable modification of f in terms of the growth of its Laplace transform along the imaginary axis. This technique is applied to obtain decay estimates of smooth orbits of bounded C₀-semigroups whose infinitesimal generators have an arbitrary finite boundary spectrum. These results are close to those given recently by C.J.K. Batty and T. Duyckaerts.     

Density estimates for a random noise propagating through a chain of differential equations

We here provide two sided bounds for the density of the solution of a system of n differential equations of dimension d, the first one being forced by a non-degenerate random noise and the n−1 other ones being degenerate. The system formed by the n equations satisfies a suitable Hörmander condition: the second equation feels the noise plugged into the first equation, the third equation feels the noise transmitted from the first to the second equation and so on … , so that the noise propagates one way through the system. When the coefficients of the system are Lipschitz continuous, we show that the density of the solution satisfies Gaussian bounds with non-diffusive time scales. The proof relies on the interpretation of the density of the solution as the value function of some optimal stochastic control problem.     

Synthesis of formulas whose complexity and depth do not exceed the asymptotically best estimates of high degree of accuracy

We study problems concerning optimal realizations of arbitrary Boolean functions by formulas in the standard basis {&;, V, ¬} in the presence of two optimality criteria: the depth and the complexity. Both the complexity and depth of Boolean functions are investigated from the point of view of so-called asymptotically best estimates of high degree of accuracy for the corresponding Shannon functions. Such estimates produce an asymptotics not only for the Shannon function, but also for the first remainder term of its standard asymptotic expansion. For any Boolean function depending on n variables, we prove that it is possible to construct a realizing formula in the basis {&;, V, ¬} so that its depth and complexity do not exceed values of the corresponding Shannon functions for the argument equal to n in the sense of asymptotic estimates of high degree of accuracy.