Optimization in terms of interval uncertainty: The determinization method

The current approaches to the optimization of systems under uncertainty are considered. An exact formulation of the problem of the conditional optimization of the target function parameters and limitations under interval uncertainty is given. A determinization method allowed us to solve the problem by reducing it to two completely determined ones of conditional optimization of the same type is stated and proved.     

Approximation Algorithms for Quickest Spanning Tree Problems

Let $G=(V,E)$ be an undirected multigraph with a special vertex ${\it root} \in V$, and where each edge $e \in E$ is endowed with a length $l(e) \geq 0$ and a capacity $c(e) > 0$. For a path $P$ that connects $u$ and $v$, the {\it transmission time} of $P$ is defined as $t(P)=\mbox{\large$\Sigma$}_{e \in P} l(e) + \max_{e \in P}\!{(1 / c(e))}$. For a spanning tree $T$, let $P_{u,v}^T$ be the unique $u$--$v$ path in $T$. The {\sc quickest radius spanning tree problem} is to find a spanning tree $T$ of $G$ such that $\max _{v \in V} t(P^T_{root,v})$ is minimized. In this paper we present a 2-approximation algorithm for this problem, and show that unless $P =NP$, there is no approximation algorithm with a performance guarantee of $2 - \epsilon$ for any $\epsilon >0$. The {\sc quickest diameter spanning tree problem} is to find a spanning tree $T$ of $G$ such that $\max_{u,v \in V} t(P^T_{u,v})$ is minimized. We present a ${3 \over 2}$-approximation to this problem, and prove that unless $P=NP$ there is no approximation algorithm with a performance guarantee of ${3 \over 2}-\epsilon$ for any $\epsilon >0$.     

User assisted separation of reflections from a single image using a sparsity prior

When we take a picture through transparent glass the image we obtain is often a linear superposition of two images: the image of the scene beyond the glass plus the image of the scene reflected by the glass. Decomposing the single input image into two images is a massively ill-posed problem: in the absence of additional knowledge about the scene being viewed there are an infinite number of valid decompositions. In this paper we focus on an easier problem: user assisted separation in which the user interactively labels a small number of gradients as belonging to one of the layers. Even given labels on part of the gradients, the problem is still ill-posed and additional prior knowledge is needed. Following recent results on the statistics of natural images we use a sparsity prior over derivative filters. This sparsity prior is optimized using the terative reweighted least squares (IRLS) approach. Our results show that using a prior derived from the statistics of natural images gives a far superior performance compared to a Gaussian prior and it enables good separations from a modest number of labeled gradients.     

The effect of an oxide layer on the ohmicity of contacts to amorphous silicon

Surface oxide layers on CVD films of amorphous silicon can cause the metallic contacts to the films to be non-ohmic. Dipping an amorphous silicon sample in a solution of hydrofluoric acid prior to electrode deposition removes the oxide layer and permits the formation of ohmic contacts.     

A Path-Based Algorithm for the Cross-Nested Logit Stochastic User Equilibrium Traffic Assignment

Abstract:   This article investigates the single-class static stochastic user equilibrium (SUE) problem with separable and additive link costs. A SUE assignment based on the Cross-Nested Logit (CNL) route choice model is presented. The CNL model can better represent route choice behavior compared to the Multinomial Logit (MNL) model, while keeping a closed form equation. The article uses a specific optimization formulation developed for the CNL model, and develops a path-based algorithm for the solution of the CNL-SUE problem based on adaptation of the disaggregate simplicial decomposition (DSD) method. The article illustrates the algorithmic implementation on a real size network and discusses the trade-offs between MNL-SUE and CNL-SUE assignment .     

Effect of static errors in time-domain analyzers on the accuracy of determination of S-parameters

Translated from Izmeritel'naya Tekhnika, No. 7, pp. 11–13, July, 1989.     

Mnemonic facilitation of learning disabled students' memory for expository prose.

In two experiments, learning disabled adolescents read passages that described North American minerals, using materials that incorporated either mnemonic or nonmnemonic illustrations. In Experiment 1, the students were instructed to remember attribute dichotomies (e.g., hard vs. soft minerals), whereas in Experiment 2, the students had to remember specific attribute values (e.g., Hardness Levels 1–20). In both experiments, students learned significantly more when they studied passages with mnemonic pictures. This was true both on immediate performance tests (Experiments 1 and 2) and following a 1-week delay (Experiment 2). Moreover, in Experiment 2, mnemonic instruction facilitated student's ability to make correct inferences about attribute dichotomies—even though such information was never explicitly presented in the lesson. (PsycINFO Database Record (c) 2010 APA, all rights reserved)