Misclassification minimization

The problem of minimizing the number of misclassified points by a plane, attempting to separate two point sets with intersecting convex hulls inn-dimensional real space, is formulated as a linear program with equilibrium constraints (LPEC). This general LPEC can be converted to an exact penalty problem with a quadratic objective and linear constraints. A Frank-Wolfe-type algorithm is proposed for the penalty problem that terminates at a stationary point or a global solution. Novel aspects of the approach include: (i) A linear complementarity formulation of the step function that counts misclassifications, (ii) Exact penalty formulation without boundedness, nondegeneracy or constraint qualification assumptions, (iii) An exact solution extraction from the sequence of minimizers of the penalty function for a finite value of the penalty parameter for the general LPEC and an explicitly exact solution for the LPEC with uncoupled constraints, and (iv) A parametric quadratic programming formulation of the LPEC associated with the misclassification minimization problem.This material is based on research supported by Air Force Office of Scientific Research Grant F49620-94-1-0036 and National Science Foundation Grants CCR-9101801 and CDA-9024618.     

Rate conservation laws: A survey

We survey the rate conservation law, RCL for short, arising in queues and related stochastic models. RCL was recognized as one of the fundamental principles to get relationships between time and embedded averages such as the extended Little's formulaH=G, but we show that it has other applications. For example, RCL is one of the important techniques for deriving equilibrium equations for stochastic processes. It is shown that the various techniques, including Mecke's formula for a stationary random measure, can be formulated as RCL. For this purpose, we start with a new definition of the rate with respect to a random measure, and generalize RCL by using it. We further introduce the notion of quasi-expectation, which is a certain extension of the ordinary expectation, and derive RCL applicable to the sample average results. It means that the sample average formulas such asH=G can be obtained as the stationary RCL in the quasi-expectation framework. We also survey several extensions of RCL and discuss examples. Throughout the paper, we would like to emphasize how results can be easily obtained by using a simple principle, RCL.     

Trend to equilibrium of weak solutions of the Boltzmann equation in a slab with diffusive boundary conditions

Recently R. Illner and the author proved that, under a physically realistic truncation on the collision kernel, the Boltzmann equation in the one-dimensional slab [0, 1] with general diffusive boundary conditions at 0 and 1 has a global weak solution in the traditional sense. Here it is proved that when the Maxwellians associated with the boundary conditions atx=0 andx=1 are the same MaxwellianM _w , then the solution is uniformly bounded and tends toM _w fort.     

On the Landau-Teller approximation for energy exchanges with fast degrees of freedom

We revisit the Landau-Teller heuristic approach to adiabatic invariants and, following Rapp, use it to investigate the energy exchanges between the different degrees of freedom, in simple Hamiltonian systems describing the collision of fast rotating or vibrating molecules with a fixed wall. We critically compare the theoretical results with particularly accurate numerical computations (quite small energy exchanges, namely of one part over 10³⁰, are measured).     

Evaluation of the limit of performance of an analytical method based on a statistical calculation of its critical concentrations according to ISO standard 11843: Application to routine control of banned veterinary drug residues in food according to European Decision 657/2002/EC

Traceability of the measurement of analytical parameters capable of evaluating the performance of methods is an important concept for the assessment of quality for the routine control, especially for residue monitoring of non-authorized medicinal substances in food from animal origin. The European Decision no. 657/2002/EC recommends to calculate two statistical limits, CCα and CCβ, which allow to evaluate the critical concentrations above which the method reliably distinguish and quantify a substance taking into account the variability of the method and the statistical risk to take a wrong decision. The calculation, which can be derived from the ISO standard no. 11843 is applied on a routine basis. An example displays a very simple way for evaluating the performance of an LC-MSMS method which has been validated a few years ago and is qualified onto a Micromass Quattro LCZ tandem mass spectrometer to monitor and confirm the nitrofuran metabolite residues in food from animal origin. Community Reference Laboratory for Antimicrobial Veterinary Drug Residue Control in Food from Animal Origin     

On some variational approximations in two-dimensional classical lattice systems

A new derivation is presented of some variational approximations for classical lattice systems that belong to the class of cluster-variation methods, among them the well-known Bethe-Peierls and Kramers-Wannier approximations. The limiting behavior of a hierarchical sequence of cluster-variation approximations, the so-calledC hierarchy, is discussed. It is shown that this hierarchy provides a monotonically decreasing sequence of upper boundsf _n on the free energy per lattice sitef and thatf _n f asn . Our results are based on extension theorems for states given on subsets of the lattice, which might be of some independent interest, and on an application of transfer matrix concepts to the variational characterization of translation-invariant equilibrium states.     

A Robust and Efficient Method for the Computation of Equilibrium Composition in Gaseous Mixtures

This paper presents in detail a robust, efficient and accurate methodology for the computation of equilibrium composition in gaseous mixtures. The methodology is founded on the concept of the chemical basis, which is defined and formalized using a powerful matricial approach. The method is specially designed to be general, thus providing basic thermodynamic data in several areas, such as combustion, plasma chemistry and, more generally speaking, computational fluid dynamics. The performance of the method is given in terms of CPU usage and the computed results are compared with those in the published literature. The method is shown to yield results of very high quality in terms of accuracy and smoothness.     

LiCl—HOAc和LiCl—HOAc—H2O体系（25℃）相图

测定了LiCl-HOAc和LiCl-HOAc-H_2O体系25℃时的溶度,绘制了溶度图.在LiCl-HOAc体系的相图中有LiCl及LiCl·1/2HOAc2种固相,后者是固液异成分化合物,体系中未发现LiCl·5HOAc或LiCL·HOAc的生成.LiCl-HOAc-H_2O体系25℃时的相图中有3种固相,即LiCl·H_2O、LiCl和LiCl·(1/2)HOAc.相变LiCl·H_2O→LiCl→LiCl·(1/2)HOAc表明HOAc在该三元体系中具有盐溶、脱水和溶剂合3种作用.此结果提供了一种制备无水LiCl的新方法.     

Vapor pressure of R227ea + ethanol at 343.13 K by molecular simulation

A new molecular model for 1,1,1,2,3,3,3-heptafluoropropane (R227ea) was developed on the basis of quantum chemical calculations and optimized using experimental vapor pressure and bubble density data. In combination with an existing model for ethanol, a molecular model for the binary mixture R227ea + ethanol was defined, using the Lorentz–Berthelot combining rule. It was validated at 283.17 K, where, considering the statistical uncertainties, it agrees to the experimental vapor pressure. The vapor–liquid equilibrium, comprising both bubble line and dew line data, was predicted at 343.13 K by molecular simulation. The Peng–Robinson equation of state fails for this system.     

Determining PPARγ-ligand binding affinity using fluorescent assay with cis-parinaric acid as a probe

Upon the study of small-molecules binding to proteins, the traditional methods for calculating dissociation constants (Kd and Ki) have shortcomings in dealing with the single binding site models. In this paper, two equations have been derived to solve this problem. These two equations are independent of the total concentration or initial degree of saturation of receptor and the activity of the competitive molecule. Through nonlinear fitting against these two equations, Kd value of a probe can be obtained by binding assay, and Ki value of a ligand can be obtained by competitive assay. Moreover, only the total concentrations of receptor([R]t), ligand([L]t) and probe([P]t) are required for the data fitting. In this work, Ki values of some typical ligands of PPARγ were successfully determined by use of our equations, among which the Ki value of PPARγ-LY171883 was reported for the first time.