Model selection approaches for non-linear system identification: a review

The identification of non-linear systems using only observed finite datasets has become a mature research area over the last two decades. A class of linear-in-the-parameter models with universal approximation capabilities have been intensively studied and widely used due to the availability of many linear-learning algorithms and their inherent convergence conditions. This article presents a systematic overview of basic research on model selection approaches for linear-in-the-parameter models. One of the fundamental problems in non-linear system identification is to find the minimal model with the best model generalisation performance from observational data only. The important concepts in achieving good model generalisation used in various non-linear system-identification algorithms are first reviewed, including Bayesian parameter regularisation and models selective criteria based on the cross validation and experimental design. A significant advance in machine learning has been the development of the support vector machine as a means for identifying kernel models based on the structural risk minimisation principle. The developments on the convex optimisation-based model construction algorithms including the support vector regression algorithms are outlined. Input selection algorithms and on-line system identification algorithms are also included in this review. Finally, some industrial applications of non-linear models are discussed.

The Association Between Childhood Trauma and Memory Functioning in Schizophrenia

Objective: Both neurocognitive impairments and a history of childhood abuse are highly prevalent in patients with schizophrenia. Childhood trauma has been associated with memory impairment as well as hippocampal volume reduction in adult survivors. The aim of the following study was to examine the contribution of childhood adversity to verbal memory functioning in people with schizophrenia. Methods: Eighty-five outpatients with a Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) diagnosis of chronic schizophrenia were separated into 2 groups on the basis of self-reports of childhood trauma. Performance on measures of episodic narrative memory, list learning, and working memory was then compared using multivariate analysis of covariance. Results: Thirty-eight (45%) participants reported moderate to severe levels of childhood adversity, while 47 (55%) reported no or low levels of childhood adversity. After controlling for premorbid IQ and current depressive symptoms, the childhood trauma group had significantly poorer working memory and episodic narrative memory. However, list learning was similar between groups. Conclusion: Childhood trauma is an important variable that can contribute to specific ongoing memory impairments in schizophrenia.

Structure of warm dense matter via angularly resolved x-ray scatter

In this paper we describe experimental results on angularly resolved x-ray scatter from a sample of warm dense aluminium that has been created by double sided laser-driven shock compression. The experiment was carried out on the Central Laser Facility of the Rutherford Appleton Laboratory, using the VULCAN laser. The form of the angularly resolved scatter cross-section was compared with predictions based on a series of molecular dynamics simulations with an embedded atom potential, a Yukakwa potential and a bare Coulomb potential. The importance of screening is evident from the comparison and the embedded atom model seems to match experiment better than the Yukawa potential.

Diagnostic-based modeling on a micro-scale atmospheric-pressure plasma jet

Diagnostic-based modeling (DBM) actively combines complementary advantages of numerical plasma simulations and relatively simple optical emission spectroscopy (OES). DBM is applied to determine spatial absolute atomic oxygen ground-state density profiles in a micro atmospheric-pressure plasma jet operated in He–O2. A 1D fluid model with semi-kinetic treatment of the electrons yields detailed information on the electron dynamics and the corresponding spatio-temporal electron energy distribution function. Benchmarking this time- and space-resolved simulation with phase-resolved OES (PROES) allows subsequent derivation of effective excitation rates as the basis for DBM. The population dynamics of the upper O(3p3P) oxygen state (? = 844 nm) is governed by direct electron impact excitation, dissociative excitation, radiation losses, and collisional induced quenching. Absolute values for atomic oxygen densities are obtained through tracer comparison with the upper Ar(2p1) state (? = 750.4 nm). The resulting spatial profile for the absolute atomic oxygen density shows an excellent quantitative agreement to a density profile obtained by two-photon absorption laser-induced fluorescence spectroscopy.

The influence of the relative phase between the driving voltages on electron heating in asymmetric dual frequency capacitive discharges

The influence of the relative phase between the driving voltages on electron heating in asymmetric phase-locked dual frequency capacitively coupled radio frequency plasmas operated at 2 and 14 MHz is investigated. The basis of the analysis is a nonlinear global model with the option to implement a relative phase between the two driving voltages. In recent publications it has been reported that nonlinear electron resonance heating can drastically enhance the power dissipation to electrons at moments of sheath collapse due to the self-excitation of nonlinear plasma series resonance (PSR) oscillations of the radio frequency current. This work shows that depending on the relative phase of the driving voltages, the total number and exact moments of sheath collapse can be influenced. In the case of two consecutive sheath collapses a substantial increase in dissipated power compared with the known increase due to a single PSR excitation event per period is observed. Phase resolved optical emission spectroscopy (PROES) provides access to the excitation dynamics in front of the driven electrode. Via PROES the propagation of beam-like energetic electrons immediately after the sheath collapse is observed. In this work we demonstrate that there is a close relation between moments of sheath collapse, and thus excitation of the PSR, and beam-like electron propagation. A comparison of simulation results to experiments in a single and dual frequency discharge shows good agreement. In particular the observed influence of the relative phase on the dynamics of a dual frequency discharge is described by means of the presented model. Additionally, the analysis demonstrates that the observed gain in dissipation is not accompanied by an increase in the electrode’s dc-bias voltage which directly addresses the issue of separate control of ion flux and ion energy in dual frequency capacitively coupled radio frequency plasmas.

A photospheric bright point model

Aims. A magneto-hydrostatic model is constructed with spectropolarimetric properties close to those of solar photospheric magnetic bright points.
Methods. Results of solar radiative magneto-convection simulations are used to produce the spatial structure of the vertical component of the magnetic field. The horizontal component of magnetic field is reconstructed using the self-similarity condition, while the magneto-hydrostatic equilibrium condition is applied to the standard photospheric model with the magnetic field embedded. Partial ionisation processes are found to be necessary for reconstructing the correct temperature structure of the model.
Results. The structures obtained are in good agreement with observational data. By combining the realistic structure of the magnetic field with the temperature structure of the quiet solar photosphere, the continuum formation level above the equipartition layer can be found. Preliminary results are shown of wave propagation through this magnetic structure. The observational consequences of the oscillations are examined in continuum intensity and in the Fe I 6302 angstrom magnetically sensitive line.

Panel Stationarity Test with Structural Breaks

In this paper, we extend the heterogeneous panel data stationarity test of Hadri [Econometrics Journal, Vol. 3 (2000) pp. 148–161] to the cases where breaks are taken into account. Four models with different patterns of breaks under the null hypothesis are specified. Two of the models have been already proposed by Carrion-i-Silvestre et al.[Econometrics Journal,Vol. 8 (2005) pp. 159–175]. The moments of the statistics corresponding to the four models are derived in closed form via characteristic functions.We also provide the exact moments of a modified statistic that do not asymptotically depend on the location of the break point under the null hypothesis. The cases where the break point is unknown are also considered. For the model with breaks in the level and no time trend and for the model with breaks in the level and in the time trend, Carrion-i-Silvestre et al. [Econometrics Journal, Vol. 8 (2005) pp. 159–175]showed that the number of breaks and their positions may be allowed to differ acrossindividuals for cases with known and unknown breaks. Their results can easily be extended to the proposed modified statistic. The asymptotic distributions of all the statistics proposed are derived under the null hypothesis and are shown to be normally distributed. We show by simulations that our suggested tests have in general good performance in finite samples except the modified test. In an empirical application to the consumer prices of 22 OECD countries during the period from 1953 to 2003, we found evidence of stationarity once a structural break and cross-sectional dependence are accommodated.

The top-down mechanism for body-mass-abundance scaling

Scaling relationships between mean body masses and abundances of species in multitrophic communities continue to be a subject of intense research and debate. The top-down mechanism explored in this paper explains the frequently observed inverse linear relationship between body mass and abundance (i.e., constant biomass) in terms of a balancing of resource biomasses by behaviorally and evolutionarily adapting foragers, and the evolutionary response of resources to this foraging pressure. The mechanism is tested using an allometric, multitrophic community model with a complex food web structure. It is a statistical model describing the evolutionary and population dynamics of tens to hundreds of species in a uniform way. Particularities of the model are the detailed representation of the evolution and interaction of trophic traits to reproduce topological food web patterns, prey switching behavior modeled after experimental observations, and the evolutionary adaptation of attack rates. Model structure and design are discussed. For model states comparable to natural communities, we find that (1) the body-mass-abundance scaling does not depend on the allometric scaling exponent of physiological rates in the form expected from the energetic equivalence rule or other bottom-up theories; (2) the scaling exponent of abundance as a function of body mass is approximately -1, independent of the allometric exponent for physiological rates assumed; (3) removal of top-down control destroys this pattern, and energetic equivalence is recovered. We conclude that the top-down mechanism is active in the model, and that it is a viable alternative to bottom-up mechanisms for controlling body-mass-abundance relations in natural communities.

TESTING FOR STATIONARITY IN HETEROGENEOUS PANEL DATA IN THE CASE OF MODEL MISSPECIFICATION

This paper investigates the performance of the tests proposed by Hadri and by Hadri and Larsson for testing for stationarity in heterogeneous panel data under model misspecification. The panel tests are based on the well known KPSS test (cf. Kwiatkowski et al.) which considers two models: stationarity around a deterministic level and stationarity around a deterministic trend. There is no study, as far as we know, on the statistical properties of the test when the wrong model is used. We also consider the case of the simultaneous presence of the two types of models in a panel. We employ two asymptotics: joint asymptotic, T, N -> infinity simultaneously, and T fixed and N allowed to grow indefinitely. We use Monte Carlo experiments to investigate the effects of misspecification in sample sizes usually used in practice. The results indicate that the assumption that T is fixed rather than asymptotic leads to tests that have less size distortions, particularly for relatively small T with large N panels (micro-panels) than the tests derived under the joint asymptotics. We also find that choosing a deterministic trend when a deterministic level is true does not significantly affect the properties of the test. But, choosing a deterministic level when a deterministic trend is true leads to extreme over-rejections. Therefore, when unsure about which model has generated the data, it is suggested to use the model with a trend. We also propose a new statistic for testing for stationarity in mixed panel data where the mixture is known. The performance of this new test is very good for both cases of T asymptotic and T fixed. The statistic for T asymptotic is slightly undersized when T is very small (

Simulation of a collisionless planar electrostatic shock in a proton-electron plasma with a strong initial thermal pressure change

The localized deposition of the energy of a laser pulse, as it ablates a solid target, introduces high thermal pressure gradients in the plasma. The thermal expansion of this laser-heated plasma into the ambient medium (ionized residual gas) triggers the formation of non-linear structures in the collisionless plasma. Here an electron-proton plasma is modelled with a particle-in-cell simulation to reproduce aspects of this plasma expansion. A jump is introduced in the thermal pressure of the plasma, across which the otherwise spatially uniform temperature and density change by a factor of 100. The electrons from the hot plasma expand into the cold one and the charge imbalance drags a beam of cold electrons into the hot plasma. This double layer reduces the electron temperature gradient. The presence of the low-pressure plasma modifies the proton dynamics compared with the plasma expansion into a vacuum. The jump in the thermal pressure develops into a primary shock. The fast protons, which move from the hot into the cold plasma in the form of a beam, give rise to the formation of phase space holes in the electron and proton distributions. The proton phase space holes develop into a secondary shock that thermalizes the beam.

Scaling in the time-dependent failure of a fiber bundle with local load sharing

We study the scaling behaviors of a time-dependent fiber-bundle model with local load sharing. Upon approaching the complete failure of the bundle, the breaking rate of fibers diverges according to r(t)proportional to(T-f-t)(-xi) where T-f is the lifetime of the bundle and xi approximate to 1.0 is a universal scaling exponent. The average lifetime of the bundle [T-f] scales with the system size as N-delta, where delta depends on the distribution of individual fiber as well as the breakdown rule. [S1063-651X(99)13902-3].

Complex fiber bundle model for optimization of heterogeneous materials

We propose a complex fiber bundle model for the optimization of heterogeneous materials, which consists of many simple bundles. We also present an exact and compact recursion relation for the failure probability of a simple fiber bundle model with local load sharing, which is more efficient than the ones reported previously. Using a ''renormalization method'' and the recursion relation developed for the simple bundle, we calculate the failure probabilities of the complex fiber bundle. When the total number of fibers is given, we find that there exists an optimum way to organize the complex bundle, in which one gets a stronger bundle than in other ways.

Predictions of large events on a spring-block model

We study the predictability of a theoretical model for earthquakes, using a pattern recognition algorithm similar to the CN and M8 algorithms known in seismology. The model, which is a stochastic spring-block model with both global correlation and local interaction, becomes more predictable as the strength of the global correlation or the local interaction is increased.

BURST-SIZE DISTRIBUTION IN FIBER-BUNDLES WITH LOCAL LOAD-SHARING

A critical load x(c) is introduced into the fiber-bundle model with local load-sharing. The critical load is defined as the average load per fiber that causes the final complete failure. It is shown that x(c) --> 0 when the size of the system N --> infinity. A power law for the burst-size distribution, D(DELTA) is-proportional-to DELTA(-xi) is approximately correct. The exponent xi is not universal, since it depends on the strength distribution as well as the size of the system.

Physical Models of Wind Instruments : A Generalized Excitation Coupled with a Modular Tube Simulation Platform

To develop real-time simulations of wind instruments, digital waveguides filters can be used as an efficient representation of the air column. Many aerophones are shaped as horns which can be approximated using conical sections. Therefore the derivation of conical waveguide filters is of special interest. When these filters are used in combination with a generalized reed excitation, several classes of wind instruments can be simulated. In this paper we present the methods for transforming a continuous description of conical tube segments to a discrete filter representation. The coupling of the reed model with the conical waveguide and a simplified model of the termination at the open end are described in the same way. It turns out that the complete lossless conical waveguide requires only one type of filter.Furthermore, we developed a digital reed excitation model, which is purely based on numerical integration methods, i.e., without the use of a look-up table.