arXiv subjects
Publications and source records attributed to Michael Goldstein.
We characterize spectra of Schr\"odinger operators with small quasiperiodic analytic potentials in terms of their comb domains, and study action variables motivated by the KdV integrable system.
The study of transmission and progression of human papillomavirus (HPV) is crucial for understanding the incidence of cervical cancers, and has been identified as a priority worldwide. The complexity of the disease necessitates a detailed model of HPV transmission and its progression to cancer; to infer properties of the above we require a careful process that can match to imperfect or incomplete observational data. In this paper, we describe the HPVsim simulator to satisfy the former requirement; to satisfy the latter we couple this stochastic simulator to a process of emulation and history matching using the R package hmer. With these tools, we are able to obtain a comprehensive collection of parameter combinations that could give rise to observed cancer data, and explore the implications of the variability of these parameter sets as it relates to future health interventions.
We prove that GOE random band matrix localization length is $\le C\left(\log W\right)^3 W^2$, where $W$ is the width of the band and $C$ is an absolute constant. Our method consists of Green function edge-to-edge vector action approach to the Schenker method. That allows to split and decouple the action, so that it becomes transparent that \emph{the magnitudes of two consecutive Schur complements vector actions can not be both larger than an absolute constant}. That is the central technological ingedient of the method. It comes from rather involved estimates $($ the main estimates of the metod $)$, in combination with an equation relating two magnitudes in question. We call the latter \emph{recurrence equation}. The method results in the \emph{lower bound of the variance of the $\log$--norm of the vector action at $\gtrsim NW^{-1}$}, where $N$ is the total number of GOE blocks, condition $N\lesssim W^D$ with an absolute constant $D\gg 1$ applies.
Modelling complex real-world situations such as infectious diseases, geological phenomena, and biological processes can present a dilemma: the computer model (referred to as a simulator) needs to be complex enough to capture the dynamics of the system, but each increase in complexity increases the evaluation time of such a simulation, making it difficult to obtain an informative description of parameter choices that would be consistent with observed reality. While methods for identifying acceptable matches to real-world observations exist, for example optimisation or Markov chain Monte Carlo methods, they may result in non-robust inferences or may be infeasible for computationally intensive simulators. The techniques of emulation and history matching can make such determinations feasible, efficiently identifying regions of parameter space that produce acceptable matches to data while also providing valuable information about the simulator's structure, but the mathematical considerations required to perform emulation can present a barrier for makers and users of such simulators compared to other methods. The hmer package provides an accessible framework for using history matching and emulation on simulator data, leveraging the computational efficiency of the approach while enabling users to easily match to, visualise, and robustly predict from their complex simulators.
Organisations, whether in government, industry or commerce, are required to make decisions in a complex and uncertain environment. The way models are used is intimately connected to the way organisations make decisions and the context in which they make them. Typically, in a complex organisation, multiple related models will often be used in support of a decision. For example, engineering models might be combined with financial models and macro-economic models in order to decide whether to invest in new production capability. Different parts of a complex organisation might operate their own related models which might then be presented to a central decision maker. Yet in practice, there is little awareness of the practical challenges of using models in a robust way to support decision making. There is significant scope to improve decision making though an enhanced understanding of the role and limitations of modelling and through the application of cutting edge methodologies and organisational best practice. This report is in the form of a 'white paper', whose purpose is to identify key issues for consideration whist postulating tentative approaches to these issues that might be worthy of further exploration, focussing on both technical and organisational aspects. It begins with a framework for consideration of how model-based decisions are made in organisations. It then looks more closely at the questions of uncertainty and multiple models. It then postulates some technical statistical and organisational approaches for managing some of these issues. Finally, it considers the way forward, and the possible focus for further work.
We explore the effect of finite population sampling in design problems with many variables cross-classified in many ways. In particular, we investigate designs where we wish to sample individuals belonging to different groups for which the underlying covariance matrices are separable between groups and variables. We exploit the generalised conditional independence structure of the model to show how the analysis of the full model can be reduced to an interpretable series of lower dimensional problems. The types of information we gain by sampling are identified with the orthogonal canonical directions. We first solve a variable problem, which utilises the powerful properties of the adjustment of second-order exchangeable vectors, which has the same qualitative features, represented by the underlying canonical variable directions, irrespective of chosen group, population size or sample size. We then solve a series of group problems which in a balanced design reduce to the sampling of second-order exchangeable vectors. If the population sizes are finite then the qualitative and quantitative features of each group problem will depend upon the sampling fractions in each group, mimicking the infinite problem when the sampling fractions in each group are the same.
We study multi-frequency quasiperiodic Schrödinger operators on $\mathbb{Z} $. We prove that for a large real analytic potential satisfying certain restrictions the spectrum consists of a single interval. The result is a consequence of a criterion for the spectrum to contain an interval at a given location that we establish non-perturbatively in the regime of positive Lyapunov exponent.
We study multi-frequency quasi-periodic Schrödinger operators on $\mathbb{Z}$ in the regime of positive Lyapunov exponent and for general analytic potentials. Combining Bourgain's semi-algebraic elimination of multiple resonances with the method of elimination of double resonances via resultants, we establish exponential finite-volume localization as well as the separation between the eigenvalues. In a follow-up paper we develop the method further to show that for potentials given by large generic trigonometric polynomials the spectrum consists of a single interval, as conjectured by Chulaevski and Sinai.
We prove a version of the doubling Bernstein inequalities for the trace of an analytic function of two variables on an analytic subset of $\mathbb{C}^2$. The estimate applies to the whole analytic set in question including its singular points. The proof relies on a version of the Cartan estimate for maps in $\mathbb{C}^2$ which we establish in this work.
Background: Many mathematical models have now been employed across every area of systems biology. These models increasingly involve large numbers of unknown parameters, have complex structure which can result in substantial evaluation time relative to the needs of the analysis, and need to be compared to observed data. The correct analysis of such models usually requires a global parameter search, over a high dimensional parameter space, that incorporates and respects the most important sources of uncertainty. This can be an extremely difficult task, but it is essential for any meaningful inference or prediction to be made about any biological system. It hence represents a fundamental challenge for the whole of systems biology. Results: Bayesian statistical methodology for the uncertainty analysis of complex models is introduced, which is designed to address the high dimensional global parameter search problem. Bayesian emulators that mimic the systems biology model but which are extremely fast to evaluate are embedded within an iterative history match: an efficient method to search high dimensional spaces within a more formal statistical setting, while incorporating major sources of uncertainty. The approach is demonstrated via application to two models of hormonal crosstalk in Arabidopsis root development, which have 32 rate parameters, for which we identify the sets of rate parameter values that lead to acceptable matches to observed trend data. The biological consequences of the resulting comparison, including the evaluation of gene functions, are described.
In this paper, we are concerned with attributing meaning to the results of a Bayesian analysis for a problem which is sufficiently complex that we are unable to assert a precise correspondence between the expert probabilistic judgements of the analyst and the particular forms chosen for the prior specification and the likelihood for the analysis. In order to do this, we propose performing a finite collection of additional Bayesian analyses under alternative collections of prior and likelihood modelling judgements that we may also view as representative of our prior knowledge and the problem structure, and use these to compute posterior belief assessments for key quantities of interest. We show that these assessments are closer to our true underlying beliefs than the original Bayesian analysis and use the temporal sure preference principle to establish a probabilistic relationship between our true posterior judgements, our posterior belief assessment and our original Bayesian analysis to make this precise. We exploit second order exchangeability in order to generalise our approach to situations where there are infinitely many alternative Bayesian analyses we might consider as informative for our true judgements so that the method remains tractable even in these cases. We argue that posterior belief assessment is a tractable and powerful alternative to robust Bayesian analysis. We describe a methodology for computing posterior belief assessments in even the most complex of statistical models and illustrate with an example of calibrating an expensive ocean model in order to quantify uncertainty about global mean temperature in the real ocean.
We study the Cauchy problem for the KdV equation $\partial_t u - 6 u \partial_x u + \partial_x^3 u = 0$ with almost periodic initial data $u(x,0)=V(x)$. We consider initial data $V$, for which the associated Schr\"odinger operator is absolutely continuous and has a spectrum that is not too thin in a sense we specify, and show the existence, uniqueness, and almost periodicity in time of solutions. This establishes a conjecture of Percy Deift for this class of initial data. The result is shown to apply to all small analytic quasiperiodic initial data with Diophantine frequency vector.
We consider the KdV equation $$ \partial_t u +\partial^3_x u +u\partial_x u=0 $$ with quasi-periodic initial data whose Fourier coefficients decay exponentially and prove existence and uniqueness, in the class of functions which have an expansion with exponentially decaying Fourier coefficients, of a solution on a small interval of time, the length of which depends on the given data and the frequency vector involved. For a Diophantine frequency vector and for small quasi-periodic data (i.e., when the Fourier coefficients obey $|c(m)| \le \varepsilon \exp(-κ_0 |m|)$ with $\varepsilon > 0$ sufficiently small, depending on $κ_0 > 0$ and the frequency vector), we prove global existence and uniqueness of the solution. The latter result relies on our recent work \cite{DG} on the inverse spectral problem for the quasi-periodic Schrödinger equation.
We consider the one-dimensional discrete Schr\"odinger operator $$ \bigl[H(x,\omega)\varphi\bigr](n)\equiv -\varphi(n-1)-\varphi(n+1) + V(x + n\omega)\varphi(n)\ , $$ $n \in \mathbb{Z}$, $x,\omega \in [0, 1]$ with real-analytic potential $V(x)$. Assume $L(E,\omega)>0$ for all $E$. Let $\mathcal{S}_\omega$ be the spectrum of $H(x,\omega)$. For all $\omega$ obeying the Diophantine condition $\omega \in \mathbb{T}_{c,a}$, we show the following: if $\mathcal{S}_\omega \cap (E',E")\neq \emptyset$, then $\mathcal{S}_\omega \cap (E',E")$ is homogeneous in the sense of Carleson (see [Car83]). Furthermore, we prove, that if $G_i$, $i=1,2$ are two gaps with $1 > |G_1| \ge |G_2|$, then $|G_2|\lesssim \exp\left(-(\log \mathrm{dist} (G_1,G_2))^A\right)$, $A\gg 1$. Moreover, the same estimates hold for the gaps in the spectrum on a finite interval, that is, for $\mathcal S_{N,\omega}:=\cup_{x\in\mathbb T}\mathrm{spec} H_{[-N,N]}(x,\omega) $, $N \ge 1 $, where $H_{[-N, N]}(x, \omega)$ is the Schr\"odinger operator restricted to the interval $[-N,N]$ with Dirichlet boundary conditions. In particular, all these results hold for the almost Mathieu operator with $|\lambda| \neq 1$. For the supercritical almost Mathieu operator, we combine the methods of [GolSch08] with Jitomirskaya's approach from [Jit99] to establish most of the results from [GolSch08] with $\omega$ obeying a strong Diophantine condition.
We study the quasi-periodic Schrödinger equation $$ -ψ"(x) + V(x) ψ(x) = E ψ(x), \qquad x \in \IR $$ in the regime of "small" $V$. Let $(E_m',E"_m)$, $m \in \zv$, be the standard labeled gaps in the spectrum. Our main result says that if $E"_m - E'_m \le \ve \exp(-κ_0 |m|)$ for all $m \in \zv$, with $\ve$ being small enough, depending on $κ_0 > 0$ and the frequency vector involved, then the Fourier coefficients of $V$ obey $|c(m)| \le \ve^{1/2} \exp(-\frac{κ_0}{2} |m|)$ for all $m \in \zv$. On the other hand we prove that if $|c(m)| \le \ve \exp(-κ_0 |m|)$ with $\ve$ being small enough, depending on $κ_0 > 0$ and the frequency vector involved, then $E"_m - E'_m \le 2 \ve \exp(-\frac{κ_0}{2} |m|)$.
We study the quasi-periodic Schr\"odinger operator $$ -\psi"(x) + V(x) \psi(x) = E \psi(x), \qquad x \in \mathbb{R} $$ in the regime of "small" $V(x) = \sum_{m\in\mathbb{Z}^\nu}c(m)\exp (2\pi i m\omega x)$, $\omega = (\omega_1, \dots, \omega_\nu) \in \mathbb{R}^\nu$, $|c(m)| \le \varepsilon \exp(-\kappa_0|m|)$. We show that the set of reflectionless potentials isospectral with $V$ is homeomorphic to a torus. Moreover, we prove that any reflectionless potential $Q$ isospectral with $V$ has the form $Q (x) = \sum_{m \in \mathbb{Z}^\nu} d(m) \exp (2\pi i m\omega x)$, with the same $\omega$ and with $|d(m)| \le \sqrt{2 \varepsilon} \exp(-\frac{\kappa_0}{2} |m|)$. Our derivation relies on the study of the approximation via Hill operators with potentials $\tilde V (x) = \sum_{m \in \mathbb{Z}^\nu} c(m) \exp (2 \pi i m \tilde \omega x)$, where $\tilde \omega$ is a rational approximation of $\omega$. It turns out that the multi-scale analysis method of \cite{DG} applies to these Hill operators. Namely, in \cite{DGL} we developed the multi-scale analysis for the operators dual to the Hill operators in question. The main estimates obtained in \cite{DGL} allow us here to establish the estimates for the gap lengths and the Fourier coefficients in a form which is considerably stronger than the estimates known in the theory of Hill operators with analytic potentials in the general setting. Due to these estimates, the approximation procedure for the quasi-periodic potentials is effective, despite the fact that the rate of approximation $|\omega - \tilde \omega| \thicksim \tilde T^{-\delta}$, $0 < \delta < 1/2$ is slow, on the scale of the period $\tilde T$ of the Hill operator.
We present an abstract multiscale analysis scheme for matrix functions $(H_{\varepsilon}(m,n))_{m,n\in \mathfrak{T}}$, where $\mathfrak{T}$ is an Abelian group equipped with a distance $|\cdot|$. This is an extension of the scheme developed by Damanik and Goldstein for the special case $\mathfrak{T} = \mathbb{Z}^\nu$. Our main motivation for working out this extension comes from an application to matrix functions which are dual to certain Hill operators. These operators take the form $H_{\tilde\omega}=-\frac{d^2}{dx^2} + \varepsilon U(\tilde\omega x)$, where $U$ is a real smooth function on the torus $\mathbb{T}^\nu$, $\tilde\omega\in \mathbb{R}^\nu$ is a vector with rational components, and $\varepsilon$ is a small parameter. The group in this particular case is the quotient $\mathfrak{T} = \mathbb{Z}^\nu/\{m\in\mathbb{Z}^\nu:m\tilde\omega=0\}$. We show that the general theory indeed applies to this special case, provided that the rational frequency vector $\tilde\omega$ obeys a suitable Diophantine condition in a large box of modes. Despite the fact that in this setting the orbits $k + m\omega$, $k \in \mathbb{R}$, $m\in\mathbb{Z}^\nu$ are not dense, the dual eigenfunctions are exponentially localized and the eigenvalues of the operators can be described as $E(k+m\omega)$ with $E(k)$ being a "nice" monotonic function of the impulse $k \ge 0$. This enables us to derive a description of the Floquet solutions and the band-gap structure of the spectrum, which we will use in a companion paper to develop a complete inverse spectral theory for the Sturm-Liouville equation with small quasi-periodic potential via periodic approximation of the frequency. The analysis of the gaps in the range of the function $E(k)$ plays a crucial role in this approach.
We consider the quasi-periodic Schr\"odinger operator $$ [H \psi](x) = -\psi"(x) + V(x) \psi(x) $$ in $L^2(\mathbb{R})$, where the potential is given by $$ V(x) = \sum_{m \in \mathbb{Z}^\nu \setminus \{ 0 \}} c(m)\exp (2\pi i m \omega x) $$ with a Diophantine frequency vector $\omega = (\omega_1, \dots, \omega_\nu) \in \mathbb{R}^\nu$ and exponentially decaying Fourier coefficients $|c(m)| \le \varepsilon \exp(-\kappa_0|m|)$. In the regime of small $\varepsilon > 0$ we show that the spectrum of the operator $H$ is homogeneous in the sense of Carleson.