Spectral Galerkin methods for transfer operators in uniformly expanding dynamics (Q1740642)

Markov uniformly expanding maps are an important class of chaotic dynamical systems. In this paper, the author develops a spectral Galerkin method for the transfer operators of these maps. More precisely, he provides estimates of some statistical quantities using finite submatrices of the infinite Fourier or Chebyshev basis coefficient matrices of the transfer operators. The transfer operator is given as \(\mathcal{L} : BV (\Lambda) \rightarrow BV (\Lambda)\) with action \[(\mathcal{L}\phi)(x)=\sum_{f(y)=x}\frac{1}{|f^{\prime}(y)|}\phi(y)\], where \(BV (\Lambda)\) denotes the space of functions of bounded variation on \(\Lambda\). The author mentions that transfer operator problems cannot in general be solved analytically, and numerical approaches are of major importance. Ulam's method scheme is widely used in the literature, whereby one projects the transfer operator onto a subspace of characteristic functions and compute statistical properties. The author considers the Fourier exponential basis \(e_{k}(x) = e^{ikx}\), \(k \in \mathbb{Z}\), which is orthogonal in \(L^{2}([0, 2\pi))\), and the Chebyshev polynomial basis \(T_{k}(x) = \cos k \cos^{-1} x\), \(k \in \mathbb{N}\), which is orthogonal on \([-1, 1]\) with respect to the weight \((1-x^{2})^{-1/2}\). The author firstly introduces the two generic classes of maps (circle maps \(U_{P}\) and interval maps \(U_{N P}\)) and then a set of so-called distortion conditions (i) \(DD_{r}\) for some \(r\in \mathbb{N}^{+}\) if \(\sup_{\iota \in I, x\in \Lambda}|\frac{v_{\iota}^{(n+1)}(x)}{v_{\iota}^{\prime}(x)}|=C_{n}<\infty\), \(n=1, \dots, r\); (ii) \(AD_{\delta}\) for some \(\delta>0\) if \(\sup_{z\in \Lambda_{\delta}^{\beta}}|\frac{v^{\prime\prime}(z)}{v^{\prime}(z)}|=C_{1, \delta}<\infty\), \(\Lambda=\mathbb{R}/2\pi \mathbb{Z}\), \(\sup_{\iota\in I, z\in \check{\Lambda}_{\delta}}|\frac{v^{\prime\prime}(z)}{v^{\prime}(z)}|=C_{1, \delta}<\infty\), \(\Lambda=[-1, 1]\) that maps from these classes may optionally hold. The author's main theoretical results are that spectral Galerkin estimates and the 1-resolvent converge exponentially fast in the approximation order \(N\) for analytic maps, and as \(O(N^{2.5-r})\) for \(C^{r}\) maps. The algorithmic outlay of his method is \(O(N^{3})\). The author sets up the problem in the beginning and then gives proof of results very rigorously. Such results characterise an operator that explicitly solves many typical transfer operator problems. The main theorem gives convergence of spectral operator estimates as \(\left\| \mathcal{L}_{N}\phi-\mathcal{L}\phi \right\|_{BV} < N\sqrt{N}K(N) \left\| \phi \right\|_{BV}\) and \(\left\| \mathcal{S}_{N}\phi-\mathcal{S}\phi\right\|_{BV} < N\sqrt{N}\bar{K}(N) \left\|\phi \right\|_{BV}\). Next the author provides bounding the magnitude of transfer operator spectral coefficient matrix entries \(L_{jk}\). To implement the results, the author describes two algorithms and use them to demonstrate the capacities of transfer operator spectral methods: a rigorously-validated algorithm, and a fast, more convenient adaptive algorithm. The rates of convergence obtained here are very good in comparison with the ones obtained with other approaches. While set-based approaches cover a much larger class of maps than the one that the author consider here, those have an optimal convergence rate of only \(O(\mathrm{log}(N)/N)\) irrespective of regularity, where \(N\) is the size of the Ulam matrix. Spectral methods are also significantly more efficient than algorithms that use periodic orbits to calculate statistical quantities: in the case of analytic maps, these algorithms converge super-exponentially in the order, but the number of periodic orbits that must be computed, and hence the computational cost, grows exponentially with the order. Numerical results indicate that the actual rates of convergence are slightly better than what the author proves in this paper, and a different theoretical approach may yield the optimal convergence rates. The author adaptive algorithms provides a much higher degree of accuracy for practical uses than previous algorithms, that is, produces double floating-point accuracy estimates in a fraction of a second on a personal computer. The author hopes that his spectral methods become a useful tool in theoretical and numerical study of chaotic systems. The author also demonstrates that adaptive spectral methods allow for very fast and user friendly computation of statistical properties, via an implementation of transfer operator spectral methods in the Julia package Poltergeist to estimate quantities under 0.1 s on a personal computer to 13 decimal places of accuracy. Finally, the author proves extremely accurate rigorous bounds on maps in \(U_{P}\) and \(U_{N P}\) satisfying sufficiently strong distortion conditions and presents two rigorously validated bounds on the Lyapunov exponent and a diffusion coefficient of the Lanford map, obtained via a rigorous implementation of his spectral method. The author considers the Lanford map \(f: [0, 1]\rightarrow [0, 1]\), \(f(x)=2x+\frac{1}{2}x(1-x)\) mod 1. (a) The Lanford map's Lyapunov exponent \(L_{exp}:=\int_{\Lambda}\log |f^{\prime}|\rho dx\) lies in the range \(L_{exp}=\)0.657 661 780 006 597 677 541 582 413 823 832 065 743 241 069 580 012 201 953 952 802 691 632 666 111 554 023 759 556 459 752 915 174 829 642 156 331 798 026 301 488 594 89 \(\pm 2 \times 10^{-128}\). (b) The diffusion coefficient for the Lanford map with observable \(\phi (x)=x^2\) lies in the range \(\sigma_{f}^{2}(\phi)=\) 0.360 109 486 199 160 672 898 824 186 828 576 749 241 669 997 797 228 864 358 977 865 838 174 403 103 617 477 981 402 783 211 083 646 769 039 410 848 031 999 960 664 7 \(\pm 6 \times 10^{-124}\). The numerical bounds the author obtains here are well beyond the practicable capabilities of other numerical results. The author uses a comparable amount of computational power of spectral methods to obtain 127 and 123 decimal places validated significant figures. The author says that his results may be extended to higher dimensions, possibly including maps with contracting directions. By constructing efficient numerical inducing schemes, it observers that his methods can provide fast and accurate estimates of statistical properties for almost all major classes of one dimensional chaotic maps, such as non-Markov expanding maps, intermittent maps and quadratic maps. At the end of the paper, the author provides the relationship between uniform expansion and uniform \(C\)-expansion, some results on distortion conditions, explicit bounds on the norm of the solution operator in \(BV\) and a very rigorous proof of lemma which shows that the relevant bounds on \(\nu_{\iota^{\prime}}\) and \(h_{\iota}\) hold uniformly for all \(\iota\).