Independent scholarship · Open accessThe monthly digest

Indexed text attachment

partial-differential-equations-in-higher-categories.tex

Source for Partial Differential Equations in Higher Categories

View paper · Download original file

Lines 3321–3420 of 3921. Source text is displayed literally.

3321\node[ink2] at (6.66,2.554) {\scriptsize second};3322\end{scope}3323\end{tikzpicture}3324\figcap{The bifurcation diagram of the buckling problem: the amplitude $u(0)$ against the load. The trivial solution is the only one up to $\lambda_1$ and persists, unstable, beyond it (dashed). The first pair of branches is $\lambda/\lambda_1=(\frac2\pi K(\sin\frac A2))^2$, with the three computed points marked; the second leaves at $4\lambda_1$. All branches bend to the right and approach $\pm\pi$, the rod folded back on itself.}3325\end{center}3326 3327\section{Water waves}3328 3329The motion of water under gravity is a free boundary problem: Laplace's equation for the velocity potential in a3330domain whose upper boundary is the unknown \src{\S14.5}. In the formulation of Zakharov and of Craig and Sulem the3331unknowns are the surface elevation and the potential on it, and the dynamics is driven by the3332Dirichlet-to-Neumann operator of \Cref{cor:dtn} for the moving domain \cite{CrSu93}. Linearizing at still water of3333depth $h$ gives the dispersion relation $\omega^2=gk\tanh kh$: waves travel at $\sqrt{gh}$ when they are long and at3334$\sqrt{g/k}$ when they are short. \chk{X10} The Korteweg--de Vries equation is the first correction to the3335long-wave limit, which is how the soliton was first seen on a canal.3336 3337\section*{Notes}3338 3339The source's Chapter 14 is followed for the examples. For conservation laws and entropy solutions see3340\cite{Lax73,Eva10}; for the Cole--Hopf transformation \cite{Hop50,Col51}; for inverse scattering3341\cite{GGKM67,Lax68}; for Noether's theorem and the variational bicomplex \cite{Olv93,And92}; for bifurcation from a3342simple eigenvalue \cite{CrRa71}; for the Hamiltonian formulation of water waves \cite{CrSu93}.3343 3344\section*{Exercises}3345 3346\begin{exercise}\label{exer:break}3347Find the breaking time for $u_t+uu_x=0$ with $\varphi=1/(1+x^2)$.3348\end{exercise}3349 3350\begin{exercise}\label{exer:rh}3351Verify the two shock speeds of \Cref{prop:consform}.3352\end{exercise}3353 3354\begin{exercise}\label{exer:ch}3355Prove the first statement of \Cref{thm:colehopf}.3356\end{exercise}3357 3358\begin{exercise}\label{exer:sol}3359Verify that $f=\tfrac c2\sech^2(\tfrac{\sqrt c}2\xi)$ satisfies $-cf+f''+3f^2=0$.3360\end{exercise}3361%% ---- end ch/14-nonlinear.tex ----3362%% ---- begin ch/15-coda.tex ----3363\chapter{Assessment}\label{ch:coda}3364 3365\section{What the language did}3366 3367Eight things, roughly in the order of how much they changed our reading of the source.3368 3369\emph{Boundary conditions became objects.} A boundary condition is a character of a reflection group when the3370domain is a chamber (\Cref{thm:images}) and a Lagrangian plane of boundary values in general3371(\Cref{thm:lagrangian}). The first explains when the method of images works and extends it to diffusion, where3372the source says it fails; the second explains the source's ``symmetric'' conditions and puts Dirichlet, Neumann,3373Robin and periodic conditions in one three-dimensional family.3374 3375\emph{Two methods turned out to be one.} The sum over images and the sum over eigenfunctions are exchanged by3376Pontryagin duality (\Cref{thm:kernels}, \Cref{thm:fourier}), and the same exchange computes the heat trace3377(\Cref{prop:rect}). The source presents Fourier's method as the replacement for a method that does not3378generalize; they are dual.3379 3380\emph{The comparison of waves and diffusions was reduced to three lines.} Group or semigroup; symplectic form or3381positivity; and the six remaining rows of the source's table follow (\Cref{ch:diffusions}).3382 3383\emph{Stability of schemes was identified with structure.} For diffusion it is membership in the Markov category3384(\Cref{thm:explicit}); for waves it is respect for the causal structure (\Cref{thm:cfl}); for elliptic problems it is3385being a map of complexes (\Cref{thm:feec}).3386 3387\emph{Problems could be glued.} Dirichlet-to-Neumann maps, transfer matrices and boundary conditions compose as3388Lagrangian relations (\Cref{thm:gluing}, \Cref{thm:transfer}).3389 3390\emph{The type was located.} It is invisible to the algebra of formal solutions and lives in the real points of the3391characteristic variety (\Cref{rem:type}); it reappears as the singular support of the fundamental solution3392(\Cref{thm:fund}) and as the kind of functor that time is.3393 3394\emph{Solvable nonlinear equations were explained by morphisms.} Cole--Hopf and the scattering transform are maps3395of equations onto linear ones (\Cref{thm:colehopf}, \Cref{thm:ist}).3396 3397\emph{Isospectrality became an isomorphism of representations} that is not an isometry (\Cref{thm:sunada}).3398 3399\section{What it did not do}3400 3401It proved no theorem of analysis. Existence of solutions, in every chapter, rests on estimates that the language3402does not supply: the maximum principle, energy inequalities, compactness. The definition of well-posedness as an3403isomorphism in a category of topological vector spaces (\Cref{def:wellposed}) says where the difficulty is and3404does not reduce it. The machinery cited in \Cref{ch:equations}, the jet comonad and $\DD$-modules, was used only to3405say what a morphism of equations is. Microlocal analysis and hyperfunctions, where sheaf theory does real work3406for linear equations, are described and not used. The nonlinear chapter treats two integrable examples and one3407scalar conservation law; for systems of conservation laws, for the Euler and Navier--Stokes equations and for the3408full water wave problem the categorical viewpoint has, so far as we can see, nothing to offer beyond the remarks3409made. The observations tagged as such are short and may well be known in other forms.3410 3411\section{How much of this is higher}3412 3413\begin{center}\small3414\begin{tabular}{@{}>{\raggedright}p{0.2\textwidth}>{\raggedright}p{0.4\textwidth}>{\raggedright\arraybackslash}p{0.31\textwidth}@{}}3415\toprule3416Source & Structure used here & Level\\3417\midrule3418Ch.~1 & solution functor; symbol; quotient by characteristics & $1$-categorical\\3419Ch.~2--3 & group and Markov semigroup; reflection groups; causal functor & $1$-categorical\\3420Ch.~4--5 & Lagrangian planes; Pontryagin duality & $1$-categorical; relations compose\\