querent

Spam yourself with math PhD qualifying-exam questions.

Showing 261–270 of 816
Stanford Practice, 3 Applied mathematics, Hamilton-Jacobi equations, eikonal equation, viscosity solutions

Fix a bounded smooth function 𝑔(π‘₯). That is, both 𝑔(π‘₯) and all derivatives of 𝑔(π‘₯) are uniformly bounded on ℝ𝑛. Fix also a constant 𝑣>0. Consider the function

𝑒(𝑑,π‘₯)=max|π‘¦βˆ’π‘₯|≀𝑣𝑑𝑔(𝑦),

defined for 𝑑β‰₯0 and π‘₯βˆˆβ„π‘›.

(i) Show that 𝑒(𝑑,π‘₯) is differentiable in 𝑑 and π‘₯ for almost every (𝑑,π‘₯).

(ii) Suppose that 𝑒(𝑑,π‘₯) is continuously differentiable in a space-time neighborhood of a point (𝑑,π‘₯). Show that then

πœ•π‘’(𝑑,π‘₯)πœ•π‘‘=𝑣|βˆ‡π‘’(𝑑,π‘₯)|.

(iii) Give an example of a smooth function 𝑔(π‘₯) such that 𝑒(𝑑,π‘₯) is not differentiable in π‘₯ everywhere for some 𝑑>0.

(iv) Do there exist smooth functions 𝑔(π‘₯) such that 𝑒(𝑑,π‘₯) is differentiable in π‘₯ everywhere for all 𝑑>0?

Stanford Practice, 30 Applied mathematics, finite element method, CΓ©a's lemma, coercivity

Consider the PDE

βˆ’(π‘Ž(π‘₯)𝑒′(π‘₯))β€²+𝑐(π‘₯)𝑒(π‘₯)=𝑓(π‘₯),π‘₯∈(0,1), 𝑒′(0)=0,𝑒(1)=0,

with π‘Ž(π‘₯)β‰₯π‘Ž0>0 and 𝑐(π‘₯)β‰₯𝑐0>0 smooth functions on [0,1] and 𝑓(π‘₯)∈𝐿2.

(i) Write down the bilinear form of this PDE in the 𝐻01 space on (0,1).

(ii) Show that this bilinear form is bounded and coercive on 𝐻01.

(iii) Consider a piecewise linear finite element discretization with mesh size β„Ž. Denote the finite approximation space by π‘†β„Ž and the finite element solution by π‘’β„Ž. Show that there exist some positive constants 𝐢1,𝐢2 such that

β€–π‘’β„Žβˆ’π‘’β€–1≀𝐢1𝐢2minπœ’βˆˆπ‘†β„Žβ€–πœ’βˆ’π‘’β€–1,

where β€–β‹…β€–1 is the 𝐻01 norm.

Stanford Practice, 31 Applied mathematics, wave equation, symmetric hyperbolic systems, Friedrichs scheme, stability analysis

Consider the equation

𝑝𝑑𝑑=𝑝π‘₯π‘₯,π‘₯βˆˆβ„,𝑑>0

with initial condition 𝑝(π‘₯,0)=𝛼(π‘₯) and 𝑝𝑑(π‘₯,0)=𝛽(π‘₯).

(i) Write down an equivalent symmetric hyperbolic system. Hint: use 𝑝𝑑 and 𝑝π‘₯.

(ii) Write down the Friedrichs scheme for the system obtained in (i). Denote the time step by π‘˜, the spatial step by β„Ž, and the numerical approximation by π‘ˆπ‘—π‘›.

(iii) Assume that the finite difference method is defined for all π‘₯βˆˆβ„ and consider it as a map πΈπ‘˜,β„Ž from π‘ˆπ‘›(π‘₯) to π‘ˆπ‘›+1(π‘₯)=(πΈπ‘˜,β„Žπ‘ˆπ‘›)(π‘₯). Write down the Fourier multiplier πΈΜƒπ‘˜,β„Ž using the convention:

(πΈπ‘˜,β„Žπ‘ˆπ‘›)(π‘₯)=βˆ‘π‘π‘Žπ‘π‘ˆπ‘›(π‘₯βˆ’β„Žπ‘)β‡’πΈΜƒπ‘˜,β„Ž(πœ‰)=βˆ‘π‘π‘Žπ‘π‘’βˆ’2πœ‹π‘–(π‘β„Ž)πœ‰.

(iv) Find the right condition on β„Ž and π‘˜ so that the method is stable.

Stanford Practice, 32 Applied mathematics, conservation laws, Godunov scheme, Riemann problem

Consider

𝑒𝑑+(𝑓(𝑒))π‘₯=0,π‘₯βˆˆβ„

for a smooth convex flux function 𝑓(𝑒).

(i) Assume the domain ℝ is partitioned into cells 𝐢𝑗 with size Ξ”π‘₯ and center π‘₯𝑗. Describe the Godunov scheme. Discuss the constraint on the time step Δ𝑑.

(ii) Prove that the numerical flux of the Godunov scheme defined using the Riemann problems has the following form

𝑓̂(𝑒𝑗,𝑒𝑗+1)={min𝑒𝑗≀𝑒≀𝑒𝑗+1𝑓(𝑒)if𝑒𝑗≀𝑒𝑗+1max𝑒𝑗+1≀𝑒≀𝑒𝑗𝑓(𝑒)if𝑒𝑗β‰₯𝑒𝑗+1
Stanford Practice, 33 Applied mathematics, symplectic integrators, Hamiltonian systems, Verlet method

Consider a Hamiltonian 𝐻(π‘ž,𝑝)=12𝑝2+𝑉(π‘ž).

(i) Describe the Euler-B and Euler-A methods.

(ii) Prove that Euler-B and Euler-A both preserve the symplectic form.

(iii) Describe the Verlet method.

(iv) Explain why the Verlet method also preserves the symplectic form.

Stanford Practice, 34 Applied mathematics, subharmonic functions, mean value inequality, growth estimates

Suppose that 𝑒:ℝ3→ℝ satisfies Δ𝑒β‰₯6.

(i) Show that

𝑒(0)+35𝑅2≀34πœ‹π‘…3βˆ«π΅π‘…(0)𝑒(𝑦)d𝑦.

(ii) Show that there's π‘₯π‘–β†’βˆž so that lim supπ‘–β†’βˆž|π‘₯𝑖|βˆ’2𝑒(π‘₯𝑖)β‰₯1.

Stanford Practice, 35 Applied mathematics, semilinear heat equation, blow-up, Neumann boundary conditions

Suppose that Ξ©βŠ‚β„π‘› is a compact domain with smooth boundary. Assume that |Ξ©|=1. Let 𝑒 solve

𝑒𝑑=Δ𝑒+𝑒2inΩ×[0,𝑇], πœ•πœˆπ‘’=0onπœ•Ξ©Γ—[0,𝑇], 𝑒(π‘₯,0)=𝑒0(π‘₯)inΞ©,

where 𝜈 is the outwards pointing unit normal to πœ•Ξ©. Assume that βˆ«Ξ©π‘’0=1. Show that the time of existence satisfies 𝑇<1.

Stanford Practice, 36 Applied mathematics, distribution function, layer cake formula, L^p spaces

Let 𝑓(𝑑)βˆˆπΏπ‘(Ξ©), where Ξ©βŠ‚β„π‘›. We say that

πœ‡(𝑑)=π‘š{π‘₯∈Ω:|𝑓(π‘₯)|>𝑑}

is the distribution function of |𝑓|.

(i) Show that

πœ‡(𝑑)≀1𝑑𝑝‖𝑓‖𝐿𝑝(Ξ©)𝑝

and

‖𝑓‖𝐿𝑝(Ξ©)𝑝=π‘βˆ«0βˆžπ‘‘π‘βˆ’1πœ‡(𝑑)d𝑑.

(ii) More generally, show that for any differentiable increasing function πœ‘(𝑠) such that πœ‘(0)=0 we have

βˆ«Ξ©πœ‘(|𝑓(π‘₯)|)dπ‘₯=∫0βˆžπœ‘β€²(πœ†)πœ‡(πœ†)dπœ†.
Stanford Practice, 37 Applied mathematics, L^p spaces, L^infinity norm, measure theory

Let πœ‡ be a positive measure on 𝑋, πœ‡(𝑋)<∞, π‘“βˆˆπΏβˆž(𝑋;dπœ‡) and let

𝛼𝑛=βˆ«π‘‹|𝑓|𝑛dπœ‡,π‘›βˆˆβ„•.

Prove that

limπ‘›β†’βˆžπ›Όπ‘›+1/𝛼𝑛=β€–π‘“β€–βˆž.
Stanford Practice, 38 Applied mathematics, Rademacher functions, orthogonal series, almost-everywhere convergence

Let

πœ‘0(𝑑)={1ifπ‘‘βˆˆ[0,1]βˆ’1ifπ‘‘βˆˆ[1,2)

extend it periodically to all of ℝ, and define πœ‘π‘›(𝑑)=πœ‘0(2𝑛𝑑), π‘›βˆˆβ„•. Assume that βˆ‘|𝑐𝑛|2<∞ and show that the series

βˆ‘π‘›=1βˆžπ‘π‘›πœ‘π‘›(𝑑)

converges for almost every 𝑑.

Questions per page