This action might not be possible to undo. Are you sure you want to continue?

LI MA

Date : Starting from the Summer of 1991 to 2007. 1991 Mathematics Subject Classiﬁcation. 35-XX. Key words and phrases. wave equation, heat equation, harmonic functions, characteristic line method, Fourier method, maximum principle, fundamental solution.

1

2

LI MA

This is a lecture note for the courses of Equations from Mathematical Physics or Methods of Mathematical Physics given by me to senior students in Tsinghua University, Beijing

EQUATIONS FROM MATHEMATICAL PHYSICS

3

To An, for patience and understanding

Monotonicity formulas. Method of separation of variables. Louville theorem. Duhamel principle. Maximum principle. .4 LI MA Contents: Wave equations. Method of Characteristic lines. Fundamental solutions. Variational principle. Fourier transform. Heat equations. Laplace transform. Laplace equation. Mean value property.

EQUATIONS FROM MATHEMATICAL PHYSICS 5 Preface As I always believe that the course about equations from mathematical physics is of fundamental importance to every student in college. many of them are nonlinear equations. Hence. we reduce them into ordinary diﬀerential equations (in short. which will lead to many identities for solutions. it will be a hard time for us to ﬁrstly learn it. However. we have to classify the equations and treat them case by case. The equations from mathematical physics are very important objects in both mathematics and applied mathematics. The crucial observation is that many equations enjoy symmetry properties like translation invariant or rotation invariant. we can not expect to have a unifying theory to handle them. we need to use the divergence theorem from the diﬀerential calculus. Theory of Fourier transformations is one of most beautiful mathematical theories so that there are many books about Fourier transformations. Even for linear partial diﬀerential equations. solving a problem means that we reduce the problem into some easy problems which can be solved by us. The most beautiful method for solving linear partial diﬀerential equations or systems of ﬁrst order is the characteristic line method. which can be solved by us. ODE). For solving our problems with partial diﬀerential equations (in short. Can you image what is up when you enter a new world where you know just a little? I want to point out some key points in our course at this very beginning. PDE). The object in our course is to try to explain some elementary part of solving simple linear partial diﬀerential equations of ﬁrst order and second order. which can be used to translated the problems into ODE again. We shall derive some typical partial diﬀerential equations from the conservation laws of mass or energy or momentum. However. The principle for studying these equations is reduce them into linear ordinary diﬀerential equations. The most powerful methods for solving linear partial diﬀerential equations or systems of second order are the methods of separation for variables (also called Fourier method ) and Fourier transformation method. The basic principle used here is that the change (rate) of the physical quantity in a body in space is created from the change on the boundary and by the source in the interior of the body. Such equations are called equations from mathematical physics. which can be solved by tools from the diﬀerential calculus and linear algebra. Therefore. To us. The other method is to construct the fundamental function to .

for our beginners. which can be used to get Liouville type theorem or regularity theorem. which lead to Bernstein type theorem. ”I suddenly ﬁnd that I know the world after I hear your words ”. Fourier method. or the fundamental solutions to set up energy bounds for solutions. Sections with ahead should be skipped by ﬁrst reading. To understand the behavior of the solutions to PDE.6 LI MA the equation with the homogeneous boundary condition and initial condition. This trick is also very useful in the study of nonlinear parabolic and elliptic equations or systems. we use divergence theorem. Energy bounds or gradient estimates are important to understand the compactness of solution spaces. The aim of this course is to let our students/readers know the basic ingredients in the study of simple linear PDE’s of ﬁrst order and second order. There is another important tool in the study of heat equation and elliptic equations. The nice forms of energy bounds for solutions are monotonicity formulae or Pohozaev type identities . Do you remember an old Chinese proverb? Here it is. one can only get a good understanding of our subject by doing more exercises and by reading more related books. This experience makes me can understand many deep lectures from some famous guys. before we do ”PDE”. In the latter method. It is the maximum principle trick with its brother called the comparison lemma. What did I feel in my college life? I did feel so lucky that I had attended many courses in mathematics when I was in Nankai University. People usually use the maximum principle to derive gradient estimates or Harnack inequalities of non-negative solutions. the divergence theorem plays an essential role. Written by Li Ma in 5-1-2007 . let us do its partner ”ODE”. In conclusion. I hope you understand my words here. Just as in the study of other courses in mathematics. I hope you remember that we try to understand PDE by ODE! In other words.

f (x) = 0}.. ∂xn f ). Let x = (x1 ... PDE=Partial Diﬀerential Equation. We denote by (k ) k1 k 2 kn ∂x f = (∂x ∂ . + kn n 1 x2 the k-th derivatives of the function f with respect to the variables x = (x1 .∂x f )... We denote by C 2 (D) the space of twice diﬀerentiable smooth functions inside the domain DRn . we assume that the functions is decay to zero at inﬁnity.EQUATIONS FROM MATHEMATICAL PHYSICS 7 Notations and conventions: The functions used in our course are bounded smooth function unless we point out explicitly. We denote by Rn the Euclidean space of dimension n.. Here the support of a function f is the closure of the set {x ∈ D. . where r = |x − p| is the radial coordinate with respect to the center p = (pj ) ∈ Rn . we assume the nice conditions for functions such that we can perform the diﬀerentiation into integrands or the change the order of integration of variables. F-transform=Fourier transform.xn ).. If the domain is unbounded. ODE=Ordinary Diﬀerential Equation. . ∇r f (x) := j (xj − pj ) ∂xj f (x) r is the radial derivative of the function f ... ∞ (D) the space of inﬁnitely diﬀerentiable smooth We denote by C0 functions with support inside the domain DRn . ∇f := (∂x1 f. .. ∇j f = ∂xj f.. In other word. .xn ) the standard coordinates in Rn . k = k1 + .

divX = ∂j u and for some smooth function u ∈ C 1 (D) ∩ C (D ∂j udx = D ∂D uνj dσ. Note that when n = 1. . In dimension two. ∂xn u) of the function u. Then. say j − th component.1. We recall the classical divergence theorem: Theorem 1. the divergence theorem is the famous Newtonian-Leibniz formula: b f (b) − f (a) = a f (x)dx. we have (Qx − Py )dx = D ∂D P dx + Qdy.0) ¯ ). ∂xi ∂u If X i = ∂x i for some smooth function.0. ν = (νi ) and X · ν = i X i νi . A special case we often use is that X has only one non-zero component. The classical divergence theorem is a basic theorem used in our course. closure D Then we have the following formula: divXdx = D i X · νdσ. . u. Recall that the Green formula is the following: for any smooth function P. . Let (xi ) be the standard coordinates in the Euclidean space Rn . Q on the closure of the domain D. X is the gradient ∇u := (∂x1 u. 0. then divX = i ∂2u = ∆u (∂xi )2 which is the Laplacian of the function u. i... Let D be a bounded domain with ﬁnite piece-wise C 1 ¯ → Rn be a smooth vector ﬁeld on the boundary ∂D in Rn . . Let ν be the unit outer normal to ∂D. X = (0.. Divergence theorem.. We ask the readers to remember this basic formula. lecture one 1.e..8 LI MA 1.. Let D be a bounded smooth domain in the plane R2 . ∂D Here divX = i ∂X for X = (X i ) . Let X : D ¯ of the domain D.. the divergence theorem is Green’s theorem in differential calculus..

Then we have divX = Qx − Py . y (s)). Choose X 1 = Q and X 2 = −P . Green’s formula implies that divXdx = D ∂D X · νds. . x ) is the interior unit normal to the curve p(s) such that {T. The outer unit normal to the boundary ∂D is ν = −N = (y . We want to prove that they are equivalent each other.EQUATIONS FROM MATHEMATICAL PHYSICS 9 It is clear that the form of divergence theorem in dimension two is diﬀerent from that of the Green formula. and along the boundary ∂D. Assume that the boundary ∂D is given by the smooth curve p(s) = (x(s). Then N = (−y . P dx + Qdy = (P x + Qy )ds = X · νds Hence. Write by x = dx/ds. which is the divergence theorem in dimension two. N } forms a right-hand orthonormal basis system. Then T = (x . y ) is the unit tangent vector along p(s). −x ). etc. where s is the counter-clock arc-length parameter.

Let u = u(x. t) be the position function of the string at (x. 0 where c is a physical constant of the string. t) in the direction of y -axis. t) ∈ C0 ((0. < δH (u). t) at (x.10 LI MA 1. d < δH (u). We consider a vibrating string of length l with its two ends ﬁxed at x = 0 and x = l. l) × (0. 2 0 The potential (or spatial) energy is c U (u) = 2 l ρu2 x dx. t). φ >:= We compute that t l d H (u + ξ )| =0 = 0.2. φ > = 0 t dτ 0 l (ρut φt − cρux φx + f0 φ)dx = 0 dτ 0 (−ρutt + cρuxx + f0 )φdx . Assume that the string is acted by the exterior force with force density f0 (x. Then the kinetic (or temporal) energy of the string is 1 l 2 T (u) = ρut dx. Assume that ρ = constant is the mass density of the string. for every ξ = ξ (x. t1 )). One dimensional wave equation. Then the energy obtained by the exterior force is l U (f ) = − 0 f0 udx. 2 that is. The Hamilton principle implies that the motion of the vibrating string obeys δH (u) = 0. Then the Hamiltonian action of the string is t H (u) = 0 dτ (T (u) − U (u) − U (f0 )) t l 1 2 c − 2 = dτ 0 t 0 l ρu2 t dx t l dτ 0 0 ρu2 x dx + 0 dτ 0 f0 udx.

t) + f (x. So we know that u(x. which is called the one dimensional wave equation.EQUATIONS FROM MATHEMATICAL PHYSICS 11 Using the variational lemma (see Appendix). we then get utt − cuxx = f0 /ρ. 0) = φ(x). By physical testings. and ut (x. t) = a2 ∆u(x. t) = 0 = u(l. and the boundary condition u(0. we can ﬁnd the initial position and velocity of the string. t). . t). 0) = ψ (x). We can also add other type of initial data and boundary data for the wave equations. The higher dimensional wave equation in its simple form is utt (x. which is a typical hyperbolic partial diﬀerential equation of second order.

Let u be the velocity ﬁeld of the ﬂuid. t)) = 0. Let ρ = ρ(x. Hence. t) be the mass density of a moving ﬂuid in the region W . So the total amount ﬂowing into W across the boundary is the integration over ∂W : − ρu · νdσ. Let ν be the outer unit normal to the boundary of the domain W . Take a small part dσ in the boundary ∂W . ∂t This is the continuity equation in Fluid mechanics. Then the change rate is only across ﬂuid rate from the boundary. then the continuity equation is ρt + ∂x (a(x)ρ(x. we have ∂ ( ρ + div (ρu))dx = 0. Take a small part dx in W . 0. where x = x1 . Conservation of mass. .3. which is ρdx. dt W Assume that there is no ﬂuid generating source. W Then.12 LI MA 1. the total mass in W at time t is the integration of ρdx over W . If u = (a(x1 ). ∂W So we have d ρu · νdσ. its change rate at time t is d ρdx. ρdx = − dt W ∂W Using the divergence theorem. we have ∂ ρ + div (ρu) = 0. Then the mass in dx is ρdx. Then the amount of −ρu · νdσ ﬂow into W . 0). W ∂t Since we can take W be any sub-domain.

If a sequence (fj ) ⊂ E satisfy |fj − fk | → 0 as j.. the norm | · | comes from an inner product < ·. t . we must try to solve it and get the explicit expression for the solutions in terms of initial data and boundary data. for example. We also need to know the uniqueness and stability of our solutions to our equations or problems. then we call (fj ) a Cauchy sequence in (E. |f |2 =< f. f >. where t ∈ [0. By deﬁnition. A vector space E equipped with a norm | · | is called a normed space. | · |) a Banach space. ∂x ) = j =0 (j ) aj (x)∂x is a partial diﬀerential operator of the variables (x). . A norm | · | on a vector space E is a function from E to non-negative numbers R+ such that (1) |f | = 0 if and only if f = 0.. < ·. t) is the stable solution (in the norm | · |) to the problem A[u](x. then we call (E. |w(0) − u(0)| ≤ δ. with initial data u(x.. g >= D f (x)g (x)dx. < f. (3) |f + g | ≤ |f | + |g |. · >. we call (E. t) := (∂t u − Lu)(x. · >) a Hilbert space.EQUATIONS FROM MATHEMATICAL PHYSICS 13 1. we say that u(x.4. ∂x . If every Cauchy sequence has a limit in E . t) = 0. and k L := (k ) P (x. 0) if for any > 0. k → ∞. t) to A[w] = 0 with initial data w(0). f. there exists δ > 0 such that for all solutions w(x. | · |). Here we may introduce the concepts of Hilbert space and normed space (see any textbook about Functional Analysis ). What are important for our equations? What I want to show here is that once we have a pde from scientiﬁc background. If further. that is. g ∈ L2 (D). we have sup |w(t) − u(t)| < . (2) |af | = |a||f | for any real number a. T ].

we need to know some typical methods to solve linear pde’s of ﬁrst order and second order. we need to ﬁnd some nice norm for solutions and derive a Gronwall type inequality. Historically. we have to use maximum principle trick or monotonicity formulae for solutions. which depend on what we wanted in reality. the solution expressions were guessed by scientiﬁc workers from physical intuition. So it is important for us know the scientiﬁc background of the equations. For stationary problems. To get uniqueness and stability of solutions to evolution problems. In our course.14 LI MA There are also many other deﬁnitions for stabilities. .

Using the Fourier law. t) ∈ ∂ Ω × (0.EQUATIONS FROM MATHEMATICAL PHYSICS 15 2. we have cρ∂t u = div (k ∇u) + ρf0 . which is a typical parabolic partial diﬀerential equation of second order. we can measure the heat density on the boundary ∂ Ω. t) = αu(x. we have cρ∂t u = −div q ¯ + ρf0 . ∂t D cρudx = ∂D q ¯ · (−ν )dσ + D ρf0 dx. t) be the density of heat source. x ∈ Ω. ∞). we have. Then as in the case of continuity equation. t). (x. which is called the heat equation with source f0 . (x. The heat equation coupling with initial and boundary conditions is called a determined problem of heat equation. we have q ¯ = −k ∇u. t) ∈ Ω × (0. and let q ¯ be the heat ﬂowing velocity. since D can be any domain in the body. ∞). Using the divergence theorem we have q ¯ · (−ν )dσ = − ∂D D div (¯ q )dx. lecture two 2. ∞). By Fourier’s law. Physically. where k > 0 is a coeﬃcient of physical quantity. and we may assume that u(x. Heat equation and Laplace equation. The initial heat density at t = 0 can also be detected. t) ∈ ∂ Ω × (0. Hence. Let f0 (x. we may have other kind of boundary conditions like k ∂u (x. t) = 0 for (x. Then we have the heat equation ut = ∆u. for a physical constant c > 0 of the region D ⊂ Ω. Heat equation and Laplace equation are very important objects in PDE theory. We use the energy conservation law to derive the heat equations. Of course.1. ∂ν . Let u be the heat density of the body Ω. Assume that c = 1 = k = ρ and f0 = 0. 0) = φ(x). So we assume that u(x.

that ut = 0. In dimension one. if the dimension n ≥ 2. it is not easy to solve the Laplace equation on domains coupling with its boundary condition. However. we have the following equation: ∆u + f = 0. So u = ax + b is a linear function of the single variable x. the Laplace equation becomes u = 0. . which is a typical elliptic partial diﬀerential equation of second order.16 LI MA When we have stationary heat in Ω. we try to solve some simple pde of ﬁrst order. where f = ρf0 . it is called the Laplace equation in Ω. or in other form −∆u = f which is called the Poisson equation. In the following subsection. If f = 0.

Using integration along the curve C (t). t) with the same initial data as u. We introduce the C -curve C (t) (called the characteristic curve) by solving dx/dt = a. t) = f (at + A. τ )dτ. we have u(A. Then we have x(t) = x and we can ﬁnd that initial point on the curve is x(0) = A = x − at. One may study examples in our textbook. we have x(0) = A. Solve pde of ﬁrst order. we ﬁnd (FF): t u(x. t) = u0 (x − at) + 0 f (x + a(τ − t).EQUATIONS FROM MATHEMATICAL PHYSICS 17 2. t) on the curve C (t). In fact. When t = 0. 0) + 0 f (aτ + A. we have t u(x(t). So one should know how to solve an ODE. This lemma is based on a very simple observation from the formula (FF) that if u0 = 0 and f ≥ 0. we have du/dt = ut + aux = f (x. We study a ﬁrst order partial diﬀerential equation. . So by the initial value condition. let g (x. On the curve C . Our question is to solve the following equation (FDE): ut + aux = f (x. t) = u(A. 0) = u0 (A) = u0 (x − at). The method above is called characteristic curve method in the literature. The principle for solving a ﬁrst order PDE is to reduce it into ODE along characteristic curves. T ]. t). Given the point (x. Let w be a smooth function satisfying wt + awx ≥ f (x. t).2. t) = wt + awx − f (x. τ )dτ. In case when f = 0. we have u(x. Then we have C := C (t) : x = at + A. We point out the following comparison lemma: Let u be a smooth solution to (FDE) with the initial data u(0) = u0 . then u ≥ 0 for all t. Assume that we have the initial condition u = u0 (x) at t = 0. Inserting A = x − at into the expression above. t). t) = u0 (x − at). We shall review some ODE theory when it is necessary. Then we have w ≥ u on R × [0.

Then we have vt + avx = g (x. t) ≥ 0 with v = 0 at t = 0. Usually. t) + g (x. t) ≥ 0 and wt + awx = f (x. t).18 LI MA Then we have g (x. we know that v ≥ 0. Let v = w − u. Using the formula (FF) for v. the comparison lemma is a very important property for PDE’s. . Hence w ≥ u.

. a pde (PDE) P (u. 0) = x for x ≥ 0. Let’s study the following nonlinear PDE of ﬁrst order: 1 ut + (1 + u)ux = 2 √ with the initial value condition u(x. it is called a nonlinear PDE. 4 2 We now solve u along the curve x(t) and get x(t) = A + t + u(x(t). ∇u. . dt Note that along the characteristic curve. we have 1 du = dt 2 and t t + u(x(0). we get √ t A = u(x(t). we ﬁnd the curve satisfying u(x(t). We have just used The characteristic curve method to study the linear pde of ﬁrst order. t)2 . t) = t t x(t) = A + t + 0 u(x(τ ). A nonlinear PDE example. D2 u. Otherwise.. 0) = + x(0) 2 2 On one hand. t) = x(t) − t. t) − . We ﬁrst ﬁnd the characteristic curve x(t) by solving dx = 1 + u(x.EQUATIONS FROM MATHEMATICAL PHYSICS 19 2. We show here that this method can also be used to study the nonlinear pde of ﬁrst order. t). for any given two solutions u and v of (PDE). Then we have t √ u(x(t). we set x(0) = A. au + v is also a solution of (PDE) for every constant a ∈ R. By deﬁnition.3.) = 0 is called a linear equation if.. 2 On the other hand. 2 Then we have t2 √ t √ + At = t + ( + A)2 = t + u(x(t). τ )dτ = A + t + 0 τ √ ( + A)dτ. t) = + A 2 √ Solving A from this relation.

t) on the curve x(t). . we have x(t) = x. Hence. t) = x − t.20 LI MA Given any point (x. the solution is √ u(x.

Let fτ = f (τ ) and solve (ODE) at t = τ as above to get the solution (S). Then its solution to (ODE) is (S): t u(t. Then the function deﬁned by t u(t. with initial condition u = fτ at t = τ . lecture three 3.1. We now try to solve the equation (F): u + p(t)u = f (t) with initial condition u = 0 at t = 0.EQUATIONS FROM MATHEMATICAL PHYSICS 21 3. τ )dτ 0 is the solution to (F). τ ) = fτ exp(− τ p(s)ds). Duhamel principle. . Consider (ODE) u + p(t)u = 0. This kind of method is called the Duhamel principle.

dse− Rs a (τ ) b(s). Then we have f = βu. Let g (t) = w − β (t)f (w).T]. Hint: Let t f (t) = a + β 0 u(τ )dτ. we get f ≤ βf. Then we have v == β (t)(f (w) − f (u)) + g (t) = β (t)B (t)v (t) + g (t) . By assumption.2. we have (F): u(t) = e Rt t a(τ ) Rt t a(τ ) a(τ ) +e dse− a (τ ) b(s). Set v = w − u. Then g (t) ≥ 0. We consider the following nonlinear ODE u = β (t)f (u) with initial data u = a at t = 0.22 LI MA 3. Assume that w ≥ β (t)f (w) with initial data w = a at t = 0. First order ODE and a comparison lemma. Then we have w(t) ≥ u(t) for all t. u ≤ f . For the given ODE u = a(t)u + b(t) we can write it as (e− Rt a (τ ) u) = e− Rt Rt a (τ ) b(t) Rs So it has a general solution u(t) = u(0)e If u(0) = 0.T] satisfying t u(t) ≤ a + β 0 u(τ )dτ. Let β (t) and f (s) be two smooth functions. Then we have u(t) ≤ aeβt on [0. One often use the argument above to prove the famous Gronwall inequality: Let β > 0 and let u be a continuous function on [0.

That is w ≥ u for all t. Let a(t) = β (t)B (t) and b(t) = g (t) in (F). Then we have v ≥ 0.EQUATIONS FROM MATHEMATICAL PHYSICS 23 with v = 0 at t = 0 and B (t) = f (zw + (1 − z )v )dz. . We remark that one can use Sturm-Liouville theory and variational principle to set up comparison lemma for second order ODE.

t) dt is the solution to utt − uxx = 0 with initial data u|t=0 = u0 and ut |t=0 = 0. Let’s ﬁrst assume that u0 = 0. We shall follow our textbook [1] for the explanation. then we can verify that d u(x. 0) = u1 (x). We consider the one-dimensional wave equation (OW): utt − uxx = 0. We now solve v : v (x.24 LI MA 3. t u(x.3. Note that 2 2 ∂t − ∂x = (∂t − ∂x )(∂t + ∂x ). Then we reduce the problem into two pde of ﬁrst order. t > 0. Using initial condition we have v (x. t) = [ u1 (s)ds] := Mu1 . Let v = ∂t u + ∂x u. So. . t) = v0 (x + t) = u1 (x + t). τ )dτ. x ∈ R. t) = Mu0 (x. but u1 = 0. with initial data u|t=0 = u0 and ut |t=0 = u1 . Changing the variable we can get 1 x+ t u(x. s = x + 2τ − t. t) = 0 v (x + (τ − t). Solving u we get t u(x. 2 x− t If u0 is non-trivial. One dimensional wave equation. t) = 0 u1 (x + (τ − t) + τ )dτ. and ∂t v − ∂x v = 0.

Remark One: We remark that if u0 = 0 and u1 ≥ 0. From this formula we know that the solution to the problem (OW) is unique with the formula (EW).EQUATIONS FROM MATHEMATICAL PHYSICS 25 This is the Duhamel principle for the wave equation. Hence. One can show that. We leave this as exercise for readers. 2 R t . This is the ﬁnite propagation property of the wave equation. then u ≥ 0 for all t. For each t. x− t u(x. we can get the D’Alembert formula (EW): 1 u(x. t) = [(u0 (x + t) + u0 (x − t) + 2 for the solution to utt − uxx = 0 with the initial data u|t=0 = u0 and ut |t=0 = u1 . we have u = 0. Similar bound is also true for ux and ut at each t provided |u0x (x)| ≤ C0 and |u1 (x)| ≤ C1 . t − s)ds is the solution to utt − uxx = f (x. u1 = 0 for all |x| > R for some R > 0. t) = 0 Mfs (x. t x+ t u1 (s)ds]. t)| ≤ M0 + tM1 is bounded provided |u0 (x)| ≤ M0 and |u1 (x)| ≤ M1 . We now assume that u0 = 0. for fs (x) = f (x. s). |u(x. ut = 0 for |x| > R + t. Remark Two: Deﬁne the energy for u by 1 E (t) = (u2 + u2 x )dx. using the Duhamel principle. t) with the initial data u|t=0 = 0 and ut |t=0 = 0. Then from the expression (EW).

26

LI MA

Then we have

d E (t) = (ut utt + ux uxt )dx. dt R Using the integration by part to the last term we get d ut (utt − uxx )dx = 0. E (t) = dt R So the energy E (t) is a constant, which is 1 (u2 + u2 0x )dx. 2 R 1 Remark Three: We now study the one dimensional wave equation on half line (HWE): utt − uxx = 0, x ∈ R+ , t > 0, with initial data u|t=0 = u0 and ut |t=0 = u1 . Note that u0 (t + x) is the special solution to (HWE). So we can use it to transform the problem into the case when u0 = 0. Then we can use the odd extension method on the x-variable to extend the function u1 and the equation on the whole line. Then we can use the D’Alembert formula to solve (HWE). One may also use this method to treat one dimensional wave equation on an interval, however, we have a more beautiful method—Fourier method/Separation variable method to study this case. See the next section. Remark Four: We point out that we can use the D’Alembert formula to treat some special case of higher dimensional wave equation. Here is the example : Consider the following problem for 2-dimensional wave equation utt − uxx − uyy = 0 with the initial data u|t=0 = φ(x) and ut |t=0 = ψ (y ). Then we split the problem int to two one-dimensional problems: One is utt − uxx = 0 with the initial data u|t=0 = φ(x)

EQUATIONS FROM MATHEMATICAL PHYSICS

27

and ut |t=0 = 0. The solution for this one dimensional problem is 1 u(x, t) = [(φ(x + t) + φ(x − t)] 2 The other is utt − uyy = 0 with the initial data u|t=0 = 0 and ut |t=0 = ψ (y ). Its solution is 1 y +t ψ (s)ds]. u(x, t) = [ 2 y −t Hence, the solution to our original 2-dimensional problem is 1 u(x, y ; t) = [(φ(x + t) + φ(x − t) + 2

y +t

ψ (s)ds].

y −t

28

LI MA

4. lecture four 4.1. The method of separation of variables 1. The separation variable method for boundary value problems is looking for a solution as a sum of special solutions of the form X (x)T (t). Let u(x, t) = X (x)T (t) be a solution to the wave equation utt = a2 uxx on [0, l] × [0, ∞) with the homogeneous boundary condition u(0, t) = 0 = u(1, t). The boundary condition gives us (BC): X (0) = 0 = X (1) Using the wave equation we have T X (x) = a2 T X , or T X = . a2 T X The left is a function of variable t and the right is a function of variable x. To make them be the same, they should be the same constant. Say −λ. Then we have one equation for T : T + a2 λT = 0 and the other equation for X : X + λX = 0 with the boundary condition (BC). The latter is the simple case of the Sturm-Liouville problem. The key step in the Separation Variable Method is to solve this problem at ﬁrst. Case 1: When λ < 0, the general solution for X is

√

X (x) = c1 e Using (BC) we have

−λx

+ c2 e−

√

−λx

c1 + c2 = 0 and c1 e −λ + c2 e− −λ = 0, which implies that c1 = c2 = 0. Case 2: When λ = 0, the general solution for X is X (x) = c1 + c2 x. Again (BC) implies that c1 = c2 = 0.

√ √

the expansion can be used . t) = k=1 [ak cos(kπat) + bk sin(kπat)] sin(kπx) to the wave equation utt = a2 uxx on [0. For general boundary conditions. the general solution for X is √ √ X (x) = c1 cos( λx) + c2 sin( λx). t) = 0 = u(1...EQUATIONS FROM MATHEMATICAL PHYSICS 29 Case 3: When λ > 0. .. t). The boundary condition (BC) gives us that c1 = 0 and √ c2 sin( λ) = 0.. Hence X (x) = Xk (x) = ck sin(kπx) for k = 1. t) = [ak cos(kπat) + bk sin(kπat)] sin(kπx) is a typical solution to the one-dimensional wave equation with boundary condition. t = 0 and the boundary condition u(0. we get Tk (t) = ak cos(kπat) + bk sin(kπat). We now look for a solution of the form u(x. but we have to invoke the Sturm-Liouville theory (see Appendix). 2. We need to use the initial condition to determine the constant ak and bk . 2.. So we have +∞ φ(x) = k=1 ak sin(kπx) . So Uk (x. We remark here that a easy way to see λ ≥ 0 is using the integration relation: λ +∞ |X |2 = |X |2 . k = 1. ut = ψ (x). ∞) with the initial condition u = φ(x). 1] × [0. The reason for us to use such an expansion is the Fourier series theory. . and then λ = k 2 π 2 . Returning to the equation for T (t) with λ = k 2 π 2 .

t) = α(t).30 LI MA and ψ (x) = +∞ kπabk sin(kπx). This is a basic method to solve linear pde. t) = β (t). t) + α(t) + x(β (t) − α(t)) and reduce the problem for v with the homogeneous boundary condition. kπa In this way. . . k=1 Using the Fourier series expansion for φ(x) and ψ we have ak = 2 and φ(ξ )sin(kπξ )dξ 2 ψ (ξ )sin(kπξ )dξ. saying bk = u(0. If the boundary condition is not homogeneous. we get at least formally a solution to our problem. we have to make a transformation u(x. t) = v (x. u(1.

where ω = for j = 1. Then we have t jπx ajπ Laj u(x. The resonance problem. 2. 0) = 0 = ut (x..2. t) = u(L. (x. t) ∈ [0.EQUATIONS FROM MATHEMATICAL PHYSICS 31 4. t) = sin( ) sin(ωτ ) sin( (t − τ ))dτ ajπ L L 0 j ajπ L = j ωj (ω 2 aj jπx (ω sin ωj t − ωj sin ωt) sin( ) 2 − ωj ) L ωj = ajπ L jπx )dx.. L where and 2 aj = L Because we have 0 L A(x) sin( 2 ω 2 − ωj which may cause the coeﬃcient aj → ∞. 0). 2 2 ωj (ω − ωj ) We say that we may have the resonance phenomenon caused by ω . t) = 0 and u(x. Consider utt − a2 uxx = A sin(ωt). .. . We now brieﬂy discuss the resonance problem of the one dimensional wave equation. ∞) with homogeneous initial and boundary conditions u(0. L] × (0.

t) = 0. B (R) = 0. As before. Introduce the polar coordinates (r. we ﬁnd that β = βn = n and hn (θ) = An cos(nθ) + Dn sin(nθ). y = r sin θ. y ) be the solution to our wave problem. h(θ + 2π ) = h(θ). Then ∆u = urr + r−1 ur + r−2 uθθ . The reason that we use polar coordinates here is because of the boundary condition in our problem being on the circle. t) = 0. we have h (θ) r2 B + rB + k 2 r2 B =− = −β 2 . we have B + r−1 B + (k 2 − n2 r−2 )B = 0. Then T + a2 λT = 0 and ∆U + λU = 0. x ∈ ∂BR (0) and the initial conditions u|t=0 = u0 and ut |t=0 = u1 . The last equation is called the Bessel equation. We denote BR (0) ⊂ R2 the ball in the plane R2 . ∞) with the homogeneous boundary condition u(x. θ) by x = r cos θ. Let’s consider the 2-dimensional wave equation utt = a2 (uxx + uyy ) on BR (0) × [0. and B + r−1 B + (k 2 − β 2 r−2 )B = 0. h(θ) B Then we get two problems: h + β 2 h = 0. . Setting U = h(θ)B (r) in the latter equation. It is easy to see that λ = k 2 > 0. Wave equation in two dimensional disk.32 LI MA 4. B (R) = 0. U (R. Inserting β = n into the Bessel equation. Let u = T (t)U (x.3.

Some people would like to deﬁne the Bessel functions of order n as the coeﬃcients of the Fourier expansion: ∞ exp(ir sin θ) = where i = √ n=−∞ Bn (r)einθ . and then we can use this basis to solve the wave problem above. One may look for reference book for more about the Bessel functions. Bessel functions form an orthonormal basis. y ) = X (x)Y (y ). Bn (r) = . −1 and e inθ = cos(nθ) + i sin(nθ). if the domain is not a ball but a rectangle. we just let U (x. 2π 1 exp(ir sin θ − inθ)dθ. 2π 0 We remark that. Just like trigonometric polynomials. Hence.EQUATIONS FROM MATHEMATICAL PHYSICS 33 The general solutions of this equation are called the Bessel functions of order n.

t) = 1 4πa2 t ψ (x)dσx . let 1 I (x.4. t) to the homogeneous equation utt − ∆u = 0. In fact. t) ∈ Rn × R+ with the initial condition u = φ(x). In dimension three. t) = rI (x.34 LI MA 4. r = |x − p| = at}. we ﬁnd that 1 I (x. f or (x. at t = 0 by assuming that we know the solution u(x. r. The other way to understand the higher dimensional wave equation for us is to try the Duhamel principle to the wave equation in Rn utt − ∆u = f . at t = 0. r. ut = 0. at t = 0. ut = ψ (x). we ﬁrst study the wave equation problem (3W): utt − ∆u = 0. t) ∈ Rn × R+ with the initial condition u = 0. g ) = udσ 4π ∂Br (x) for g = g (y ) ∈ C 2 (R3 ) and J := J (x. ut = ψ (x). r. Sat (p) where Sat (p) = {x ∈ R3 . g ) = u(x + rθ)dθ 4π {|θ|=1} . at t = 0. u) for u = u(y. R3 × {t > 0} with the initial data u = 0. We then use the descent method to study utt = ∆u in Rn ×R in one dimension higher space Rn+1 ×R with initial conditions u = 0. ut = ψ (x). f or (x. r. t) be the solution to the problem (3W). We ﬁnd that u(p. Wave equation in R3 . Changing variables.

R. g )] 2 = 2R∂R I (x. R. g ). g ). r. g )dr. g )dr = R2 ∂R I (x. We now compute R ∆x 0 4πr2 I (x. g )dr 0 = ∆x ∂R 2 = R ∆x I (x. r. r.EQUATIONS FROM MATHEMATICAL PHYSICS 35 and g (x + z )dz = BR (0) R 2 0 R r dr {|θ|=1} u(x + rθ)dθ = = 0 4πr2 I (x. R. Note that R ∂R ∆x 0 R r2 I (x. g )dr g (x + z )dz BR (0) = ∆x = BR (0) ∆x g (x + z )dz ∆z g (x + z )dz BR (0) = = ∂R g (x + z )dσz ∂BR (0) = ∂B1 (0) ∂R g (x + Rθ)R2 dθ = = 4πR2 ∂R I (x. So. g )dr r2 I (x. R. we have R ∆x 0 r2 I (x. g ). g ) + R2 ∂R I (x. r. R. R. r. . g ) and ∂R [R2 ∂R I (x.

36 LI MA Hence. and Jt (x. ψ )dr. g ) 2 = = 2∂R I (x. R. at−R . R. R. g ) + R∂R I (x. 0) = 0. R. we have ∆x RI (x. ∂R This is a one dimensional wave equation. we have the boundary condition J (x. t) = 1 2a at+R rI (x. Now. R. R. R. R. r. We now determine its boundary condition and initial condition. R. g ) 2 = ∂R (RI (x. Then we have J (x. t)|R=0 = 0. t)dθ ∂B1 (0) = a2 ∆x (RI (x. R. g )). u)) = a2 ∆x (RI (x. R. R. t)|t=0 = RI (x. ∂2J ∂t2 = R 4π utt (x + Rθ. 2 R2 ∆x I (x. R. At t = 0. u)) ∂2J = a2 2 . t)|t=0 = RI (x. Dividing both sides by R. g ) + R2 ∂R I (x. R. t)dθ ∂B1 (0) a2 ∆x R 4π u(x + Rθ. R. g ) = 2R∂R I (x. Clearly. g ). t)dθ ∂B1 (0) a2 R = 4π = ∆x u(x + Rθ. ψ ). J (x.

4πa2 t Sat (x) Then we use the Duhamel principle in the study of the wave equation u(x. One may see other textbooks for the other understanding of the Kirchhoﬀ formula. t) = utt − ∆u = f . ψ )dr R→0 2aR at−R 1 [RI (x. and we have the famous Kirchhoﬀ formula for the solution: 1 r−1 f (x. 1 ψ (y )dσy . at. ψ )]|R=at a tI (x. 4π ∂B1 (0) . we can see the Huygens principle that the wave at x0 at time t = 0 inﬂuences the solution only on the boundary of a cone originating from x0 . R3 × {t > 0} with the initial data u = φ(x).EQUATIONS FROM MATHEMATICAL PHYSICS 37 Hence. r = |x − p| ≤ at}. r. t) R→0 R at+R 1 lim rI (x. one can get a solution formula for two dimensional wave equation. R. t) = 4πa2 Bat (p) 1 + ψ (x)dσx 4πa2 t Sat (p) 1 φ(x)dσx ] + ∂t [ 4πa2 t Sat (p) where Bat (p) = {x ∈ R3 . ut = ψ (x). at t = 0. u(x. J (x. R. t − a−1 r)dx u(p. From this formula. We just remark that using the descent method. ψ ) t ψ ((x + atθ)dθ. t) = lim = = = = That is.

Say −λ. Let u(x. Then we have T + a2 λT = 0 and (EX): X + λX = 0.. t) = X (x)T (t) be a solution to the heat equation ut = a2 uxx on [0. The method of separation of variables 2. As in the wave equation case. 2 aT X The left is a function of variable t and the right is a function of variable x. we impose the homogeneous boundary condition X (0. t) = 0 = X (1. lecture five 5. Then we have X (x) = Xk (x) = ck sin(kπx) for k = 1. Then we solve for T and get Tk (t) = T (0)exp(−a2 λk t) by imposing initial conditions for T . l] × [0. t) to solve X and λ.38 LI MA 5. We begin with a typical example for heat equation with ﬁxed boundary conditions. we can solve the heat equation by looking at solutions of the form Ak exp(−a2 k 2 π 2 t) sin(kπx) k X2 = |X |2 > 0. Using the superposition law.. . Then formally. and λ = λk = k 2 π 2 . we have X T = . The lesson we have learned is that the solution to heat equation has exponential decay to 0 in t-variable as t → ∞. Multiplying both sides of (EX) by X and integrating over [0. We now describe the main idea. ∞). 1]. with the initial data φ(x) = k Ak sin(kπx) . 2. To make them be the same. they should be the same constant.1. we have λ So λ > 0.

t).EQUATIONS FROM MATHEMATICAL PHYSICS 39 Assume that we are given the heat equation ut = a2 uxx with non-homogenous boundary condition. we do the expansion for f f (x. Here uj (0) are determined by φ(x) = j uj (0)Xj (x). with the boundary condition X (0) = 0 = Xx (L). we need to use transformations to make the boundary data into a homogenous condition. we need to solve for {Xj } and λ = λj to the Sturm-Liouville problem X + λX = 0. Hence. Then we reduce the non-homogenous problem into the ODE uj = −a2 λj uj + fj (t) with suitable initial condition. . Then we have to meet the following non-homogenous heat equation ut = a2 uxx + f (x. Once this done. t). t) = j e−λj t (uj (0) + 0 eλj τ fj (τ )dτ )Xj (x). we have the expression t uj (t) = e−λj t (uj (0) + 0 eλj τ fj (τ )dτ t and u(x. t) = 0 = ux (L. We omit the detail here. t) such that u(x. To study this problem. t) = j uj (t)Xj (x). t) = j fj (t)Xj (x) and look for solution u(x. saying u(0.

. uj (0) = 0 for all j . we have ∞ 2 (uj (t)2 + λ2 j uj (t) )dt ≤ 0 0 ∞ fj (t)2 . ∞) deﬁned by f = j 0 |fj (t)|2 dt < ∞. Then we have T 2 2 (uj (t)2 + λ2 j uj (t) )dt − λj uj (T ) = 0 0 T fj (t)2 .2. Hence. Note that we have assumed that λj < 0. 2 which implies that ∂t u 2 + λ2 u ≤ f .2 (D × [0. Then we have uj (t) = λj uj (t) + fj (t). Deﬁne the function space E such that the functions {φj } form an orthonormal basis for the space E with its norm given by u = a2 j for u = aj φj (x) ∈ E . with the initial data u = 0 at t = 0.40 LI MA 5. Let L be a partial diﬀerential operator on the domain D ⊂ Rn such that for some 0 > λj ∈ R. with a ﬁxed homogeneous boundary condition. Energy bound. as before. Assume that the expansion u = uj (t)φj (x) is a solution to the partial diﬀerential equation: ut = Lu + f. Hence. we have Lφj (x) = λj φj (x). Hence. Note that 2 2 (uj (t) − λj uj (t))2 = uj (t)2 + λ2 j uj (t) − 2λj uj (t)uj (t) = fj (t) . Assume that the known function f has the expansion f= fj (t)φj (x) ∞ with its norm on D × [0. we have the expression t uj (t) = eλj t 0 e−λj τ fj (τ )dτ. Given a smooth function f ∈ C 1. t > 0. ∞)).

EQUATIONS FROM MATHEMATICAL PHYSICS 41 where λ2 = infj {λ2 j }. This is the energy bound for the solution as we wanted. .

t2 u = 2|x|tβ −1 f ∂2 = u ∂ |x|2 = 2tβ −1 f + 4|x|2 tβ −2 f = 2tβ −1 f + 4rtβ −1 f n−1 ∆u = u + u |x| = 2tβ −1 f + 4rtβ −1 f + 2(n − 1)tβ −1 f = 4rtβ −1 f + 2ntβ −1 f . Take A = 0. λ2 t) is also a solution of (HE). So we try to look for solutions of the form u(x. t) = u(λx. with the initial data u = g (x) at t = 0. +∞): ut = ∆u. Then we have 4f = −f. 4rn/2 f + rn/2 f = A. t) = tβ f (|x|2 /t). r r Let β = −n/2. Then for some constant A.3. Then we ﬁnd that ut = βtβ −1 f + tβ (− u = ∂ u ∂ |x| |x|2 )f = βtβ −1 f − rtβ −1 f . . Note that for λ > 0. β 2n − f + f + 4f + f = 0.42 LI MA 5. the function uλ (x. Then we have 4(rn/2 f ) + (rn/2 f ) = 0. So. Fundamental solution to Heat equation. Set r = |x|2 /t. We study the heat equation (HE) on Rn × [0. Assume that u is a smooth solution of (HE).

Then we have f → 0 as r → +∞. 2 Step two: We take δ1 > 0 small enough such that for 0 < t < δ1 . Without loss of generality. we have some δ0 > 0 such that. and then u satisﬁes (HE) since K does. for t > 0. Rn Theorem 2. It is easy to verify that K (x. Assume that there is a positive constant M > 0 such that |g (x)| ≤ M. K (x. So f = Be−r/4 . Step one: Given an arbitrary small > 0. We use two step argument. Choose B = 1/(4π )n/2 . |g (y ) − g (x)| < . we can take derivatives into the integrand term. t)dx = 1. x ∈ R. Proof. . t)dy < {|x−y |≥δ0 } 4M . t) → g (x). e (4πt)n/2 This is called the Gaussian kernel.t) satisﬁes the heat equation (HE) and as t → 0. t) = |x|2 1 − 4t .EQUATIONS FROM MATHEMATICAL PHYSICS 43 Assume that f → C = const. Given g (x). Deﬁne. which is also called the Poisson kernel. t) = Rn K (x − y. Then u(x. For any ﬁxed point x ∈ R. K (x − y. as r → +∞. we let u(x. u(x. for |x − y | < δ0 . t) is the solution to our problem at the beginning of this subsection. we now assume n = 1. Roughly speaking. t)g (y )dy. the theorem says that this u(x. For t > 0.

t)(g (y ) − g (x))dy | {|x−y |<δ0 } + | ≤ 2M K (x − y. < 2M Here we have used the fact that K ≥ 0 and K (x − y. Assume g ∈ C 0 (Rn ) satisﬁes g (x) → 0. t)(g (y ) − g (x))dy | K (x − y. we have u(x. we have supBR (x) |g (x)| → 0. t)dy = 1 for any t > 0. Then for the solution u(x. t)dy {|x−y |≥δ0 } + {|x−y |<δ0 } K (x − y. We now study the behavior of the solution deﬁned above as |x| → ∞. t) → 0. for each t > 0. t) → g (x) as t → 0. ∇x u(x. t) − g (x)| = | ≤ | {|x−y |≥δ0 } K (x − y. Choose R > 0 suﬃcient large. t) → 0. Then by our assumption on g (x). t)(g (y ) − g (x))dy | K (x − y. as |x| → ∞.44 LI MA Then we have |u(x. t)dy 4M + 2 = . Theorem 3. This says that u(x. Fix t > 0. t) deﬁned above. . as |x| → ∞. Proof. |x| → ∞.

t)dy R n − B R ( x) ∞ ≤ supBR (x) |g (x)| + C (t)supRn |g (x)| R e−ρ 2 /4 t ρn−1 dρ → 0 + o(R) as |x| → ∞. as |x| → ∞. Using the polar coordinates (ρ. t)dy B R ( x) ≤ supBR (x) |g (x)| + supRn −BR (x) |g (x)| K (x − y. we can also show that |∇j u(x. . Hence. Here C (t) is a constant depending only on t and o(R) is as small as we like for R large. θ). we have showed that |u(x.EQUATIONS FROM MATHEMATICAL PHYSICS 45 Note that |u(x. as |x| → ∞. t)| ≤ Rn K (x − y. t)g (y )dy K (x − y. t)| → 0. t)| → 0. We omit the detail here.

R ∞ (R). t ρt (x)dx = 1 R and limt→0 ρt = δ0 . φ > . f or |x| < 1. x ∈ [−1/j. Comments on Fundamental solutions. This δx (·) is called (Rn ). The Gaussian kernel K (x − y. I would like to point out that the latter limit is more rigorous deﬁnition. For example. let ρ(x) = me − 1 1−|x|2 . qj (x) → 0 and at x = 0. 0) = δx (y ). assume that n = 1 and let 1 qj (x) = . qj (0) → ∞. If so. Then 1 ρt (x) = ρ(x/t). Here we deﬁne δx (y ) as the generalized function by < δx (y ).46 LI MA 5. One can also consider the delta function δ0 (x) as the limit of a sequence of nice functions. and for x = 0. . φ > = R qj (x)φ(x)dx 1 1/j = φ(x)dx → φ(0) 2j −1/j = < δ0 . 1/j ] and qj (x) = 0 |x| > 1/j. or for φ ∈ C0 < qj . That is. we understand it in the generalized sense that K (x − y. t) satisﬁes the heat equation for t > 0. 2j Then qj (x)dx = 1. δx is a linear functional on C0 C0 Dirac’s delta function or Dirac measure. R and let where t > 0.4. f or |x| ≥ 1 where m > 0 is a constant such that ρ(x)dx = 1. For example. For t = 0. We ask the reader use the ﬁrst concept of limit to understand delta function. we deﬁne limqj = δ0 . g (y ) >:= δx (y )(g (y )) = g (x). we consider δx (y ) as a function deﬁned on the function space ∞ ∞ (R). as j → ∞. ρ(x) = 0. Then. We may ﬁnd other sequences which have δ0 as its limit.

t). If we consider the heat heat equation on half line [0. ∞) with boundary condition u(0. t) = 0. we say K (x − y. t) + K (x + y.EQUATIONS FROM MATHEMATICAL PHYSICS 47 Generally speaking. y. t) = R+ Γ(x. Then the fundamental solution to his problem is Γ(x. For this case. t) is the fundamental solution the heat equation on Rn × R+ . We now consider the heat equation on the half line [0. ∞) with the boundary condition u(0. . t). We emphasize that we call K (x − y. and the initial condition u = φ(x). So the function deﬁned by u(x. t) = K (x − y. t) = 0. So in some sense. and we consider K (x − y. t) − K (x + y. ∞ we can also consider u as a linear functional on the space C0 (Rn ) with < u. ∞) with the boundary data ux (0. y. t)g (y )dy satisﬁes the heat equation on the half line [0. y. Here we have used the even extension to keep the boundary condition. the fundamental solution is Γ(x. t = 0. t) satisﬁes the heat equation for t > 0. With this understanding. we use the odd extension to keep the boundary condition. φ >= Rn u(x)φ(x)dx. t) is a fundamental solution to the heat equation on the whole space because K (x − y. for any given continuous function u(x) in Rn . t) = 0 and the initial value condition u = φ(x). t = 0. t) as a generalized function which is the delta function for t = 0. t) = K (x − y.

r n−1 v = 0. r c rn−1 f or n = 2. 2π and 1 Γ(x) = r 2− n . Then ∂r xi = . r and Then we have ∆u = v + Solve v + 1 (xi )2 (xi )2 u xi xi = v ( − 3 ) + v . r r r2 n−1 v. n(n − 2)|S n−1 | . Deﬁne −1 Γ(x) = log r.1. lecture six 6. Fundamental solution to Laplace equation. We ﬁnd that v = for some constant c and v = a log r + b. for some constants a and b. i Then we have rdr = i xi dxi . f or n = 2. Consider the Laplace equation (L) on Rn : ∆u = 0. A solution to the Laplace equation in a domain is called a harmonic function. and v = ar2−n + b. Choose b = 0. f or n ≥ 3. ∂xi r Let u = v (r) be a solution of (L). f or n ≥ 3. Let r ≥ 0 such that r2 = (xi )2 . Note that xi u xi = v .48 LI MA 6.

Take R > 1 large such that f = 0 outside BR . where → 0. y )f (y ))dσ. y )∇r u(y ) {|y −x|= } = lim →0 − ∇r Γ(x. Given two smooth functions deﬁned on a bounded domain D with piece-wise smooth boundary. For simplicity. y ) = Γ(x − y ) for x. We denote ν the outer unit normal to the boundary ∂D. Let u(x) = Rn Γ(x − y )f (y )dy. we need to use the second Green formula. y )(−∆u(y )) BR = lim →0 B R − B ( x) Γ(x. Here we recall the second Green formula as follows. y ) is called the fundamental solution (or Green function ) to the problem (L). If so.e. we have u(x) = < δx (y ). In fact. This function Γ(x. we can consider −∆y Γ(x − y ) = δx (y ). we assume n = 2. u(y ) > = − Rn ∆Γ(x − y )u(y )dy Γ(x − y )∆u(y )dy Rn = − = Rn Γ(x − y )f (y )dy. we get that Γ(x. Let Γ(x. . Fix x ∈ R2 . Using the second Green formula on the region BR − B (x). f is continuous in Rn and f (x) = Theorem 4. Then we have (v ∆u − u∆v )dx = D ∂D (v ∇u · ν − u∇v · ν )dσ. i. y ∈ Rn . To understand this. Then the function u satisﬁes the Poisson equation −∆u = f. u(y ) > = − < ∆Γ(x − y ). Given f ∈ C0 0 in the outside of some large ball. y )(−∆u(y )) (Γ(x. 0 (Rn ). in Rn . Proof.EQUATIONS FROM MATHEMATICAL PHYSICS 49 Here |S n−1 | is the volume of the sphere S n−1 .

To overcome this drawback.50 LI MA ∂u (y ). for each ﬁxed point x ∈ D. ·) with its boundary . y )∇r u(y )|dσ {|y −x|= } ≤ 1 2π 1 ln |∇r u(y )| r {|y −x|= } ≤ sup|∇u| |ln | → 0. Then using the second Green formula and the fact v = Γ being the fundamental solution on the whole Rn . Let’s study the Poisson equation on bounded domain. and − {|y −x|= } ∇r Γ(x. ∂r where r = |x − y and ∇r u(y ) = Note that |Γ(x. the harmonic function g (x. we introduce. Let D ⊂ Rn be a bounded domain with piece-wise smooth boundary. we get (U1): u(x) = D Γ(x − y )f (y )dy (Γ(x − y )∇u(y ) · ν − φ(y )∇Γ(x − y ) · ν )dσy . We now consider the problem (PD): −∆u = f. y )∆u(y )dy = u(x). ∂D + This expression still contains a unknown term for ∇u(y ) · ν on the boundary ∂D. y )u(y )dσ ∇r ln(r)u(y )dσ {|y −x|= } = = = 1 2π 1 2π 1 2π {|y −x|= } 1 f (y )dσ r u(y )dσ {|y −x|= } → u(x) as → 0. in D with the boundary condition u(x) = φ(x) for x ∈ ∂D. We ask our readers to verify the n ≥ 3 case. This shows that the function u satisﬁes − R2 Γ(x.

y ). y ∈ ∂D. ¯ ) satisﬁes Assume that u ∈ C 2 (D) ∩ C (D −∆u = 0. y )f (y )dy φ(y )∇G(x. x ∈ ∂D. ∂D + Set G(x. y ) · ν )dσy . y )∇u(y ) · ν − φ(y )∇g (x. Note that G(x. we obtain that u(x) = D G(x. y ) = 0. . y ) · νdσy . G(x. Let’s consider the boundary value problem for Laplace equation. Then it can be veriﬁed that the function deﬁned by u(x) = ∂D P (x. We call P (x. in D ⊂ Rn . As before. Using the expression above for v = g (x. we have u(x) = D Γ(x − y )(−∆u(y ))dy (Γ(x − y )∇u(y ) · ν − φ(y )∇Γ(x − y ) · ν )dσy ∂D + = ∂D φ(y )∇Γ(x − y ) · νdσy . y ) the Green function (or fundamental solution )for the problem (PD). ∂D − We call G(x. Adding (U1) and (U2).EQUATIONS FROM MATHEMATICAL PHYSICS 51 data g (x. for x ∈ D. we have (U2): 0 = D g (x. i. y ) = Γ(x − y ) + g (x. y )f (y )dy (g (x. y ) has zero boundary value for y . y ) the Poisson kernel to the boundary value problem in D.e. y ) = −Γ(x − y ) for y ∈ ∂D. y )φ(y )dσy . Let P (x. with the Dirichlet boundary condition u(x) = φ(x) ∈ C (∂D). y ). y ) = ∇Γ(x − y ) · ν for x ∈ D and y ∈ ∂D.

is the solution to the boundary value problem in D. L1 ) = u1 (x) and U (x.52 LI MA which is called the Poisson formula. We want to ﬁnd a harmonic function U (x. L2 ] such that U (x. y ) deﬁned on the one dimensional higher cylindrical domain D × [L1 . This function can be achieved by setting L2 − y y − L1 U (x. L 2 − L1 L 2 − L1 . L2 ) = u2 (x). Given two harmonic functions u1 and u2 on the same domain D ⊂ Rn . We now point out a nice observation about the construction of harmonic functions. y ) = u1 (x) + u2 (x).

∂Bt (p) udθ is a constant for all t ≤ R. we have u(p) = Then u(p)|∂BR (p)| = ∂BR (p) 1 |∂BR (p)| ∂BR (p) udθ. Using the mean value theorem and sending t → 0. BR (p) .EQUATIONS FROM MATHEMATICAL PHYSICS 53 6. Hence. ∂r Then we have the the right side of the identity above is tn−1 ∂B1 (p) ∂ ∂u (p + tθ)dθ = tn−1 ∂r ∂t u(p + tθ)dθ. 0 = tn−1 This implies that t1−n ∂Bt (p) ∂ 1−n (t ∂t udθ).2. where r = |x − p| and θ = x− . θ. p Note that x = p + rθ. Given a harmonic function u in a domain D. we have u(p) = Hence. If we take integration in R. ∂B1 (p) Here. It is not a composite derivative for independent variables (r. we get 0= Bt (p) ∆u = ∂Bt (p) ∂u dσ. Mean value theorem for harmonic functions. Then 1 |∂Bt (p)| udθ = ∂Bt (p) 1 |∂BR (p)| ∂BR (p) udθ. We take a ball BR (p) in the domain D. we want to point out that the notation ∂u (p + tθ) ∂r means the radial derivative of function u taking value at x = p + tθ. Using the r divergence theorem. udθ. we have 1 |BR (p)| udx.

|u(x0 ) − u(x1 ) ≤ ≤ 1 ( |BR (0)| |u|dx + B R ( x0 ) − B R ( x1 ) B R ( x1 ) − B R ( x0 ) |u|dx) 2M |BR (x1 ) − BR (x0 )| |BR (0)| → 0 . This implies that u is a constant. |BR (0)| BR (x0 )−BR (x1 ) B R ( x1 ) − B R ( x0 ) |∂xi u(x)| ≤ Hence. Then we have 1 u(p) = udθ. 1 ∂xi u(x) = ∇i u(x + y )dy |BR (0)| BR (0) 1 = νi u(x + y )dσy . Theorem 6. M |∂BR (0)| →0 |BR (0)| as R → ∞. Note that 1 u(x + y )dy. u(x) = |BR (0)| BR (0) Hence. and then. |BR (0)| ∂BR (0) where νi = xi /R. As an application we prove the following result. If u is a bounded harmonic function in the whole space Rn .54 LI MA Theorem 5. Method Two. We take a ball BR (p) in the domain D. BR (p) Mean value theorem is a basic property for harmonic functions. we have 1 ( udx − udx) u(x0 ) − u(x1 ) = |BR (0)| BR (x0 ) B R ( x1 ) 1 = ( udx − udx). Fix any two point x0 and x1 . Then for any R > 0. |∂BR (p)| ∂BR (p) and u(p) = 1 |BR (p)| udx. Method One. Given a harmonic function u in a domain D. (Liouville Theorem). Let M > 0 such that |u(x)| ≤ M for all x ∈ Rn . Then u is a constant. Proof.

.EQUATIONS FROM MATHEMATICAL PHYSICS 55 as R → ∞ (since |BR (x1 ) − BR (x0 )| ≈ Rn−1 and |BR (0)| ≈ Rn ). That is to say. u is a constant.

however. Recall that on ∂BR . This subsection is only written for readers not just for fun. i ij ∂xi ∂xi xj ∇u · ν ∇u · Xdσ + ∂BR BR ∆u∇u · X = − ∇u · ∇(∇u · X )dx.3. . we would like to show the classical monotonicity formula for harmonic functions. Monotonicity formula for Laplace equation. but also for its fundamental role in the regularity theory of nonlinear elliptic and parabolic equations. and ∇u ·∇(∇u · X )dx = BR BR ∇i u∇i (∇j uXj )dx = BR (|∇u|2 + uij ui Xj ).56 LI MA 6. u = . Recall that ∆u = 0 in the ball BR . We can not exhibit it here. Then we re-write it as ∇u · ν ∇u · Xdσ = ∂BR BR ∇u · ∇(∇u · X )dx. Let X (x) = (Xj (x)) = x − p be a radial type vector ﬁeld in BR . so is R 2− n BR |∇u|2 dx. Recently I and my students have found a monotonicity formula for solutions to a very large class of elliptic and parabolic equations. the outer unit normal is ν = (x − p)/R. To our surprise. It is easy to see that the integral |∇u|2 dx BR is a non-decreasing function of R. Why? let us prove it. It is not required for the examination for students even though it is really important for modern PDE. Note that ∇u · ν ∇u · Xdσ = R ∂BR ∂BR |∂r u|2 . We denote by ui = We now have 0=− BR ∂u ∂2u = ∇ u. Assume that u is a harmonic function in a ball BR := BR (p) with center at the point p and of radius R.

Then R 2− n BR |∇u|2 − r2−n Br |∇u|2 = 2 BR −Br |x − p|2 |∂r u|2 for all r < R. ∂BR and (n − 2) BR |∇u|2 = R ∂BR (|∇u|2 − 2|∂r u|2 )dσ. We now compute ∂R [R2−n BR |∇u|2 ] = R2−n ∂BR |∇u|2 + (2 − n)R1−n BR |∇u|2 = 2R2−n ∂BR |∂r u|2 ≥ 0. In particular.EQUATIONS FROM MATHEMATICAL PHYSICS 57 Using integration by part to the last term. we have 1 R 1 uij ui Xj = ∇j |∇u|2 Xj = |∇u|2 − 2 2 ∂BR 2 BR Then we have ∇u · ∇(∇u · X )dx = BR n|∇u|2 . Monotonicity formulas play a very important role in the regularity theory of weak solutions to nonlinear parabolic and elliptic partial diﬀerential equations/ systems. . More precisely. This implies that the function R 2− n BR |∇u|2 is a monotone non-decreasing function of R. 2−n 2 |∇u|2 + BR R 2 |∇u|2 . Theorem 7. the function r 2− n Br |∇u|2 is a monotone non-decreasing function of r. we have proved the following result. Assume that u is a harmonic function in a ball BR .

where a. From the deﬁnition. In this section.1. lecture seven From the previous sections. we get k L(y (k ) )(s) = s L(y )(s) − j =1 k y (j −1) (0)sk−j . We deﬁne. The famous Laplace and Fourier transforms are for this purpose. in particular. f or x > 0 then the Laplace transform for y is well deﬁned. To handle PDE on unbounded domains. let Y (s) = L(y )(s). people try to use integration type of transformations to reduce ODE into algebraic equations and to reduce linear PDE into to ODE. we know that solving a PDE is not easy matter. ∞). we have L(y1 + ay2 ) = L(y1 ) + aL(y2 ) for every real number a ∈ R. + as + b . PDE on whole space. the Laplace transform by ∞ L(y )(s) = 0 y (t)e−st dt. we always assume that our functions have a good decay at inﬁnity such that the integrals converge. b are constants and g (t) ∈ C 0 [0. Then we have s2 Y + asY + bY = L(g ). Remark: It is clear that if there is some constant M > 0 such that |y (x)| ≤ eM x . In fact. Note that ∞ L(y )(s) = 0 y (t)e−st dt = −y (0) + sL(y )(s). for a function y = y (t) ∈ C 0 [0. We can use this property to solve the following ODE: y + ay + by = g (t). with initial conditions y (0) = 0 = y (0). 7.58 LI MA 7. ∞). By induction. we have Y = s2 L(g ) . Laplace transforms. Hence.

1 ∞ n −st L(y )(s) = − t e d(−st) s 0 ∞ 1 tn−1 e−st dt] = [n s 0 = . Given −1 y (t) = tn . Example. So we shall invoke to special book on Laplace transforms.EQUATIONS FROM MATHEMATICAL PHYSICS 59 Denote by L−1 the inverse of the Laplace transform.. Then we have L(y )(s) = 0 ∞ tn e−st dt. We do the computation step by step. s + as + b In general. We remark that. which is a complex variable in the plane.. We now compute one example to close this part. Then we have L(g ) y (t) = L−1 [ 2 ]. s . we have a list of Laplace transforms of special functions to use. using complex analysis and inverse Fourier transform (see next section). n! = n+1 . it can be proved that a+i∞ 1 L (Y )(t) = Y (s)est ds 2πi a−i∞ where a is the real part of s = a + ib.

1 gA (ξ ) = √ 2π Note that i 2 ξ2 . 2A 2 π . that is. −Ax − ixξ = −A(x + ξ) − 2A 4A2 2 e−Ax e−ixξ dx. Deﬁne ˆ(ξ ) = √1 f f (x)e−ixξ dx. Then the Schwartz space is the space of Schwartz function ∞ (R) is a subspace of S. |f (x)|dx < ∞. A . that is. 2 and e−Ax dx = Then we have e− 4A2 gA (ξ ) = √ 2π ξ2 ξ2 1 2 e−Ax dx = √ e− 4A2 . 2π which is called the Fourier transform (in short.2. n. Deﬁne the Fourier transform of a Schwartz function. Example 1: Let’s compute the Fourier transform of the function gA (x) above. R. |x|m |f (n) (x)| → 0 as x → ∞. Fourier transforms. Rn From the deﬁnition of the F-transform. Let f be a smooth function in S. Set F (f ) = f Remark: From the deﬁnition formula. We ﬁrst introduce the Schwartz space S. F : S → S.60 LI MA 7. The important example is on R. We say f is a Schwartz function on R if f is smooth and for any non-negative integers m. Note that C0 that the function 2 gA (x) = e−Ax is in the space of Schwartz space. We denote g (x) = g1 (x). we can see that the Schwartz space is the invariant set of F-transform. we see that the Fourier transform makes sense for any function f ∈ L1 (Rn ). By deﬁnition. F-transform) of f in ˆ. where A > 0 is a constant.

1 . for (δλ g )(x) = e−λ we have ξ2 1 δλ g (ξ ) = λ−1 g ˆ(ξ/λ) = √ e− 4λ2 .EQUATIONS FROM MATHEMATICAL PHYSICS 61 From this we can see that. gA is the eigenfunction of the 4 F-transform. f or |x| > 1/j. and qj (x) = 0. 2j 1 sin jξ e−ixξ dx = √ . t)](ξ ) = e−a We shall use this expression. Deﬁne their convolution by f g (x) = f (x − y )g (y )dy. 2 − x2 1 1 √ 4a t . (δλ f )(ξ ) = λ−1 f Hence. (δλ f )(x) = f (λx). 2π Given h. Example 2: Let qj (x) = Then 1 qj (ξ ) = √ 2 2πj Hence. t) = a 2t F [g (x. 2λ 2 x2 2 ξ2 t . f or ξ = 0 as j → ∞. for A3 = 1 . Then by deﬁnition. qj (ξ ) → 0. and 1 qj (0) = √ . 2π jξ −1/j 1/j . Given two functions f and g . let A = 4a e 2 t and let g (x. we have ˆ(ξ ) (τh f )(ξ ) = e−i(hξ) f and ˆ(ξ/λ). . f or |x| ≤ 1/j. Let (τh f )(x) = f (x − h). Then For t > 0. 0 = λ ∈ R.

62 LI MA Then we have F (f 1 g (x)) = √ f g (x)e−ixξ dx 2π 1 = =√ e−ixξ dx f (x − y )g (y )dy 2π 1 = √ g (y )e−iyξ dy f (x − y )e−i(x−y)ξ dx 2π √ ˆ· g = 2π f ˆ. In fact. By an elementary computation. note that gA (x) → 1. we have −1 ∂x ∂x f = f. . u ˆ(y )g (y )dy. f Proof. which is one of the beautiful theory of modern mathematics. In short we write it by F (f ). −1 This property gives us a chance to deﬁne the inverse operator ∂x of ∂x by the deﬁnition − ˆ1 f (ξ ) = (iξ )−1 f ˆ(ξ ). The most important property is that the F-transform of the derivative fx is a multiplication: ˆ ˆ f x (ξ ) = iξ f (ξ ). ˆ ˇ= f f and ˇ ˆ = f. we have the following symmetry formula : u(x)ˆ g (x)dx = Deﬁne ˇ(ξ ) = √1 f f (x)eixξ dx. ∂x Clearly. we have Proposition 8. Let’s verify the latter equality. Let’s going back. 2π which is called the inverse Fourier transform of f . Remark: One may see more interesting properties about Fourier transforms in the subject called ”Harmonic/Fourier Analysis ”. By an elementary computation and the symmetry property.

Using the symmetry formula.EQUATIONS FROM MATHEMATICAL PHYSICS 63 as A → 0 and √ ξ2 1 gA (ξ ) = √ e− 4A2 → 2πδ0 . Then we have |u ˆ|2 = u ˆu ˆ= u ˆg = ug ˆ= |u|2 . One can deﬁne n-dimensional Fourier transform of f in Rn by ˆ(ξ ) = √ 1 f ( 2π )n f (x)e−ix·ξ dx. by symmetry property. f Using this we can prove Theorem 9. Then. Then 1 f (x) = f1 (0) = √ 2π 1 f1 (ξ )dξ = √ 2π ˆeixξ dξ. In fact. we let f1 (y ) = f (y + x). (Plancherel Theorem): Given u ∈ S. 2A as A → 0. So u ¯=g ˆ. we have that 1 u(x) = √ 2π u ˆ(y )e−ixy dy = u ˆ(x). using the inverse Fourier transform. ˇ ˆ(0) = √1 ˆ(ξ )dξ f f 2π 1 ˆ(ξ )dξ = √ limA→0 gA (ξ )f 2π 1 = √ limA→0 gA (y )f (y )dy 2π = f (0). Let g = u ˆ in the above equation. In general. . we have |u|2 dx = |u ˆ|2 dξ Proof.

n ( 2π ) Similar properties in dimension n are also true for n-dimensional Fourier transforms. Plancherel Theorem in Rn is true. The inverse Fourier transform of f in Rn is deﬁned by ˇ(ξ ) = √ 1 f f (x)eix·ξ dx. . In particular.64 LI MA Here x · ξ is the inner product in Rn . we have the spectrum property: A[eixξ ] = −|ξ |2 eixξ . One important fact lying behind the Fourier transform is that for the linear operator A[u] = u (x).

. The latter condition implies that u = Aeλx 2 2 /2 x2 u2 )( |u |2 ). Proof. The inequality above is the so called Heisenberg uncertainty principle. we have 1 u2 ≤ ( x2 u2 )1/2 ( |u |2 )1/2 . Integrating by part. as a byproduct. 4 Using the Cauchy-Schwartz inequality we have (H): 1 ≤ 4( with equality at u = λxu. where β > 0 and A4 = 2β/π . For any smooth function u (maybe complex-valued) vanishing at inﬁnity with |u|2 = 1. we then get what wanted. Using u = 1 we have λ = −2β and the relation between A and β . Note that |u |2 = ξ 2 |u ˆ|2 . . Inserting this into (H).EQUATIONS FROM MATHEMATICAL PHYSICS 65 7. we have 1= |u|2 = −2 xu · u ≤ 2 |x||u||u |.3. There are a lot of applications of Fourier transforms in the various branches in sciences and technologies. Notice that. 2 1 ξ 2 |u ˆ|2 dξ ) ≥ . 2 1 for every u ∈ C0 (R). I hope our readers can read more books on Fourier transforms and Fourier analysis. Heisenberg uncertainty principle. we have ( x2 |u(x)|2 dx)( with its equality true only at u(x) = Ae−βx . Please read as more as you can.

ut (x. 0) Set u(x. t).66 LI MA 8. lecture eight 8. 0) cos(|ξ |t) + ut (ξ. |ξ | Using the inverse F-transform we ﬁnd the solution expression: u(x.1. 0) = f (x). |ξ | 2i|ξ | we have 1 u(x. Application of Fourier transforms. − (ξ ) = (f (ξ ) − 2 i|ξ | Using the similar approach. t) = √ (f ˆ(ξ ) )e dξ. t) = √ ( 2π )n where i(ξx+|ξ |t) i(ξx−|ξ |t) ˆ ˆ (f +f )dξ. The solution can be written as 1 sin(|ξ |t) iξx ˆ(ξ ) cos(|ξ |t) + g u(x. let us use the F-transform to solve the wave equation utt = ∆u. |ξ | sin(|ξ |t) ). we get u ˆ=u ˆ(ξ. − (ξ )e + (ξ )e 1 ˆ 1 ˆ g ˆ(ξ )) f + (ξ ) = (f (ξ ) + 2 i|ξ | 1 ˆ 1 ˆ f g ˆ(ξ )). t) = −|ξ |2 u ˆ(ξ. Solving this ODE. Then we have u ˆtt (ξ. 0) = g (x). 0) sin(|ξ |t) . 0) cos(|ξ |t) + ut (ξ. First. t) = F (ˆ u(ξ. we can use properties of F-transform to solve the heat equation ut = ∆u. on Rn × R+ and . n |ξ | ( 2π ) Using the relations 1 cos(|ξ |t) = (ei|ξ|t + e−i|ξ|t ) 2 and sin(|ξ |t) 1 = (ei|ξ|t − e−i|ξ|t ). on Rn × R+ Take F-transform for the variable x on both sides.

This shows that u = f Ht is the solution to the heat equation with initial data u(x. 0)exp(−|ξ |2 t)). t). t) = −|ξ |2 u ˆ(ξ. t) = f ˆt (ξ. Let gt (x) := g (x. t) = √ ( 2π )n Let 1 2 Ht (x) := H (x. (2t)n/2 . and u Hence. t). t) = Then g ˆt (ξ ) = exp(−|ξ |2 t). u ˆ(ξ. t) = e−|x| /4t . Using the inverse F-transform we have u(x. ˆexp(−|ξ |2 t). This is a ﬁrst order ode. and we can solve it to get u ˆ=u ˆ(ξ. Given the initial data u(x. 0)exp(−|ξ |2 t). t) = F (ˆ u(ξ. u(x. n/ 2 (4πt) Then u(x. 0) = f (x). t) = −|ξ |2 u ˆ(ξ. Then we have u ˆt (ξ.EQUATIONS FROM MATHEMATICAL PHYSICS 67 Take F-transform for the variable x on both sides. Let 1 f gt (x). t) = f Ht (x). 1 2 e−|x| /4t . 0) = f (x) in Rn .

68

LI MA

8.2. Laplace equation in half space. Our problem is to solve the Laplace equation ∆u = uxx + uyy = 0 on the half space R2 + = {(x, y ); y > 0} with boundary condition u(x, 0) = f (x). To use the F-transform method, we think of uy (x, 0) = 0. Then we have −|ξ |2 u ˆ(ξ, y ) + u ˆyy (ξ, y ) = 0. Solve this and we ﬁnd ˆ(ξ )exp(−|ξ |y ). u ˆ(ξ, y ) = u ˆ(ξ, 0)exp(−|ξ |y ) = f Deﬁne Py (x) = F (exp(−| · |y ))(x). Then by an elementary computation, we have Py (x) = F (exp(−| · |y ))(x) 1 = √ e−|ξ|y+ixξ dξ 2π ∞ 1 = √ e−yξ+ixξ dξ 2π 0 0 1 eyξ+ixξ dξ + √ 2π −∞ 1 1 1 = √ ( + ). 2π y + ix y − ix This implies that 1 2y Py (x) = √ . 2 2π x + y 2 Hence 1 u ˆ = √ f Py , 2π and 1 u(x, y ) = √ f Py (x) 2π is the solution wanted for the problem above.

EQUATIONS FROM MATHEMATICAL PHYSICS

69

8.3. Derivative bound. We ask the following question. Given a smooth function u ∈ S(Rn ). Assume the bound M0 of |u| and the bound M2 of |∆u| on Rn . Can we give a bound of |∇u| on Rn in term of M0 and M2 ? We answer this question below. Let −∆u + u = f, in Rn . Using the Fourier transform, we get ˆ. (|ξ |2 + 1)ˆ u=f Then we have u ˆ= Deﬁne 1 ˆ. f |ξ |2 + 1

1 1 B2 (x) = √ F ( 2 )(x). |ξ | + 1 ( 2π )n u(x) = B2 f (x)

Then we have and |∇u(x)| = |∇B2 f (x)| ≤ |∇B2 | |f |(x). Using the knowledge from calculus we can show that |∇B2 |(y )dy := B < ∞.

Rn

Notice that |f (y )| ≤ max|u| + max|∆u|, where max means taking maximum value over the domain Rn . Hence, we have |∇u(x)| ≤ B (max|u| + max|∆u|). Using the function u(rx) (with r > 0 to be determined) to replace u in the inequality above, we have r|∇u(rx)| ≤ B (max|u| + r2 max|∆u|). Since r > 0 can be any positive number, we get |∇u(rx)|2 ≤ 4B 2 max|u|max|∆u|. Therefore, max|∇u| ≤ 2B max|u|max|∆u|.

70

LI MA

9. lecture nine 9.1. Maximum principle (MP) for Laplace operator. The Maximum Principle (MP) is a beautiful property for solutions to Laplace equations and heat equations. One can use the Maximum principle to get uniform bound for a solution and its derivatives. Consider a smooth solution u to the following elliptic partial diﬀerential inequality in a bounded domain D in the plane R2 (IPDE): uxx + uyy > 0. Note that there is no maximum point of u in the interior of the domain D. For otherwise, assume (x0 , y0 ) is and then we have ∇u(x0 , y0 ) = 0, DT DT u(x0 , y0 ) ≤ 0, for any direction T ∈ R2 . The latter condition implies that the matrix D2 u(x0 , y0 ) is non-positive, that is, D2 u(x0 , y0 ) ≤ 0. Hence, (uxx + uyy )(x0 , y0 ) ≤ 0, which contradicts to (IPDE). Therefore, we have supD u = sup∂D u. Assume that we only have uxx + uyy ≥ 0 in the domain D. Then we choose w = u + eβx for small positive constant and large positive constant β . Note that ∆eβx > 0. Then we have ∆w > 0. Hence, we have supD w = sup∂D w. Sending →, we get that supD u = sup∂D u. Clearly, the above argument is true for general domains in Rn . This is the maximum principle for elliptic pde of second order.

Assume u is a smooth function satisfying ∆u ≥ 0. y0 ). y0 ) in the interior of D. It is a basic tool in the study of these kinds of pde’s. can not reaches its maximum in the interior of the domain D. The Maximum principle is true for many nonlinear elliptic and parabolic partial diﬀerential equations of second order. Remark: (1). y0 ) ≥ 0. in D. we have uxx uyy (x0 . a contradiction! This implies that supD u = sup∂D u. Hence. uyy ≤ 0 at (x0 .EQUATIONS FROM MATHEMATICAL PHYSICS 71 Theorem 10. It is clear from our argument that the function u ∈ ¯ ) satisfying C 2 (D) ∩ C (D uxx uyy < 0. Then we have supD u = sup∂D u. In fact. in D. then we have uxx ≤ 0. (2). . The same argument applies to the diﬀerential inequality: det(D2 u) < 0. in D. Maximum Principle. if it has the maximum point at (x0 .

Assume that there is an interior point (x0 . and let ¯ . t) − t. t) = max∂p Q u(x. Note that maxQ ¯ u(x. x ∈ ∂D. This Then we have at the point (x0 . Then Lv (x. t) ∈ Q. v (x. t0 ).1 (Q) is the space of continuous functions which have continuous x-derivatives up to second order and continuous t-derivative. MP for heat equations 1. t) − < 0. Proof. a contradiction. and ut ≥ 0. t) and max∂p Q v (x. t). Assume that D is a bounded smooth domain in Rn . t) = max∂p Q v (x. Assume that u ∈ C 2. Here on Q. t) ∈ Q be the parabolic boundary of the domain Q. t) ≤ 0. Hence. We now consider the heat equation (HE): Lu := ut − a2 ∆u = f (x. . t) < 0 in Q. t) = Lu(x. As before. t). For general case when Lu(x. t) ≤ maxQ ¯ v (x. t) ¯ ) is a solution to (HE).2. t) = u(x. ¯ ) satisﬁes Lu ≤ 0 on Theorem 11. t) + T. t). Then we have maxQ ¯ u(x. t0 ) ≤ 0 and Lu(x0 . t). (x. we let. for > 0. Then we can use the result above to v (x. t0 ) ≥ 0. ∇u = 0. We assume that u ∈ C 2. x ∈ D. t0 ) = maxQ ¯ u(x.1 (Q) ∩ C (Q Q. we let Q = {(x. u can not have an interior maximum point and then maxQ ¯ u(x. (x.72 LI MA 9. t) ∈ Q. 0 < t ≤ T }. t) ≤ max∂p Q u(x. Dx implies that ∆u(x0 . 2 u ≤ 0. or t = 0} ∂p Q = {(x. t). t) = max∂p Q u(x. t0 ) ∈ Q such that u(x0 . We ﬁrst assume that Lu(x. t) to get maxQ ¯ v (x.1 (Q) ∩ C (Q C 2.

¯ . t)|. Assume that 2 . t) ≥ max∂p Q u(x. Consider the function w(x. t) ∈ Q. t) ≥ 0. Let’s consider an easy application of the maximum principle.1 ¯ u ∈ C (Q) ∩ C (Q) satisﬁes Lu = f (x. maxQ ¯ u(x. t). t) on ∂p Q.EQUATIONS FROM MATHEMATICAL PHYSICS 73 Sending → 0. we have However. ¯ ) with F := maxQ Theorem 12. t) = F t + B ± u(x. which implies the conclusion wanted. t) ∈ Q with u(x. t). t). t) deﬁned by w(x. By the Maximum principle above. We are done. t) = max∂p Q u(x. on Q and on ∂p Q. (x. we have w(x. t). we get maxQ ¯ u(x. Then we have maxQ ¯ u(x. since ∂p Q ⊂ Q maxQ ¯ u(x. Proof. t) ≤ max∂p Q u(x. t) = F T + max∂p Q |φ(x. t) = φ(x. Hence. t)|. w ≥ 0. t). Denote by B = max∂p Q |φ(x. Then Lw = F ± f ≥ 0. . Given f ∈ C (Q ¯ |f |. (x.

we have (M (t) − δ ) ≤ inf limJk (t)1/2k ≤ suplimJk (t)1/2k ≤ M (t). t) be a non-trivial smooth ¯ ) to the heat equation (HE): solution in C (D ut = ∆u. ¯ . max |u(x. we get limk→+∞ Jk (t) ≥ M (t). Let u = u(x. Theorem 13. M (t) − δ ≤ |u(x. So we only need to show that M (t) ≤ inf limJk (t)1/2k . . 0) = g (x). For ﬁxed t.3.74 LI MA 9. Since δ > 0 can be any small constant. t) = 0. Consider a new function deﬁned by Jk (t) = D |u|2k dx. M (t) := supx∈D |u(x. Proof. We remark that the last inequality is also true for any bounded continuous function deﬁned on a unbounded domain D. we note that inf limJk (t)1/2k ≤ suplimJk (t)1/2k ≤ M (t). k→+∞ First. ∂D × [0. t)| = lim Jk (t)1/2k . we In fact. u(x. for each t. Recall that. on D × [0. x ∈ D. t)| Hence (M (t) − δ )|Q|1/2k ≤ Jk (t)1/2k ≤ M (t)|D|1/2k Sending k → +∞. We show in this subsection that the maximum principle cab be proved to be true for the heat equation by another method. t)| x∈ D is a monotone decreasing function of t. MP for heat equations 2. T ) where D is a bounded piece-wise smooth domain in Rn . since u is a bounded continuous function in D ¯ have a sub-domain Q ⊂ D such that on Q . Given a bounded continuous function g (x) on the smooth bounded domain D ⊂ Rn . T ) with initial and boundary conditions u(x. Then.

Then. With an easy modiﬁcation. max |u(x. We omit the detail. Note that for any bounded continuous function f deﬁned on a unbounded domain D with D |f |p dx < ∞ for some p > 0. we have dJk = −2k (2k − 1) u2k−2 |∇u|2 < 0. which implies that M (t) ≤ M (0)e2at . The same conclusion as in the theorem above is also true. we only need to prove that suplimk→∞ ( D |f |k dx)1/k ≤ M. k→∞ D According to the remark made in the proof of the theorem above. Remark 1. dt Hence Jk (t) is a monotone decreasing function of t. If we assume u ≥ 0 in Q satisﬁes ut ≤ ∆u. we still have M := supD |f (x)| = lim ( |f |k dx)1/k . Remark 2. Therefore. we can extend the above theorem into unbounded domain D. arguing as before. If we assume u ≥ 0 in Q satisﬁes ut − a2 ∆u ≤ Au.EQUATIONS FROM MATHEMATICAL PHYSICS 75 Note that dJk = 2k u2k−1 ut = 2k u2k−1 ∆u. it is almost trivial since suplimk→∞ ( D |f |k dx)1/k ≤ suplimk→∞ (M k−p D |f |p dx)1/k = M. However. t)| x∈ D is a monotone decreasing function of t too. Remark 3. in Q. in Q. we have Jk (t) ≤ Jk (0)e2kAt . . Given a constant A > 0. dt By using integration by part.

we have w + tξ ∈ A for every t ∈ R. by the variational lemma (see our textbook [1]). [|∇w|2 + 2t∇w · ∇ξ + t2 |∇ξ |2 ]. dt Assume that w ∈ C 2 . 1 Since ξ can be any testing function in C0 (D). lecture ten 10. for all t ∈ R. we get the Euler-Lagrange equation for the functional I (·). which can not be presented here. Variational principle. Variational principle is a important tool in sciences.76 LI MA 10. that ∆w = 0. x ∈ ∂D. We denote A by such a class of functions. dt Note that I (w + tξ ) = Hence. We can only touch the the concept of variational structure. Then for every ξ ∈ C0 (D). Then using integration by part. We introduce the Dirichlet energy (so called the variational structure) I (u) = D |∇u|2 dx. We look for a function w ∈ A such that I (w) = infu∈A I (u). we have (∆wξ ) = 0. We discuss the variational method for Laplace equation in a region of dimension two. One of the beautiful result in the calculus of variations is the Noether principle. Given a bounded domain D ⊂ Rn . d I (w + tξ )|t=0 = 2 ∇w · ∇ξ = 0. .1. I (w) ≤ I (w + tξ ) and then d I (w + tξ )|t=0 = 0. for any smooth function u : D → R with ﬁxed boundary value u(x) = φ(x). in D. 1 Assume such a function w exists. So we have.

It will be really helpful for readers to know some knowledge about Sobolev spaces. in L2 (I ) is deﬁned by a Cauchy sequence {fk } ⊂ C0 |fk − fj | → 0 as k. h) = I fx hx dx. and (fk )x → g in L . It can be showed that the strong derivative is the same weak derivative. and H0 (I ) is a Hilbert space. Furthermore. If this is true. For f. the element f (which will be called a L2 -function) ∞ (I ). Hence. i. We consider L2 (I ) as the completion of ∞ the space C0 (I ) with respect to the norm |f | = ( I |f (x)|2 dx)1/2 . which is a beautiful part of the course Functional Analysis. Here are some simple facts about it. Let I = [a. Then is easy to see that the integration by part formula f φx dx = − I I 2 gφdx. ∞ (I ). In the other word. Remark: We mention that the existence of the inﬁmum w can be proved by using the Hilbert space theory. 1 1 Then one can show that (·. This is the famous Dirichlet problem in the history. j → ∞. we often call g the strong derivative of f and denote by g = fx .e. h ∈ H0 (I ). We say that the element f ∈ L2 (I ) has derivative ∞ (I ) such that g ∈ L2 (I ) if there is a Cauchy sequence {fk } ⊂ C0 fk → f. we deﬁne (f. .EQUATIONS FROM MATHEMATICAL PHYSICS 77 with the boundary condition w = φ on ∂D.. ·) is an inner product on H0 (I ). we simply call g the derivative of f . is true. b]. 1 We denote by H0 (I ) the Sobolev space which consists of L2 functions 1 with derivatives. one can show that both deﬁnitions are equivalent each other. The recent study is about the obstacle problem for harmonic functions. People often use this formula to deﬁne the for all φ ∈ C0 weak derivative of f .

We recall the Fixed Point Theorem: Let E be a Banach space and let B ⊂ E be a closed convex subset. We study the evolution problem of type ut = ∆u + F (u) or utt = ∆u + F (u) in R with nice initial data. Some comments on Nonlinear evolution problem.78 LI MA 10. One often use Fixed Point Theorem to get the local existence of solutions to non-linear evolution problem. to make f : B → B . i. It is easy to see that the solution is a u(t) = . we can reduce the problem into the form: n t u = S (t)(u0 ) + 0 S (t − τ )F (u)dτ := f (u). we need to take E to be deﬁned by the energy method such that E is a Banach algebra. In some sense.e.e.2.e. We show some simple examples below. i. The interesting feature in non-linear problems is that there is a blow up phenomenon. with the initial data u(0) = a > 0. Let f : B → B be contraction mapping. 1) such that |f (x) − f (y )| ≤ κ|x − y | for all x. i. Then we can use the interpolation inequality method to show that f is in fact a contraction mapping. f (x ) = x .. Then there is a ﬁxed point x ∈ B of f .. B is an invariant set of f and f is a contraction. Let B = BR (S (t)(u0 )). Example One: Consider the ODE ut = u2 . 1) such that |f (x) − f (y )| ≤ κ|x − y |. there exists a positive constant κ ∈ (0. the mapping f satisﬁes f : B → B. Let’s see a little more. Here | · | is the norm of E . We have to show that for small R and T . 1 − ta . Using the Duhamel principle. Generally speaking. We assume in this section that readers know the Banach ﬁxed point theorem. we try to give a few comments about recent research directions in the study of non-linear PDE. Here S (t)(u0 ) is the solution to the problem with F = 0. y ∈ B .. there is a positive constant κ ∈ (0. that is.

t) or p(t) := ρ(x)u(x. Example Three: Given a constant β ≥ 1 . T ) with the non-trivial initial data u(x. 2π ]. 0) = φ(x) > 0 and the boundary data u(x. This gives us that . t) for x ∈ [0. t) = u(x + 2π. x ∈ [0. 2π ] and the boundary data u(0.EQUATIONS FROM MATHEMATICAL PHYSICS + 79 1 which goes to inﬁnity as t → a . 2π ]. (x. t) ∈ [0. a 1 − ta for a = minx∈[0. T ) with the non-trivial initial data u(x. t) > 0. which makes p(t) go to negative as t → +∞. where k is an integer. (x. we have u(x. 2π ] × [0. t) ∈ [0. t). we can study the blow up property by introducing some monotone or concave function A(t) deﬁned by the integration like p(t) = minD u(x. Then we have p (t) ≥ p(t)2 . t)|k . For a non-linear evolution PDE. We consider the following 4 nonlinear parabolic equation p(t) ≥ ut = ∆u + u2 + βu.2π] φ(x) > 0 and u has to blow up to +∞ in ﬁnite time before a−1 . t) for x ∈ [0. We can show that p(t) will satisﬁes some diﬀerential inequality. By the Maximum principle. This is the blow up property of the ODE. 2π ] × {t > 0}. 0) = φ(x) > 0. This gives us the blow up property. This is absurd since p(t) > 0 for all t. 2π ] × [0. Let p(t) = minx∈[0. ρ(x) > 0 is a solution of some elliptic equation. t) = 0 = u(2π.2π] u(x. t)k or p(t) := ρ(x)|∇x u(x. x ∈ [0. Example Two: We consider the following nonlinear parabolic equation ut = ∆u + u2 .

t)dx. we have used similar idea to prove the long standing conjecture on non-linear Schrodinger equation on compact Riemannian manifolds. 2 Let p(t) = C 0 2π Then. . Very recently. we have p x sin( )ut 2 0 2π 2π x x sin( )∆u + C sin( )u2 + βp(t) = C 2 2 0 0 2π 1 x = C sin( )u2 + (β − )p(t) 2 4 0 2 ≥ p(t) . 2 x sin( )u(x.80 LI MA 1 4 Choose C = such that 2π C 0 x sin( ) = 1. = C 2π This gives us the blow up as before in ﬁnite time.

we have dx dy ( )2 + ( )2 = 1... dt2 This says that the curve is a straight line. Then the PDE says that the normal vector (∇u. . dt dt This implies that t is the arc-length parameter for the curve. B (x. uy = b. This is a fun model for non-linear pde of ﬁrst order. Introduce the curve (x(t). with a2 + b2 = 1. = uy . d2 x dx dy = uxx + uxy = 0. −1) of the graph is orthogonal to the vector ﬁeld (A1 (x. a2 + b2 = 1. we have ux uxx + uy uyx = 0. We consider the ﬁrst order PDE of the following form (FPDE): Ai (x. So u = ax + by + C. 2 dt dt dt Similarly. u)uxi = B (x. z ). u). which is called the eikonal equation. Then we have the simple equation: ux = a. y (t)) by dx dy = ux . z )). Hence along the curve. Let us ﬁrst consider a nice example: 2 u2 x + uy = 1. i We look at the graph z = u(x). dt dt Then by the equation. z ).3. .. we have d2 y = 0. Nonlinear PDE of ﬁrst order. y (t) = bt + d. x(t) = at + c.EQUATIONS FROM MATHEMATICAL PHYSICS 81 10. Doing diﬀerentiation to the eikonal equation. saying. An (x. We now give some material for ambitious readers (who may see [2] for more).

82 LI MA This suggests us that the characteristic curve (x(t).Hilbert and R. z (s)). For more. s ∈ Γ. If we solve the equation x(s. Courant [2]. please look at the great book of D. t). dt dt If we can solve this system starting from a given surface (x(s). . z ). t). then u = z (s(x). z (t)) should satisfy dxi dz = Ai (x. t) = x for (s = x(x). z = z (s. of dimension n − 1 to get a solution x = x(s. z ). t(x)) is the solution to (FPDE). t = t(x)). = B (x.

for every ξ ∈ C0 Proof. We have three parts. 0≤r< ∞ (D). Let f ∈ C 0 (D) for a bounded domain in Rn . u satisﬁes −u = u(t)2 + g (t). Note that ρ(|x − x0 |) ∈ C0 f (x)ρ(|x − x0 |) > 0. Assume that f (x)ξ (x)dx = 0 D ∞ (D). by deﬁnition. the lemma is true. in B (x0 ). One is for the variational lemma. In fact. L]. 2 3 where g (t) is a given continuous function on [0. 11. Therefore. L]. Then.1. Assume that u = u(t) ∈ C0 L F (u) = 0 1 1 ( |u |2 − u3 − g (t)u(t))dt. Another is about the method of separation of variables. Appendix A One may consider this section as a last one to our course. This gives a contradiction to our assumption of our lemma. 2 [0. Our readers may do research for these problems in the future. φ >= 0 (u φ − u2 φ − g )dt . Then we may assume that f (x0 ) > 0 and there is a small ball B (x0 ) such that f (x) = 0 for every x ∈ B (x0 ). Lemma 14. Then f = 0. The other is about some open problems in PDE’s. Assume that there is a point x0 ∈ D such that f (x0 ) = 0. Take ξ (x) = ρ(|x − x0 |) where ρ(r) = e − 1 1−r 2 . Hence. L 0 =< δF (u). f (x)ξ (x)dx = D ∞ (D ) C0 f (x)ρ(|x − x0 |) > 0. Variational lemma.EQUATIONS FROM MATHEMATICAL PHYSICS 83 11. Deﬁne Example. if δF (u) = 0. We argue by contradiction. and and ρ(r) = 0 for all r ≥ .

. L). By variational lemma. Integrating by part. which is equivalent to our equation above. we get for all φ ∈ C0 L 0= 0 (−u − u2 − g )φdt. we know that u + u2 + g = 0.84 LI MA ∞ (0.

In the method of separation of variables. In the old time before Fourier. a2 X (L) + b2 X (L) = 0. 0 < x < L. Sturm-Liouville theory. we meet the problem if we can use the eigenfunctions of the following eigenvalue problem (p(x)X ) + [λ + q (x)]X = 0. the answer is yes too.EQUATIONS FROM MATHEMATICAL PHYSICS 85 11. 0 0 |Xn |2 dx = 1 for λn = λn . where the constants satisfy ai ≥ 0. L] with the ﬁxed boundary condition X (0) = a. bi ≥ 0 and ai + bi = 0. Luckily. a2 f (L) + b2 f (L) = 0. such that every smooth function f with the boundary condition −a1 f (0) + b1 f (0) = 0). and the functions p(x) > 0 and q (x) are continuous on [0. can be expressed as the expansion of trigonometric functions. can be expressed as Cn Xn (x). −a1 X (0) + b1 X (0) = 0). X (L) = b. which is deﬁned on the interval [0. people often asked if any continuous function. Roughly speaking. L]. the Sturm-Liouville theory says that there are an increasing eigenvalues λn → ∞ and eigenfunctions Xn (corresponding to λn ) with L L Xn Xm dx = 0.2. f (x) = n . This is the Sturm-Liouville theory. The answer is yes since we know the beautiful theory of Fourier series.

A question from Yanyan Li and L. I and my student D.Z.Z. then we have |u (0)| ≤ for M ≥ 2u(0) . and xn →±∞ lim u(x . R]. −∆v = up . De Giorgi proposed the following conjecture: If u is a solution of the scalar Ginzburg-Landau equation: ∆u + u(1 − u2 ) = 0. p+1 q+1 n We have some progress in this problem.Du and Li Ma. then all level set of u are hyper-planes.H. if a nonnegative function satisfying |u (x)| ≤ M on [−R. Caﬀarelli. v ) to the following elliptic system on Rn : −∆u = v q . at least for n ≤ 8.Nirenberg It is not hard to show by using the Taylor expansion that. Proc. who is a PhD student of L. on Rn such that |u| ≤ 1 and ∂ u ∂xn > 0 on Rn . xn ) = ±1. 2 Serrin conjecture In the past years. Please see our paper Li Ma and D. For more reference. R2 .86 LI MA 11. De Giorgi conjecture: In 1978.Chen. where the indices p > 0. 131(2002)2415-2422.3. 3.Chen studied the following conjecture of Serrin: There is no bounded positive solution (u. The case when n ≤ 8 was solved recently by Savin. CPAA. Some remarks related to the De Giorgi conjecture. R2 2u(0)M and |u (0)| ≤ u(0) R + M R 2 for M < 2u(0) . Three open problems. 1. A Liouville type Theorem for an integral System. 5(2006)855-859. AMS. for any x ∈ Rn−1 . please see our paper: Y. q > 0 satisfy 1 1 2 + >1− .

BR Then there is a constant C depending only on n such that |∇u(x)| ≤ C if R ≥ u(0) M u(0)M ≥ 2|x|.EQUATIONS FROM MATHEMATICAL PHYSICS 87 Yanyan Li and L.Nirenberg (2004) asked what is the best constant C in the following result in the ball BR ⊂ Rn : Assume that u is a non-negative C 2 function in closure of the ball BR satisfying: max |∆u| = M. . |∇u(x)| ≤ C ( u(0) + RM ) R (0) . if ≥ 2|x| ≥ R < uM This is an open question.

Then u satisﬁes utt = a2 uxx . 12. Homework 1. ] and u(x. Assume that the vibrating string of length L moves with two end points ﬁxed at x = 0 and x = A. Prove that ∆f (u) = f (u)∆u + f (u)|∇u|2 . 0) = (A − x). Find the determined problem for this string. Verify that ∆|x|β = (nβ + β (β − 2))|x|β −2 . . ∞) with the initial data 2b A 2b A u(x. Answer: Note that ∂xi r = x . 4 4 Exercise 2. (2). Assume 0 = x ∈ Rn . Show that the deﬁnition is well-deﬁned. x ∈ [ . Exercise 4. b. and A is A2 L2 = + b2 .88 LI MA 12.1. f or (x. r Exercise 3. Exercise 1. t) = 0 and u(A. for β ∈ R. t). n ≥ 2. Assume its initial position forms a triangle with the x-axis such that the middle point of the string is the highest point and its height is b > 0. Assume that f : R → R is in C 2 . for r = |x|. In particular. x ∈ [0. Denote by |∇u|2 = i i |∂i u|2 . prove that ∆uβ = βuβ −1 ∆u + β (β − 1)uβ −2 |∇u|2 . Deﬁne the Gamma function Γ(s) for s > 0 by ∞ Γ(s) = 0 e−r rs−1 dr. Answer: Let u be the height position of the vibrating string at (x. (1). The relation between L. Appendix B This is a section for part of homework exercises in my courses in the past years. 0) = x. Assume 0 < u = u(x) ∈ C 2 . t) = 0. t) ∈ (0. A) × (0. A] A 2 A 2 and the boundary data u(0. and Γ(s +1) = sΓ(s) for s > 0.

uy (x. Answer: u(x. ∞) with initial data u = x2 at t = 0. Answer: u(x. t) = (x − 3t)2 . Exercise 2. ∞) with initial data u = 2xex /2 at t = 0. Solve the equation 2ut = ux − xu. (x. y ) = y x 1 φ(xy ) + φ( ) 2 2 y x √ xy y − 3 + s 2 φ(s)ds 4 xy x √ xy y − 3 − s 2 ψ (s)ds. Solve the equation ut + 3ux = 0. 2 we have du t = −xu = (− + A)u. 1) = ψ (x). Exercise 4. t) ∈ R × (0. (x. 2 u(x.EQUATIONS FROM MATHEMATICAL PHYSICS 89 12. t) = et cos(x − t). Homework 2. 1) = φ(x). 2 xy 2 . ∞) with initial data u = cos x at t = 0. t) ∈ R × (0. dt 2 Then. Exercise 1. Solve the problem y 2 uyy − x2 uxx = 0 with the initial data u(x. t) = (2x + t)ex /2 .2. Answer: Note that along the characteristic curve t x = − + A. Exercise 3. Solve the equation ut + ux = u. t) ∈ R × (0. Answer: u(x. (x.

. + aj Y (x). Then the solution expression is x y (x) = j bj uj (x) + 0 g (τ )Y (x − τ )dτ... Homework 3.ak y = g (x) with the initial data y (0) = b0 . uj (x) = Y (j ) (x) + a1 Y (j −1) (x) + . .. . Assume Y (x) is a solution to the the higher order ODE of the form y (k) + a1 y (k−1) + .... .. Assume that Y is a solution to −y + y = 0.. y (k−1) (0) = 1. y (0) = 0. x > 0 with the initial conditions y (0) = 0 and y (0) = 1.. Exercise 1. ak . Answer: Deﬁne. x > 0 with the initial conditions y (0) = A and y (0) = B .90 LI MA 12. Find the solution expression for the higher order ODE of the form y (k) + a1 y (k−1) + . y (k−1) (0) = bk−1 . Given constants a1 . Answer: x y (x) = AY (x) + BY (x) − 0 g (τ )Y (x − τ )dτ.. for 0 ≤ j ≤ k − 1. Exercise 2. y (0) = b1 . . Using the Duhamel principle to ﬁnd the expression for solution to the the following ODE −y + y = g (x)...3.ak y = 0 with the initial data y (0) = 0.

L Answer: u(x. and Bn = 0 for n = 2k + 1. and An = 0 for n = 2k .. Find a solution to the following Sturm-Liouville problem: X + λX = 0. utt = a2 uxx . 2. L).. Homework 4. ut (x. and X (0) = X (L) = 0. and 8L3 Bn = 4 4 an π for n = 2k . u(0. t > 0. u(x. 1. or Xk (x) = B sin( 1 2 2 (k + 2 )π L2 )π (k + 1 2 x) L where k = 0.. L L L 8 − 4)π for n = 2k + 1. t) = n≥1 (An cos( anπ anπ nπ t) + Bn sin( t)) sin( x).EQUATIONS FROM MATHEMATICAL PHYSICS 91 12. 0) = sin2 πx . An = − n(n2 where . Exercise 1.4.. x ∈ (0. t). 0 < x < L. Exercise 2. t) = 0 = ux (L. 0) = x(L − x). Answer: λk = and X = 0.

5. y ). (ζ. y ). y ). 2 Deﬁne D = I (D). Exercise 1. Given a domain Ω ⊂ R2 . Compute the Poisson kernel in such D. Deﬁne G1 ((x. Let 0 ∈ D be a bounded domain in Rn . Let B = BR (0) and assume φ ∈ C (∂B ). (3). (x. −η )). η )) − Γ((x. Answer: We just remark that on the boundary ∂D. y ). Answer: (1). Exercise 3. Given u ∈ C (D) with ∆u(x) = f (x). η )) = G1 ((x. y ). Ω = {x > 0. y ). 2kL + η )). Ω is the half upper plane. (ζ. (−ζ. Deﬁne G2 ((x. y ). that is. (1). η )) = k=−∞ G1 ((x. (3). (2). −1). I (I (x)) = x for x ∈ D. (ζ. Consider the following problem: −∆u = f (x. (ζ. y ). y ). Exercise 2. Prove that I 2 = identity . y > 0}. η )). Deﬁne u(x) = R2 − |x|2 nωn R φ(y ) dσy . (2). Deﬁne ∞ G3 ((x.92 LI MA 12. y > 0} is the upper half plane. . x ∈ B. (ζ. 0 < y < L}. y ) ∈ Ω with the boundary condition u = φ on ∂ Ω. Assume D = {(x. Find the Green function for the problem when (1). Prove that ∆u(x) = 0. (ζ. y ). (ζ. |x − y |n ∂B where ωn = |B1 (0) is the volume of the unit ball. where L > 0 is a ﬁxed constant. . Ω = {−∞ < x < ∞. Homework 5. where I is the inversion transform x ∈ D → I (x) = x/|x|2 ∈ D on the unit sphere {|x| = 1}. the unit our normal is ν = (0. η )) = Γ((x. x ∈ D. Exercise 4. This implies that D = I (D ). η )) − G1 ((x.

Deﬁne. Using the conclusion of (2). Then ∆w(x) = |x|2−n f (x/|x|2 ). x ∈ D . y ) = Γ(x. G(x. y ) − |y |2−n Γ(x. x ∈ D which is called the Kelvin transform of the function u. y ) is the Green function on B1 (0). G(x. Then. . for x. y ∈ B1 (0). Let w(x) = |x|2−n u(x/|x|2 ). (3). ﬁnd the Green function on B1 (0). Answer: (3). I (y )).EQUATIONS FROM MATHEMATICAL PHYSICS 93 (2).

Using the Fourier transform to solve the following problem ut = a2 uxx + cu + f (x. Deﬁne. dx2 d A[u](x) = ( + x)u(x). t) = e−ct f (x. Answer: F (f )(ξ ) = 2 sin(Bξ ) . Given two constants a > 0. t). dx d C [u](x) = (− + x)u(x). Method One : Let w = e−ct u and f1 (x. u ˆt = (−a2 |ξ |2 + c)ˆ u+f with ˆ(ξ ). + 0 We can obtain this result by two methods. t) and φ(x). Exercise 1. for u ∈ S. B ξ Exercise 2. Then we have wt = a2 wxx + f1 with w(x. c and two bounded smooth functions f (x. t > 0. Homework 6. Answer: u(x.6. 0) = φ(x). u ˆ(ξ. t). t) = φ(x). τ )dξ ]. t). dx . H [u](x) = (− d2 + x2 )u(x). Method Two : Note that for u ˆ(ξ. −∞ < x < ∞. u(x. 0) = φ Exercise 3. Given a constant B > 0. t − τ )f (ξ. t) = ect [ t K (x − ξ.94 LI MA 12. t). t)φ(ξ )dξ e−cτ dτ K (x − ξ. Compute the Fourier transform of the function f (x) = 1 for |x| ≤ B and f (x) = 0 for |x| > B . ˆ(ξ.

Denote by u ˆ the Fourier transform of u. k = 1. . n=−∞ which is called the Poisson summation formula. t > 0. Then setting t = 0. −∞ −∞ Exercise 5. U (t) is a periodic function such that U (t + 2π ) = U (t) on R. 2. H is called the Hermite operator. We remark that C k (e ) are eigenfunctions of the operator H and the polynomials deﬁned by hk (x) = ex 2 /2 C k (e−x 2 /2 ) are called the Hermite functions. we get ∞ √ −1 ∞ u(n) = ( 2π ) u ˆ(n). Clearly. for u ∈ S.. H (e−x H (C k (e−x 2 /2 ) = e−x − x2 / 2 2 /2 2 /2 2 /2 )) = (2k + 1)C k (e−x )..EQUATIONS FROM MATHEMATICAL PHYSICS 95 Show that CA[u] = (H − 1)[u]. you show that t−1/2 ϑ(1/t) = ϑ(t). 2 Using the Poisson summation. you show that √ −1 ∞ U (t) = ( 2π ) u ˆ(n)eint . which is called the periodization of u. which plays an important role in modern methods of mathematical physics and its higher dimensional analogue L = −∆ + |x|2 plays an important role in Bose-Einstein condensate.. 1 + |ξ |2 . Recall that the theta function ϑ(t). Using Fourier series theory. Assume that ˆ (ξ ) = B 1 . a new function ∞ U (t) = n=−∞ u(t + 2πn). AC [u] = (H + 1)[u]. Deﬁne. Exercise 6. Exercise 4. is deﬁned by ϑ(t) = n=−∞ ∞ e−πn t .

M. Exercise 7. press. GTM120. . Assume that 1 ˆβ (ξ ) = B . or W. Answer: First we write ∞ 1 2 e−s(1+|ξ| ) ds. we have ∞ B (x) = 2−n/2 0 s−n/2 e−s− |x|2 4s ds.Ziemer’s book. which is called the second Bessel potential.1989. Princeton Univ.96 LI MA Find B (x).1970. Hint: One can ﬁnd the expression of Bβ in E. e−s ds B (x) = √ ( 2π )n/2 0 Rn Recall that a2 π 2 eiay−sy dy = ( )1/2 e− 4s . = 2 1 + |ξ | 0 Then we have ∞ 1 2 eix·ξ−s|ξ| dξ. Springer. P. (1 + |ξ |2 )β/2 Find Bβ (x). s Rn Hence. s R Then we have |x|2 π 2 eix·ξ−s|ξ| dξ = ( )n/2 e− 4s . Given β > 0. Stein’s book.

Then maxD |u(x)| ≤ max∂D |φ(x)| + where d = diam(D) := supx. Let ξ (x) = Note that ξ (x) ± u(x) ≥ 0 on ∂D and −∆(ξ (x) ± u(x)) ≥ 0 Applying the Maximum principle to the function w(x) = ξ (x) ± u(x). x ∈ ∂D. 2n . Exercise 1. we get that w(x) ≥ 0 on D. M = max∂D |φ(x)| Take p ∈ D.EQUATIONS FROM MATHEMATICAL PHYSICS 97 12. Let D be a bounded domain in Rn .y∈∂D |x − y |. Proof. This implies that d maxD |u(x)| ≤ max∂D |φ(x)| + sup|f (x)|. We denote F = sup|f (x)|. 2n F 2 (d − |x − p|2 ) + M. x ∈ D u(x) = φ(x). 2n d sup|f (x)|. Consider u ∈ C 2 (D) ∩ C (D) such that −∆u = f (x). Homework 7.7.

and the boundary data u(x. t). Note that ∆w = ∆u and at t = 0. t0 ) ≥ 0. t1 ). 0 < t ≤ T }.1 (Q) satisﬁes ut = a2 uxx + f (x. Prove that on Q. t)|. t). we have w(x1 . 0 < t ≤ T. Let Q = {(x. t)u. Answer: There is a uniform constant C = C (T. a) such that maxQ ¯ |ut | ≤ C [|f |C 1 (Q ¯ ) + |φ |C [0. there is a small 0 < t0 ≤ t1 ¯ × (0. t) ≤ M supD |φ(x)| for some constant M > 0 depending only on n. t) ∈ Q and the initial data u(x. 0 ≤ u(x. T. Proof. Consider w =u+ t+ 2 . t0 ) ≤ 0. . w = φ + > 0. Assume that u < 0 somewhere. t)| can be bounded by the suitable norm of f and φ. t) = 0 for x ∈ ∂D. t). u(x. (x. 0) = φ(x) ≥ 0. Let a > 0 be a constant. u(x1 . t) = 0 = ux (L. t0 ) = −( t0 + Hence. saying. ∆u(x0 . Given φ ∈ C (D ¯ |a(x. u(0. w(x0 . 0) = φ(x) Show that maxQ ¯ |ut (x.8. t1 ) = −( t1 + for some arbitrary small At (x1 . 0 < x < L. t0 ) = 0. we have Lw(x0 . Hence. t) ≥ 0 for (x. Exercise 1.98 LI MA 12.L] ]. t1 ) = 0.1 ¯ ) satisﬁes Assume that u ∈ C (Q) ∩ C (Q Lu := ut − a2 ∆u = −u2 + a(x. ¯ ) and a(x. t) ∈ C (Q ¯ ) with A = maxQ Exercise 2. t)D x0 ∈ D for w such that 2 2 2 ) and t1 > 0. 2. 0 < x < L. and A. Assume that u ∈ C 2. t0 ) = ∆w(x0 . Homework 8. which implies that u(x0 . t0 ] and for some minimum point such that w(x. 2 ).

we can use the maximum principle 2 (see Section 9. we get −m2 + a(x2 . − ≥ ( t0 + 2 2 ) − a(x0 . we have Lu ≥ 0. t) ≤ M supD φ(x). Then at (x2 . t0 ). t2 ). Hence. .EQUATIONS FROM MATHEMATICAL PHYSICS 99 which gives us that Lu(x0 . on Q. Hence. t0 )( t0 + 2 ). we have m2 ≤ Am and m ≤ A. t0 ) ≤ − .3) to conclude that u(x. However. we have u ≥ 0. t2 )m ≥ 0. Using the equation. t2 ) = maxQ ¯ u > 2supD φ. Remark. So this method can not be used to give the result wanted. Then. using the equation for u at (x0 . One may try another method below. Once we have u ≥ 0. this is impossible for small and t1 . Assume that m := u(x2 .

Show that d dt F (u)dx = 0. Let G be a vector-valued smooth function in R. Here ∂x G(u) = G (u)ux . t) ∈ C 1 (Rn × (0. x ] + c n u [u n x (n) for suitable constants cn and polynomials qn . t) be a smooth solution to the equation ut − uxx + cu2 x = 0.. t) = 0 uniformly in t and ∂t F (u) + ∂x G(u) = 0. R We call the quantity R F (u)dx the conservation law. Consider the Burgers’ equation ut − uxx + uux = 0. u( )dx. T )) is a bounded function such that limx→±∞ u(x. 3. t) ∈ C 1 (R × (0.. T )) is a bounded function such that lim|x|→∞ u(x. Find a function φ such that the function w = φ(u) satisﬁes the heat equation wt − wxx = 0. Assume that u(x.. 2.. Let F and G be two smooth functions in R.100 LI MA 12. Let u = u(x.9. which is called the Cole-Hopf transformation. Hint: Deﬁne En (u) = R n) 2 (n−1) 2 n−2) ([u( ] − qn (u. x ∈ R . Exercise 2. −cu Answer: w = e .. Exercise 5. Using the conclusions of Exercise 1 and Exercise 2 to ﬁnd at least 5 conservation laws of the KdV equation: ut + uux + uxxx = 0. Show that d dt uk dx = 0 Rn for each k = 1. Homework 9. Exercise 4. t) = 0 uniformly in t and ∂t u + divx G(u) = 0. . Assume that u(x. Exercise 3. Exercise 1. Here ux denotes the x-derivative of order n for the function u..

EQUATIONS FROM MATHEMATICAL PHYSICS 101 with initial data u = f (x). t) = −∞ u(y. t)dy. Let x w(x. Show that w satisﬁes 1 2 wt − wxx + wx =0 2 and use the conclusion of Exercise 4 to ﬁnd an explicit expression of u. t) = . Answer: f (y ) −|x−y |2 − 02 4t dy ( x − y ) e R u(x. . f0 (y ) −|x−y |2 t R e 4t − 2 dy x where f0 (x) = −∞ f (y )dy .

uxn )dx. Exercise 3. Homework 10. A Exercise 4. 2 2 2 1 + ux + uy 1 + u2 x + uy Answer: 1 cosh(Ar + B ). There is a rich literature about this equation. Find the Euler-Lagrange equation for the integral f (y ) = I (u) = D L(x. ux1 . Assume u = u(x. r = x2 + y 2 . j =1 . u.10. A cos(Ax) This solution is called the Sherk’s surface. we know the famous minimal surface equation: −div ( ∇u 1 + |∇u|2 ) = 0.. Exercise 2. Find solution u(x. u(xu) = 1 cos(Ay ) log | |. As an exercise for pleasure. Exercise 1. y ) = f (x) + g (y ) satisﬁes uy ux ) + ∂y ( ) = 0.. . we invite readers to ﬁnd the Euler-Lagrange equation for the area functional A(u) = D 1 + |∇u|2 dx among the class A above. Answer: From this variational structure. to the equation uy ux ∂x ( ) + ∂y ( ) = 0. n Answer: ∂u L − ∂xj (∂uxj L) = 0. Try to use the method of separation variables to the Minimal surface equation in dimension two in the following way. y ) of the form u = f (r). ∂x ( 2 2 2 1 + ux + uy 1 + u2 x + uy Then we have f g + = 0.. 1 + f (x)2 1 + g (y )2 Prove that for some constant A > 0.102 LI MA 12.

D Answer: −∆u = λu.EQUATIONS FROM MATHEMATICAL PHYSICS 103 Exercise 5. Find the Euler-Lagrange equation for the integral J (u) = D |∇u|2 dx on the space 2 C0 (D) with the constraint |u|2 dx = 1. .

edu.math.1999.Hilbert. For students who would like to take the ﬁnal examination of this course. Lectures on equations from Mathematical Physics.tsinghua.Courant. . please look my homepage at http://faculty.104 LI MA Summary This is a brief lecture note for the courses on ”Introduction to Equations from Mathematical Physics” or ”Methods for Mathematical Physics”. 1962. Interscience Publicaters. Methods of Mathematical Physics. Yazhe Chen. References [1] Jiang Lishan. heat equations. Beijing. and the Laplace equation. Thank you very much for your patience in attending my courses. Higher education press. 2. Our motivation to write this note is to show the mainstream methods used in solving some typical partial diﬀerential equations like wave equation.in Chinese. Vol. Fahuai Yi. [2] D. Xiheng Liu. R.cn/~lma for the highlight of the course.

11 Heisenberg. 51 Hermite operator. 42 Gronwall. 48 string.cn . 9 D’alembert.EQUATIONS FROM MATHEMATICAL PHYSICS 105 Index Bessel function. 39 divergence. 44 monotonicity formula. 90 Huygens. 5. 18 harmonic function. 24 fundamental solution. 35 Gaussian kernel. 36 Green. 60 Characteristic curve method. 17 Fourier. 11 Liouville. 67 maximum principle. 4 variational lemma. 67 support. 21 Dirac’s delta function. Beijing 100084. 5 Duhamel principle. 13 continuity equation. 10 Plancherel. 5. 65 wave equation. 41 heat equation. 51 mean value theorem. 59 Poisson. 88 Kirchhoﬀ. 42 Schwartz.7 Sturm. 7 Department of mathematical sciences.edu. 28 blow up. 47 partial diﬀerential operator. 87 Kelvin transform. China E-mail address : lma@math. Tsinghua university. 30 Inversion transform. 29 Laplacian.tsinghua.

- aece_2016_3_3
- 11.VIistrian MATIES, Olimpiu HANCU, Ciprian Radu RAD.pdf
- Mi2004-AirGapFlux
- d0304-Program90SMarcus
- 78242801InMemoriam-GDragan_1_2015_pp3-5
- Su Paper
- 48561371 Simulating a DC DC Boost Converter Using Matlab Part 2
- 119245295 Marnay Et Al 2008 Optimal Technology Selection and Operation of Commercial Building Microgrids
- wang_2012
- d1210-PremiiPresa2011-2013
- Guide
- Fimet_ac Motors 1.000 Kw Serie Mcv-Acv_i-e
- Casa_Nicu
- Locomotiva Cu Aburi-DODEA ROXANA
- Fimet_ac Motors Ie2 Serie Ema-em_i-e
- Fimet_ac Motors 1.000 Kw Serie Mcv-Acv_i-e
- Romania a Piece of Paradise
- Tec05 Batt Modl
- Cartea de Algoritmi
- Op to Electronic A
- a39_Automatica
- Rugaciunea Mea La Sfarsit de An
- 2011marENE Marian CV Februarie 2011
- Curs C-Programarea Calculatoarelor Si Limbaje de Programare
- 415_rez-ro

Sign up to vote on this title

UsefulNot usefulLectures Pdecourse

Lectures Pdecourse

- Info Sheet 2013
- c 0361519
- FDheat
- SAN_Maths
- PARTIAL DIFFERENTIAL EQUATIONS
- FINAL SEM 2 08-09Q
- Class 19
- 1st Ord Linear Const Coeff
- Lecture 6
- lec25.pdf
- Chapter 7 PDE
- ed034pa552.1
- Lecture 1
- HS Math May 2014
- Bsc_mathsfirstyear
- AITS2012-01-PT1P1-CMP-161011
- Maxima Book Chapter 5
- Rr322104 Computational Aerodynamics i
- Differential Equations Corel
- Guideleine for Paper
- RPT ADD MT F4
- RPT ADD MT F4 2012
- Instrumentation Engineering
- New Microsoft Word Document
- Lab 09
- Add Mth f4 Binder PDF October 11 2007-6-18 Am 376k
- The Algebra of Holonomic Equations
- Matematics and Computing
- Two Dimensional Block Pulse Functions and Application to Solve Volterra-Fredholm Integral Equations with Galerkin Method
- Grid
- Lectures Pdecourse