Processing math: 100%

2024年3月8日 星期五

Solution to Partial Differential Equations: Methods and Applications (Robert McOwen) Section 5.3

Solution to Partial Differential Equations: Methods and Applications (Robert McOwen) Section 5.3

  1. Suppose u is a distribution solution of the nonhomogeneous heat equation ut=Δu+f(x,t) in URn+1, where fC(U). Show that uC(U).
  2. SolutionPick ξ=(x0,t0)U and ϵ>0 so small that the ball B4ϵ(ξ)U. Choose ϕC0(B4ϵ(ξ)) with ϕ1 on B3ϵ(ξ). Let w(x,t)=ϕ(x,t)u(x,t) and v(x,t)=wt(x,t)Δw(x,t)f(x,t) in U, which are both distributions in U with support in B4ϵ(ξ). Clearly, v=0 on B3ϵ(ξ). In particular, the convolution

    (˜Kv)(x,t)=Rn+1˜K(xy,ts)v(y,s)dyds=Rn+1˜K(y,s)v(xy,ts)dyds

    is well-defined as a distribution. But

    v(xy,ts)=wt(xy,ts)Δw(xy,ts)f(xy,ts)=(tΔx)w(xy,ts)f(xy,ts)=(sΔy)w(xy,ts)f(xy,ts),

    so for gC0(U), we find

    ˜Kv,g=URn+1˜K(y,s)[(sΔy)w(xy,ts)f(xy,ts)]g(x,t)dydsdxdt=URn+1(sΔy)˜K(y,s)w(xy,ts)g(x,t)dydsdxdt+URn+1˜K(y,s)f(xy,ts)g(x,t)dydsdxdt=URn+1[δ(y,s)w(xy,ts)+˜K(y,s)f(xy,ts)]g(x,t)dydsdxdt=U[w(x,t)+(˜Kf)(x,t)]g(x,t)dxdt=w+˜Kf,g.

    In other words, ˜K(vf)=w as distributions.

    Now we choose ψC(Rn+1) with ψ0 on Bϵ(0) and ψ1 on Rn+1B2ϵ(0). Then ψ vanishes in a neighborhood of the singularity of ˜K, so ψ˜K is smooth function, and hence so is (ψ˜K)(vf). But for (x,t)Bϵ(ξ) and (y,s)supp(v)B4ϵ(ξ)(Rn+1B3ϵ(ξ)), we have |(x,t)(y,s)|2ϵ, so ψ(xy,ts)=1. In other words, (ψ˜K(vf))(x,t)=(˜K(vf))(x,t)=w(x,t) for (x,t)Bϵ(ξ), which shows that w is smooth on Bϵ(ξ). But w(x,t)=ϕ(x,t)u(x,t), so u is also smooth on Bϵ(ξ). Since ξU was arbitrary, we conclude that u is smooth on U.


  3. Consider the heat equation ut=Δu in (x,t)Rn×(0,) with the initial condition u(x,0)=δ(x). Since u(x,t)dx represents the total heat content in Rn at time t, we encounter the condition u(x,t)dx=1 for all t0.
    1. Show that the choice α=n/2, which was made to simplify (41) to (42), is a consequence of the scale invariance of u(x,t)dx for t>0.
    2. Show that (43b) is a consequence of u(x,t)dx<.
    3. Use u(x,t)dx=1 to evaluate the constant C in (44), and conclude that u(x,t) coincides with the heat kernel.
  4. Solution
    1. Suppose it is known that u(x,t)=tαw(x/t) (see (39)). Then we use the change of variable to find

      C=Rnu(x,t)dx=tαRnw(x/t)dx=tαRnw(y)tndy=tα+n/2Rnw(y)dy.

      Due to the t-independence, we have α=n/2. The proof is complete.
    2. From (43a), we have

      w+r2w=Ar1n.

      Multiplying it by er2/4, it yields

      (er2/4w)=Ar1ner2/4,

      which gives

      w(r)=w(0)er2/4+Aer2/4r0s1nes2/4dsfor r(0,).

      From part (a), the condition Rnu(x,t)dx< implies 0rn1w(r)dr<. This means

      A0rn1er2/4r0s1nes2/4dsdr<.

      Here we have used the fact that 0w(0)rn1er2/4dr is finite. However, we apply L'Hôpital's rule to observe that

      limr[rrn1er2/4r0s1nes2/4ds]=limrr0s1nes2/4dsrner2/4=limrr1ner2/4nrn1er2/4+rner2/4r/2=limr22nr2+1=2.

      This means the function rn1er2/4r0s1nes2/4ds1r as r sufficiently large, and the integral 0rn1er2/4r0s1nes2/4dsdr must be divergent. Therefore, A must be zero.
    3. By (44) and Rnu(x,t)dx=1, we have

      1=Ctn/2Rnexp(|x|24t)dt=Ctn/2ni=1Rexp(x2i4t)dxi=Ctn/2ni=14πt=Ctn/2(4πt)n/2=(4π)n/2C.

      Therefore, we get C=(4π)n/2 and hence

      u(x,t)=1(4πt)n/2exp(|x|24t),

      which conincides with the heat kernel.

  5. Apply the similarity method to the linear transport equation ut+aux=0 to obtain the special solutions u(x,t)=c(xat)α.
  6. SolutionLet ˜u=ϵαu, ˜x=ϵβx and ˜t=ϵγt, where α, β and γ are constants, i.e.,

    ˜u(˜x,˜t)=ϵαu(ϵβ˜x,ϵγ˜t).

    To scale out the t-dependence means that we should take ϵγ=t1 so that ˜t=1; that is if we introduce w(˜x)=˜u(˜x,1) and z=˜x, then u(x,t)=tα/γw(z) and z=tβ/γx defines the similarity transformation to the similarity variables w and z. For convenience, we take γ=1 and hence

    u(x,t)=tαw(z),z=tβx.

    The linear transport equation becomes

    0=ut+aux=αtα1w(z)+tαw(z)βtβ1x+atαw(z)tβ=αtα1w(z)+atαβw(z)βtαβ1xw(z)=αtα1w(z)βtα1zw(z)+atαβw(z)=tα1(αw(z)βzw(z))+atαβw(z).

    If we choose β=1, then

    u(x,t)=tαw(z),z=x/t.

    Here w satisfies

    αw(z)+(az)w(z)=0,

    which can be solved by w(z)=c|za|α. Therefore, we obtain

    u(x,t)=tαc|za|α=tαc|xta|α=c|xat|α.


  7. Consider the inviscid Burgers equation ut+uux=0.
    1. Use the similarity method to reduce this equation to the ODE ww(α+1)zw+αw=0.
    2. Take α=1 and obtain the rarefaction wave solutions discussed in Section 1.2.
  8. Solution
    1. Let ˜u=ϵαu, ˜x=ϵβx and ˜t=ϵγt, where α, β and γ are constants, i.e.,

      ˜u(˜x,˜t)=ϵαu(ϵβ˜x,ϵγ˜t).

      To scale out the t-dependence means that we should take ϵγ=t1 so that ˜t=1; that is if we introduce w(˜x)=˜u(˜x,1) and z=˜x, then u(x,t)=tα/γw(z) and z=tβ/γx defines the similarity transformation to the similarity variables w and z. For convenience, we take γ=1 and hence

      u(x,t)=tαw(z),z=tβx.

      The inviscid Burgers equation becomes

      0=ut+uux=αtα1w(z)+tαw(z)βtβ1x+tαw(z)tαw(z)tβ=αtα1w(z)βtα1zw(z)+t2αβw(z)w(z)

      If we choose β=α+1, then

      u(x,t)=tαw(z),z=tα1x.

      Here w satisfies

      w(z)w(z)(α+1)zw(z)+αw(z)=0,

      which is desired.
    2. Taking α=1, we get w(z)1=0, which gives w(z)=z+C. Therefore, we obtain

      u(x,t)=t1(z+C)=t1(x+C)=x+Ct.


    1. For n=1 and α=0, express the solution of (41) in terms of the error function

      erf(y)=2πy0es2ds.

    2. Use (a) to solve the heat equation on a half-line: ut=uxx for x,t>0 with u(x,0)=1 for x>0, and u(0,t)=0 for t>0.
  9. Solution
    1. For n=1 and α=0, equation (41) becomes

      w+12rw=0.

      Multiplying it by er2/4 and then integrating over [0,t], we get(er2/4w)=0 and

      er2/4w(r)=w(0).

      It is clear that

      w(r)=w(0)+w(0)r0et2/4dt=w(0)+2w(0)r/20es2ds=w(0)+w(0)πerf(r/2).

    2. By (a), we get the solution u in the following form:

      u(x,t)=w(z)=w(xt)=w(0)+w(0)πerf(x4t)for x,t>0.

      Clearly, erf(0)=0 and limyerf(y)=1. By the boundary condition u(0,t)=0, we get w(0)=0. Using the initial data u(x,0)=1 for x>0, we get w(0)=1/π. Therefore, we obtain the solution u given by

      u(x,t)=erf(x4t)for x,t>0.


  10. Apply the similarity method to the porous medium equation ut=Δ(uγ) in Rn, where γ>1, to obtain Barenblatt's solution:

    u(x,t)=tα(C(γ1)β|x|22γt2β)1/(γ1),

    where C is a positive constant, β=(n(γ1)+2)1, and α=nβ.

    Remark. Here u(x,t) is defined and positive for (γ1)β|x|2<C2γt2β, but may be extended by zero to a weak solution on Rn×(0,). In this way, u exhibits finite propagation speed, unlike the solution of the heat equation γ=1.

  11. SolutionLet ˜u=ϵαu, ˜x=ϵβx and ˜t=ϵδt, where α, β and δ are constants, i.e.,

    ˜u(˜x,˜t)=ϵαu(ϵβ˜x,ϵδ˜t).

    To scale out the t-dependence means that we should take ϵδ=t1 so that ˜=1; that if we introduce w(˜x)=˜u(˜x,1) and z=˜x, then u(x,t)=tα/δw(z) and z=tβ/δx defines the similarity transformation to the similarity variables w and z. For convenience, we take δ=1 and hence

    u(x,t)=tαw(z),z=tβx.

    The porous medium equation becomes

    0=utΔ(uγ)=αtα1w(z)+tαw(z)βtβ1xtαγΔx(wγ(z))=αtα1w(z)βtα1zw(z)tαγx(tβγwγ1(z)w(z))=αtα1w(z)βtα1zw(z)γtαγβ[tβ(γ1)wγ2(z)|w(z)|2+tβwγ1(z)Δw(z)]=αtα1w(z)βtα1zw(z)γtαγ2β[(γ1)wγ2(z)|w(z)|2+wγ1(z)Δw(z)].

    In order to obtain a time-independent partial differential equation for w, the powers of t must agree, which gives α1=αγ2β, and hence β=(α(γ1)+1)/2. Thus, we find that

    u(x,t)=tαw(z),z=t(α(γ1)+1)/2x.

    Here w satisfies

    αw(x)βzw(z)γ[(γ1)wγ2(z)|w(z)|2+wγ1(z)Δw(z)]=0.

    Now we assume that w is radial, i.e., ˜w(r)=w(z) for |z|=r. Then ˜w satisfies

    α˜w(r)βr˜w(r)γ[(γ1)˜wγ2(r)˜w2(r)+wγ1(z)(˜w(r)+n1r˜w(r))]=0,

    which implies

    α˜w(r)βr˜w(r)[(wγ(r))+n1r(˜wγ(r))]=0.

    If we choose α=nβ, then we observe that rn1(α˜wβr˜w(r))=β(rn˜w). Multiplying the differential equation by rn1, we obtain

    (rn1(˜wγ))+β(rn˜w)=0,

    which implies

    rn1(˜wγ)+βrn˜w=A.

    For the sake of simplicity, we assume that w and w decay to zero as r tends to infinity, and hence A=0. Thus, we get

    ˜wγ2˜w=βγr,

    which gives us

    ˜wγ1(r)=C(γ1)β2γr2.

    Therefore, we arrive at

    w(z)=(C(γ1)β2γ|z|2)1/(γ1),u(x,t)=tαw(tβx)=tα(C(γ1)β2γt2β|x|2)1/(γ1),

    where α=n(n(γ1)+2)1 and β=(n(γ1)+2)1.

沒有留言:

張貼留言