Differential Properties of Solutions of Some Higher-Order Differential Equations
Ya. S. BUGROV
Submitted 1963-01-01 | SovietRxiv: ru-196301.53477 | Translated from Russian

Abstract Generated abstract

The paper studies differential properties of solutions to a Dirichlet problem in the half-plane for higher-order equations with even derivatives in two variables. Using Fourier separation of variables, it derives integral representations of solutions for cases including s = 1, 2, and 3, with generalized Poisson kernels, and estimates these kernels through Hardy, Hölder, and Minkowski inequalities. The main results give sufficient, sharp regularity conditions on boundary functions in Sobolev-type H and W classes that imply corresponding anisotropic regularity of the solution, as well as weighted summability of higher derivatives. The paper also applies these results to obtain conditions ensuring boundedness of a generalized Dirichlet integral, explicitly formulated for s = 2.

Full Text

MATHEMATICS

Ya. S. BUGROV

DIFFERENTIAL PROPERTIES OF SOLUTIONS OF CERTAIN HIGHER-ORDER DIFFERENTIAL EQUATIONS

(Presented by Academician L. I. Sedov, 5 VII 1962)

In this work we investigate the differential properties of solutions of certain boundary-value problems depending on the differential properties of the boundary functions in terms of the \(H\)-classes and \(W\)-classes of S. L. Sobolev and generalized classes.

We consider the Dirichlet problem for the equation

\[ L^{(l,s)}u \equiv (-1)^{l+1}\frac{\partial^{2l}u}{\partial x^{2l}} +(-1)^{s+1}\frac{\partial^{2s}u}{\partial y^{2s}}=0 \tag{1} \]

(\(l,s\) are natural numbers, \(y>0\), \(-\infty<x<\infty\)) under the conditions

\[ \left.\frac{\partial^j u}{\partial y^j}\right|_{y=0} =\varphi_j(x)\qquad (j=1,2,\ldots,s-1) \tag{2} \]

and the function \(u(x,y)\) is bounded in \(R_2^0=\{y>0;\ -\infty<x<\infty\}\).

The conditions (2) are understood in the sense that

\[ \lim_{y\to +0}\int_{-\infty}^{\infty} \left| \frac{\partial^j u(x,y)}{\partial y^j}-\varphi_j(x) \right|^p dx=0 \qquad (1<p<\infty). \tag{3} \]

Let us note that for \(l=s=1\), \(L^{(1,1)}u\equiv \Delta u\), where \(\Delta\) is the Laplace operator. Applying the Fourier method of separation of variables, we obtained an analytic representation of the solution of the boundary-value problem (1)—(2). In particular, for \(s=1\) the solution of the boundary-value problem has the form

\[ u(x,y)=\int_{-\infty}^{\infty}K_0(t,y;l,1)\varphi_0(x+t)\,dt, \tag{4} \]

where

\[ K_0(t,y;l,1)=\frac{1}{2\pi}\int_0^\infty e^{-\lambda^l y}\cos\lambda t\,d\lambda; \]

for \(s=2\)

\[ u(x,y)=\int_{-\infty}^{\infty}K_0(t,y;l,2)\varphi_0(x+t)\,dt +\int_{-\infty}^{\infty}K_1(t,y;l,2)\varphi_1(x+t)\,dt, \tag{5} \]

where

\[ K_0(t,y;l,2)=\frac{1}{2\pi}\int_0^\infty e^{-\nu y}(\cos\nu y+\sin\nu y)\cos\lambda t\,d\lambda, \]

\[ K_1(t,y;l,2)=\frac{1}{2\pi}\int_0^\infty e^{-\nu y}\frac{\sin\nu y}{\nu}\cos\lambda t\,d\lambda \qquad \left(\nu=\frac{\sqrt{2}}{2}\lambda^{l/2}\right); \]

for \(s=3\)

\[ u(x,y)=\int_{-\infty}^{\infty} K_0(t,y;l,3)\varphi_0(x+t)\,dt+ \int_{-\infty}^{\infty} K_1(t,y;l,3)\varphi_1(x+t)\,dt+ \]

\[ +\int_{-\infty}^{\infty} K_2(t,y;l,3)\varphi_2(x+t)\,dt, \]

where

\[ K_0(t,y;l,3)=\frac{1}{2\pi}\int_0^\infty \left[ e^{-y\lambda^\alpha}+\frac{2\sqrt{3}}{3}e^{-\frac{y}{2}\lambda^\alpha} \sin\frac{\sqrt{3}}{2}y\lambda^\alpha \right]\cos\lambda t\,d\lambda \quad \left(\alpha=\frac{l}{3}\right), \]

\[ K_1(t,y;l,3)= \]

\[ =\frac{1}{2\pi}\int_{-\infty}^{\infty}\lambda^{-\alpha} \left[ e^{-y\lambda^\alpha}+e^{-\frac{y}{2}\lambda^\alpha} \left( \sqrt{3}\sin\frac{\sqrt{3}}{2}y\lambda^\alpha -\cos\frac{\sqrt{3}}{2}y\lambda^\alpha \right) \right]\cos\lambda t\,d\lambda, \]

\[ K_2(t,y;l,3)= \]

\[ =\frac{1}{2\pi}\int_{-\infty}^{\infty}\lambda^{-2\lambda} \left[ e^{-y\lambda^\alpha}+e^{-\frac{y}{2}\lambda^\alpha} \left( \frac{\sqrt{3}}{3}\sin\frac{\sqrt{3}}{2}y\lambda^\alpha -\cos\frac{\sqrt{3}}{2}y\lambda^\alpha \right) \right]\cos\lambda t\,d\lambda. \]

Using the explicit representation of the solution of the boundary-value problem, on the basis of the properties of the kernels \(K_j(t,y;,l,s)\) \((j=0,1,\ldots,s-1)\), which we call generalized Poisson kernels, and with the aid of the Hardy, Hölder, and Minkowski inequalities, we proved a number of theorems generalizing the corresponding results of S. M. Nikol’skii \((^1)\), O. V. Besov \((^3)\), S. V. Uspenskii \((^{5,6})\), L. D. Kudryavtsev \((^7)\), and others.

Let \(R_1=\{-\infty<x<\infty\}\), \(R_2(0,1)=\{0<y<1,\ -\infty<x<\infty\}\).

Theorem 1. If the functions \(\varphi_j(x)\in H_p^{\left[r+(s-1-j)\frac{l}{s}\right]}(R_1)\) \((r>0;\ 1<p<\infty;\ j=0,1,\ldots,s-1)\), then the solution of the boundary-value problem (1)—(2)

\[ u(x,y)\in H_{px}^{\left[r+\left(s-1+\frac{1}{p}\right)\frac{l}{s}\right]}\,[R_2(0,1)] \]

for any \(r>0\);

\[ u(x,y)\in H_{py}^{\left(\frac{sr}{l}+s-1+\frac{1}{p}\right)}[R_2(0,1)] \]

for \(p=2\), \(r\) arbitrary; for \(p\ne2\), \(r+(s-1-j)\frac{l}{s}\ne 2k\) \((2k\) an even number\()\)*.

Theorem 2. If the functions \(\varphi_j(x)\in W_p^{\left[r+(s-1-j)\frac{l}{s}\right]}(R_1)\) \((1<p<\infty;\ j=0,1,\ldots,s-1;\ r+(s-1-j)\frac{l}{s}\) are not integers for \(p\ne2\)), then the solution of problem (1)—(2)

\[ u(x,y)\in W_p^{(r_1,r_2)}[R_2(0,1)], \]

where

\[ r_1=r+\left(s-1+\frac{1}{p}\right)\frac{l}{s},\qquad r_2=\frac{s}{l}r_1 \]

**.

\[ \underline{\phantom{xxxxxxxxxxxxxxxx}} \]

* For the definition of the classes \(H_p^{(r_1,r_2)}\), see \((^2)\).

** For the definition of the classes \(W_p^{(r_1,r_2)}\), see \((^{8,4,5})\).

The proof of Theorems 1 and 2 was carried out by us for \(s=1,2,3\) and arbitrary natural \(l\). These theorems cannot be improved in these terms, as follows from the results of S. M. Nikol’skii \(({}^{2})\), p. 296, and O. V. Besov \(({}^{4})\), p. 79.

Further, we have shown that the solution of problem (1)—(2) has derivatives of arbitrary order, summable with a weight. For example, for \(s=1\) the following holds.

Theorem 3. Let \(\varphi_0(x)\in W_p^{(r)}(R_1)\) \((r>0;\ r=\rho+d,\ 0<\alpha<1;\ \rho\) an integer\()\); \(\frac{r}{l}+\frac{1}{p}=n+\beta,\ 0\leqslant \beta<1;\ r+\frac{l}{p}=nl+l\beta=nl+k+\gamma,\ k=0,1,\ldots,l-1,\ 0\leqslant \gamma<1,\ n\) an integer.

Then the function (4) has derivatives of order higher than \(n\) with respect to \(y\) and of order higher than \(nl+k\) with respect to \(x\), summable with a weight:

\[ \left\| \frac{\partial^{\,n+s_1+s_2}}{\partial y^{\,n+s_1}\partial x^{s_2}}\, y^{\,s_1-\beta+\frac{s_2}{l}} \right\|_{L_p[R_2(0,1)]} \leqslant c\|\varphi_0\|_{W_p^{(r)}(R_1)}, \]

where \(s_1=1,2,\ldots;\ s_2=0,1,\ldots;\)

\[ \left\| \frac{\partial^{\,nl+k+s_1+s_2}u}{\partial y^{s_1}\partial x^{\,nl+k+s_2}}\, y^{\,s_1+\frac{s_2-\gamma}{l}} \right\|_{L_p[R_2(0,1)]} \leqslant c\|\varphi_0\|_{W_p^{(r)}(R_1)}, \]

where \(s=0,1,\ldots;\ s_2=1,2,\ldots,\)

\[ \|\psi\|_{L_p[R_2(0,1)]} = \left( \int_0^1\int_{-\infty}^{\infty}|\psi|^p\,dx\,dy \right)^{\frac1p}. \]

This result is definitive and generalizes the result of S. V. Uspenskii \(({}^{6})\) (Theorem 2).

Using Theorems 1 and 2, one can obtain sufficient conditions for the boundedness of the generalized Dirichlet integral

\[ D(u;l,s)_p = \int_0^\infty\int_{-\infty}^{\infty} \left[ \left|\frac{\partial^l u}{\partial x^l}\right|^p + \left|\frac{\partial^s u}{\partial y^s}\right|^p \right]\,dx\,dy \]

in terms of \(H\)-classes and \(W\)-classes. We give these conditions for \(s=2\).

Theorem 4. Let the functions \(\varphi_0\in W_n^{\left(\frac{l}{2q}+\frac{l}{2}\right)}(R_1)\), \(\varphi_1\in W_n^{\left(\frac{l}{2q}\right)}(R_1)\), where \(\frac1p+\frac1q=1\) (for \(p\ne2\), \(\frac{l}{2q},\ \frac{l}{2q}+\frac{l}{2}\) are not integers).

Then these functions can be continued to \(R_2^0\) in the form of the solution (5), with \(D(u;l,2)_p<\infty\), and conditions (2) are satisfied.

In terms of \(H\)-classes, a similar theorem is formulated with accuracy up to an arbitrary \(\varepsilon>0\).

Blagoveshchensk State
Pedagogical Institute

Received
28 VI 1962

References

  1. S. M. Nikol’skii, Matem. sbornik, 33 (77), 2, 247 (1954).
  2. S. M. Nikol’skii, Matem. sbornik, 33 (75), 2, 261 (1953).
  3. O. V. Besov, Izv. AN SSSR, ser. matem., 20, 469 (1956).
  4. O. V. Besov, Tr. Matem. inst. im. V. A. Steklova AN SSSR, 60, 42 (1961).
  5. S. V. Uspenskii, Tr. Matem. inst. im. V. A. Steklova AN SSSR, 60, 280 (1961).
  6. S. V. Uspenskii, DAN, 138, No. 4, 385 (1961).
  7. L. D. Kudryavtsev, DAN, 107, No. 4, 501 (1956).
  8. S. L. Sobolev, Some Applications of Functional Analysis in Mathematical Physics, Leningrad, 1950.

Submission history

Differential Properties of Solutions of Some Higher-Order Differential Equations