Hochstadt H
Mathematics Department, Polytechnic Institute of New York, Brooklyn, N.Y. 11201.
Proc Natl Acad Sci U S A. 1975 Jul;72(7):2496-7. doi: 10.1073/pnas.72.7.2496.
It is known that if complete spectral data are provided, the potential function in a Sturm-Liouville operator is uniquely determined almost everywhere. If two such operators have spectra that differ in a finite number of eigenvalues, then the corresponding potential functions will no longer be the same. However, as is demonstrated when the nonidentical eigenvalues are almost equal, then the corresponding potential functions will also be nearly equal almost everywhere. Furthermore, if an operator and its spectrum are given and the potential is presumably known and if a second operator is defined in such a way such that its eigenvalues agree with the eigenvalues of the first operator except for a finite set, then the potential corresponding to the second operator can be explicitly found by solving a set of nonlinear ordinary differential equations. Lastly, it is shown that the procedures discussed here and the Gelfand-Levitan procedures are significantly different, and in fact that the Gelfand-Levitan procedure is almost certainly not well posed.
已知若提供完整的谱数据,则在一个斯特姆 - 刘维尔算子中的势函数几乎处处唯一确定。若两个这样的算子的谱在有限个特征值上不同,那么相应的势函数将不再相同。然而,正如当不相同的特征值几乎相等时所证明的那样,相应的势函数在几乎处处也将几乎相等。此外,若给定一个算子及其谱且势大概已知,并且若以这样一种方式定义第二个算子,使得其特征值除了在一个有限集上与第一个算子的特征值一致,那么通过求解一组非线性常微分方程可以明确找到与第二个算子对应的势。最后,结果表明这里讨论的过程与盖尔范德 - 列维坦过程有显著不同,事实上盖尔范德 - 列维坦过程几乎肯定是不适定的。