On uniqueness for some non-Lipschitz SDE (Q2400597)

The aim of the paper is to investigate the path-by-path uniqueness of the strong solution to stochastic differential equation \[ X_t=x_0 +\int_0^t b(X_s)ds +W_t, \] in which \(W_t\) is a noise process with continuous paths, \(b:\mathbb R\to \mathbb R\) is a function assumed to be at least measurable and \(x_0\) is a given real number. The existence and uniqueness of strong solution to this equation has been already proved in the case that \(b\) satisfies a Lipschitz condition and also under the weaker condition of boundedness and measurability of \(b\) in the case of Brownian motion \(W\). The authors apply some techniques from the theory of ODEs to study the path-by path uniqueness of strong solution to the above equation. First, they use an extension of Iyanaga's uniqueness theorem for ODEs an the Girsanov theorem to stablish the uniqueness in the case of Brownian motion \(W\). Then, they give a proof when \(b(x)=\sqrt{|x|}\) and the noise is nonnegative. At last, mimicing the method of Peano's uniqueness theorem, they analyze the uniqueness of the solution for noises with strictly negative derivative.