Estimation of densities and applications (Q1266778)

Let \(B =\{B_t,\;t \in [0,T]\}\) be a standard Brownian motion defined on some probability space and \(X=\{X_t,\;t \in [0,T]\) be a diffusion process which is the solution of the differential equation: \[ X_t=X_0+ \int\limits_0^t \sigma(s,X_s)dB_s+ \int\limits_0^t b(s,X_s)ds,\qquad t \in [0,T]. \] Under some nondegeneracy condition it is proved that \(X_t\) has a continuous density \(p_t(x)\) such that for all \(p>1\): \[ p_t(x)\leq C_p||(\int^{t}_{0}\sigma^{2}(s,x_s)ds)^{-1/2}||_p . \]