Right now the ArgumentEstimator implementations build the logarithmic derivative from the target function f and its derivative df themselves. For some functions it might be useful to be able to provide the logarithmic derivative df / f directly. This way one could e.g. simplify this derivative by hand or precompile it separately. There might even be use cases where only the logarithmic derivative is given...
Right now the
ArgumentEstimatorimplementations build the logarithmic derivative from the target functionfand its derivativedfthemselves. For some functions it might be useful to be able to provide the logarithmic derivativedf / fdirectly. This way one could e.g. simplify this derivative by hand or precompile it separately. There might even be use cases where only the logarithmic derivative is given...