A library of noise processes for stochastic systems like stochastic differential equations (SDEs) and other systems that are present in scientific machine learning (SciML)
This should fix #121 , which was reported on Slack.
Currently, NoiseWrapper has no specialized reinit! method and falls back to the general one for AbstractNoiseProcess, which only sets W.curW for a NoiseGrid, otherwise W.curW is left unchanged. So, when solving it twice, W.curW starts the second run with a value equal to the last value of the previous run.
This PR adds a specialized reinit! for NoiseWrapper that takes care of that.
I could have checked for a NoiseWrapper inside the reinit of AbstractNoiseProcess, but I though a separate reinit! would be cleaner.
This should fix #121 , which was reported on Slack.
Currently,
NoiseWrapperhas no specializedreinit!method and falls back to the general one forAbstractNoiseProcess, which only setsW.curWfor aNoiseGrid, otherwiseW.curWis left unchanged. So, when solving it twice,W.curWstarts the second run with a value equal to the last value of the previous run.This PR adds a specialized
reinit!forNoiseWrapperthat takes care of that.I could have checked for a
NoiseWrapperinside the reinit ofAbstractNoiseProcess, but I though a separatereinit!would be cleaner.