In this simplified version of the SGD update there is a clear relationship between momentum and the batch size $.$ What is that relation? Specifically, let's assume we train a model with momentum and a batch size $.$ How should we change the momentum if we now have a GPU with more memory and can use a batch size of $?$ Specify the momentum that would lead to equivalent gradient updated in the simplified SGD update equation above. Round to two decimal digits $($e$.$g$.0\mathrm{.}12).$