FTRL is identical to SGD for unconstrained problem when no noise is added

On line https://github.com/google-research/DP-FTRL/blob/main/optimizers.py#L59,

Why is `ms + (-gs - nz) / alpha`? This makes FTRL not FTRL is identical to SGD if learning rate is not 1.0. Shouldn't it be
`ms + (-gs - nz) * alpha`?

Start in optimizers.py at line 59 and inspect how the FTRL update uses ms, gs, nz, and alpha. Compare the current division by alpha with the proposed multiplication for an unconstrained problem without noise and a learning rate other than 1.0. Done means the intended FTRL behavior is established and the issue is resolved or clarified.

Written by the indexing model from the issue text.