-
Notifications
You must be signed in to change notification settings - Fork 30
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
split dual AD and mathematical types #10
Comments
It's ok @mlubin, there is no actual conflict, at least not on my part. I think we both agree that the mathematically correct definition of dual equality differs from the one that serves the practical purposes of autodiff. We also both agree that the latter is what is used in practice. For this reason, as I have previously mentioned, I don't mind if we change the existing mathematical definition of equality to match that of autodiff. By that I mean that you can simply change the definition without needing to proceed and define a second type to satisfy mathematical equality. If for any reason mathametical equality arises in the future as a programming need, then we can create the second type - for now simply change the definition of |
I also don't think that it is worth spending time on a |
There seems to be some conflict over whether we should follow the mathematical convention for dual numbers or use the definitions that allow dual numbers to be dropped in automatically to compute derivatives. I propose separating these into two separate types to settle the conflict.
As the primary use of DualNumbers, as far as I'm aware, is for computing derivatives, I'd argue that the AD type should keep the name
Dual
. This also avoids breakage downstream. PerhapsDualAlg
could be used for the algebraic/mathematical type?@Scidom @andreasnoackjensen
The text was updated successfully, but these errors were encountered: