Dear Christoph,
Am 05.11.19 um 16:36 schrieb Christoph Hertzberg:
You could create your own custom type `my_int16` for which `my_int16 * my_int16` results in a `my_int32` (this needs to be told to Eigen, similar to how real*complex products are handled).
This could indeed be a way. BTW, I was thinking in the spirit of exact scalar products for floating points like in C-XSC. Maybe an option for Eigen 7.0. Best regards, Peter
