FrankTian0012

Thank you for your answer! It helps a lot. But based on your answer, I have some other questions: Why not just do `cfh.staticPhaseRotationCompensation` and `cfh.testPhases` on the original symbols?...

Thank you very much for your clear answers and they help me a lot to understand the whole system. Really appreciate it and thank you! Now I have some other...

Thank you very much! Sorry for one more question for performance evaluation, Why not just use a series of launch powers to train separately thus getting a line of `effSNR`...

> Yes, due to limitations in computational resources and the limited personal time I could dedicate post-PhD Yeah thank you very much. Really appreciate that. You help me a lot...

Hi Rasmus~I'm sorry to bother you again. Now I have one other problem. It is about the symbol error rate and bit error rate shown in your simulations. Take [GS_GMI](https://github.com/Rassibassi/claude/blob/master/examples/tf_AutoEncoderForGeometricShapingNlinGmi.ipynbl)...

Thank you very much. Yes, GMI should be the metrics. Can I understand it this way? We focus on GMI as a metric because even though the pre-BER is high...

Thank you very much! And appologies for responding late. I asked silly questions now I get it. Also when I did simulation with PS in NLIN model. I noticed that...