Why can't I get a good fit when I use a Gaussian process

[Yzxastr]

This is my original drawing This is GPfit This is my code

[kecnry]

This looks like a flux-scaling issue to me, you likely need to set/fit pblum to an appropriate value (see the pblum tutorial for more details). I'd also suggest trying to make sure that you have a decent synthetic light curve that agrees with your observations before adding GPs into the mix.

[Yzxastr]

Thank you so much,i have get good fit by set/fit pblum to an appropriate value.