[Quantlib-users] difference between Gaussian1dSwaptionEngine and Gaussian1dJamshidianSwaptionEngine

[<log...@se...>]

Hello, 
I'm pricing a swaption where the underlying swap's floating leg is based on 
curve1.

Then I have another curve, say curve2, which I use in GSR like this

model = ql.Gsr(curve2, volStepDates, volatilities, reversions)





When I use in the swaption pricing:

A) engine = ql.Gaussian1dSwaptionEngine(model)

B) engine = ql.Gaussian1dJamshidianSwaptionEngine(model)


I'm getting quite a significant difference between NPV A) and NPV B), is not
due to technical nature of the pricer.





It feels like Gaussian1dJamshidianSwaptionEngine ignores the floating leg's 
curve1 (and usese curve2 everywhere), while the Gaussian1dSwaptionEngine 
takes into account curve1 for the floating leg. Can someone please explain 
how curves are treated in these pricers? 




Thanks, 

M.