GSR Process Kernel Crash in Python

[apapanico]

This came up on the mailing list last year but no one filed a ticket.

Setup:

import QuantLib as ql

print(ql.__version__)

YRs, nSTEP = 2, 4

gsSeqRNG = ql.GaussianRandomSequenceGenerator(
    ql.UniformRandomSequenceGenerator(nSTEP, ql.UniformRandomGenerator(0))
)

# expire  sigma   reversion

rPROC = ql.GsrProcess([1] * 1, [0.005] * 2, [0.03])

gsPathGN = ql.GaussianPathGenerator(rPROC, YRs, nSTEP, gsSeqRNG, False)

1.40

And then executing this causes a kernel crash

path = gsPathGN.next().value()

Interestingly, at least in a jupyter notebook, you can access the data before saving the variable.

print(gsPathGN.next().value()[0])
print(gsPathGN.next().value()[1])

yields

0.0
nan

[boring-cyborg]

Thanks for posting! It might take a while before we look at your issue, so don't worry if there seems to be no feedback. We'll get to it.

[lballabio]

I'm reopening this because I had to revert in #2417 the change made in #2410 — it avoided the crash, but it also caused regressions in the calibration of the process; see the Gaussian1DModels example, which started printing model prices and vols that didn't match the market inputs. I think that the calibration code actually relies on the process holding references to the original vectors, so that changes made to them are seen by the process. This needs further investigation.