compute_dtt RuntimeWarning

[gmsilveira]

I'm using msnoise to compute dv/v on a single station, considering also the cross-components.
I'm trying to apply a filter between 0.5 and 1.0 Hz and, for some days, I get the following RuntimeWarning

---

msnoise/s06compute_dtt.py:139: RuntimeWarning: invalid value encountered in double_scalars
wavg = (d * w).sum() / w.sum()
s06compute_dtt.py:141: RuntimeWarning: invalid value encountered in double_scalars
wstd = np.sqrt(np.sum(w * (d - wavg) ** 2) / ((N - 1) * np.sum(w) / N))

---

Can it be related to my filter options including the mwcs length and step?

I've just compute dv/v cross-correlations in a lower frequency band (0.1 - 1.0 Hz) and all went fine.

Many thanks for any suggestion.

Graça Silveira

[ThomasLecocq]

@asyates did you have a look at that? replacing the "/" by "np.divide"?

[asyates]

Not one I'm so familiar with. This is assuming that the source of error is a division by zero i.e. w.sum() = 0?

@gmsilveira, on the days that have the warning, do the errors in the corresponding MWCS files appear as expected? i.e. non-zero or otherwise.

That is, for a given day, there is a corresponding .txt file for each station/component pair in the MWCS directory that contains as columns the lag time, delay time measured, error, and coherence. w.sum() is based on the value in the errors column. Wondering if this shows anything unusual (errors much larger than your dtt_maxerr or alternatively non-numeric error values).

[ThomasLecocq]

Remember that msnoise2 doesn't make txt files anymore :)

[asyates]

Ah, I made the assumption that this was using release version. Is it dev version ?

[gmsilveira]

In fact I am using both, as in my new MAC OS with M2 it was easier to install msnoise master. But those zero divisions happened in version 1.6.? Professora Coordenadora Departamento de Física Instituto Superior de Engenharia de Lisboa Rua Conselheiro Emidio Navarro,1 1959-007 Lisbon Portugal & Instituto Dom Luiz IDL Edif. C8, FCUL, Campo Grande 1749-016 LISBOA Portugal Email: ***@***.*** On 2 Jul 2024, at 20:23, Alec Yates ***@***.***> wrote: Ah, I made the assumption that this was using release version. Is it dev version ? — Reply to this email directly, view it on GitHub<#341 (reply in thread)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ACHHNXTIW6PL5O6PXBHDNS3ZKL445AVCNFSM6AAAAABHLBING6VHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM4TSNBQGUYTG>. You are receiving this because you were mentioned.Message ID: ***@***.***>