implement a fref

[md-arif-shaikh]

#43
Added

• function to compute orbital average omega22 and the time to associate these values with https://github.com/vijayvarma392/Eccentricity/blob/fref/measureEccentricity/eccDefinition.py#L398
• function to get the ref time for given omega22 https://github.com/vijayvarma392/Eccentricity/blob/fref/measureEccentricity/eccDefinition.py#L416

[md-arif-shaikh]

[vijayvarma392]

More thoughts. Two other ways to compute the mean frequency come to mind:

1. As we already have omega_p and omega_a interpolants, just take the mean of those?
2. For residual methods, we could just use the zero ecc waveform.

For both of these approaches, no new interpolation is required.

[vijayvarma392]

And fref should be implemented as an input argument: Allow the user to pass either fref or tref_in, not both.
Allowing an array of frefs might be complicated because the user has to be careful about merger frequency, etc.
I think it's better to start implementing a scalar fref, which gets converted to a scalar tref in the init, and then everything goes through as before. This conversion can later be vectorized, but most users won't need that.

[md-arif-shaikh]

Using the omega_p and omega_a interpolants and taking the mean sounds nice. What do we call it?

[vijayvarma392]

Using the omega_p and omega_a interpolants and taking the mean sounds nice. What do we call it?

I suggest: omega22_average_between_extrema.
But don't remove your current function, it's worth having options, even if they are not turned on.

[md-arif-shaikh]

And fref should be implemented as an input argument: Allow the user to pass either fref or tref_in, not both. Allowing an array of frefs might be complicated because the user has to be careful about merger frequency, etc. I think it's better to start implementing a scalar fref, which gets converted to a scalar tref in the init, and then everything goes through as before. This conversion can later be vectorized, but most users won't need that.

I have implemented something along this line. But currently, it takes t_ref, and f_ref but if both are given t_ref is used. I can change this. Also, instead of get tref_in from given f_ref in the init, I do it at the beginning of the measurec_ecc and then everything follows

Looks like it's working for the test I did

[md-arif-shaikh]

Also added checks to see if the fref_in is within reasonable range
https://github.com/vijayvarma392/Eccentricity/blob/fref/measureEccentricity/eccDefinition.py#L204

[md-arif-shaikh]

if omega_average is monotonically increasing then we have a one to one map and even if a fref vector is provided we can get the corresponding tref_in and the current implementation should work without many additional changes. Only changes should be
change fref here https://github.com/vijayvarma392/Eccentricity/blob/fref/measureEccentricity/eccDefinition.py#L205 to fref[0] and later fref here to fref[-1] and make fref an array if it's a scalar here https://github.com/vijayvarma392/Eccentricity/blob/fref/measureEccentricity/eccDefinition.py#L202 using np.atleast_1d(fref)

[md-arif-shaikh]

Tried on fref array and it looks like working

[md-arif-shaikh]

I have added different averaging methods as option in extra_kwargs

[md-arif-shaikh]

now the behaviour when one gives frequencies to measure ecc is exactly the same as when time is provided. It accepts scalar/array of frequencies, returns fref_out and measured ecc and mean anomaly