3 ms·

( Aside: link to preprint: https://arxiv.org/abs/1802.08257 https://arxiv.org/abs/1802.08257 ) tl;dr: parochialism is convenient Practically anyone doing real

( Aside: link to preprint: https://arxiv.org/abs/1802.08257 https://arxiv.org/abs/1802.08257 ) tl;dr: parochialism is convenient Practically anyone doing real astronomy will use a well-defined continuous Earth-centric (or solar-system barycentric) timescale for a variety of reasons including reproducibility within the solar system (after all, astronomers on Earth only have other Earth-bound/near-Earth astronomers to communicate with at present) and the uncertainties in (spacetime) position of astronomical objects dwarfing the precision of local timescales (e.g. TT or TAI). Describing observables at one timestamp (a form of gauge fixing) is convenient, but pushes the uncertainty into the spatial slice thus fixed, and a suitable choice of coordinates can push most of that uncertainty into one of the three spacelike axes. Since there is a gauge freedom -- meaning you can choose to slice up the spacetime position arbitrarily -- you would be perfectly free to use coordinates fixed on the observed object as you are doing, but that's incovenient for discussing the relationship between a distant object and a closer one, and because uncertainties in spacetime position for astronomical objects are commonplace and demand revision when new data provides a better estimate of position. Concretely, you say "two or three days", capturing the uncertainty in distance to Prox. Cen. That's rather worse than the sub-second errors in the ALMA timestamp data (e.g. JD 2457836.8349, one of the timestamps in the preprint). Moreover, further observation of the alpha Cen. system will almost inevitably reduce uncertainties in position of Prox. Cen., which raises the question of whether one should go back and revise any timestamps like your "two or three days before New Year's Eve 2012" as better data enable better retrodiction of its position relative to Earth. Fixing a Gregorian-like date on a set of spacetime coordinates originating on an event at a celestial object is a false de-parochialization; that is, it's no more general. If there are astronomers on Proxima b, it would be a stunning coincidence if their start-of-year aligned with our start-of-year on that orbit (and we don't know enough about the rotation of Proxima b to talk about their local length-of-day). Astronomers elsewhere in the galaxy (and in particular in our future) are unlikely to be helped by swapping observations-at-JD-xxxyyy for a best guess of local-to-event observables inferred by retrodiction from those same Earth-bound observations.

danielvf 9y ago
We don’t know exactly how far away the stars are, so when we convert earth time to star local time, we are being less accurate. Better to keep times stored local to earth, and let distances get more accurate as we learn more.
- raattgift 9y ago
  Sure, that's one of the points I made. There isn't really any more precise timekeeping equipment available to us than atomic clocks here on (and near) Earth, so using those are expedient. In principle we could measure the spectra of the relic fields, but with current technology we can't do that for the cosmic neutrino background, and while the cosmic microwave background (CMB) is an attractive option (we have lots of instruments measuring it in excruciatingly fine detail) its utility as a quasi-universal clock is based on picking out an ideal observer who sees no non-cosmological redshift (i.e., no contributions from proper motion or local gravitation), and such an observer isn't really physical. Converting local measurements of the CMB to an ideal observer's in a way which achieves precision approaching that of our terrestrial atomic timescales depends on understanding our own local (~ Mpc scale) environment better than we do today. (Probably we will get there by using CMB observations to steer local oscillators, much as we do in atomic clocks, H2 masers, and so forth, and use those steered oscillators to predict CMB spectrum on the (well-supported) assumption that it is that of an ideal blackbody with a certain temperature). However, we'll still record actual observations at local CMB temperature (perhaps recording what the "universal" CMB temperature would be if we were a "universal" observer). We are still stuck with the problem of position of the observed object, since we cannot directly observe the CMB temperature at an event like this flare. In other words, the coordination (or if you like, localization) problem here isn't solved by switching to a different set of coordinates. Indeed, a different set of coordinates might make things worse (which was the thrust of what you replied to, and essentially what you write yourself).
gpderetta 9y ago
Sorry for the content free reply, but that was an awesome and insightful comment.
- raattgift 9y ago
  Thanks, it's just that the self-described "nitpick" is one which I've seen here and there several times, and I was finally motivated to say something about it. Describing an observed event seen at observer's time X and then recording it as happening at observer's time X minus several factors (the dominant one being light-travel-time) is at best complicated and likely loses accuracy as well. Making something needlessly less simple and less accurate is kinda the opposite of clever, but for various reasons relatively clever people keep doing exactly that. (You can get very far down the rathole of how to do precise timings of events generally, and astronomers do just that with e.g. barycentric julian dates. One might ask a computer-engineering question like: suppose one is watching timestamped logs. What does one want to see exactly? The time the logger commits the log message? The time the event generating the log entry occurred (which may require adjusting for network conditions etc)? Reversing that, does one want a clock display program to adjust for conditions between the program and the retina or even primary visual cortex of the user? And what if there are multiple users looking at such a clock?) Time is hard. Let the UTC/TAI/TT/IERS people sort it out, and use their results.)

danielvf 9y ago

We don’t know exactly how far away the stars are, so when we convert earth time to star local time, we are being less accurate. Better to keep times stored local to earth, and let distances get more accurate as we learn more.

raattgift 9y ago
Sure, that's one of the points I made. There isn't really any more precise timekeeping equipment available to us than atomic clocks here on (and near) Earth, so using those are expedient. In principle we could measure the spectra of the relic fields, but with current technology we can't do that for the cosmic neutrino background, and while the cosmic microwave background (CMB) is an attractive option (we have lots of instruments measuring it in excruciatingly fine detail) its utility as a quasi-universal clock is based on picking out an ideal observer who sees no non-cosmological redshift (i.e., no contributions from proper motion or local gravitation), and such an observer isn't really physical. Converting local measurements of the CMB to an ideal observer's in a way which achieves precision approaching that of our terrestrial atomic timescales depends on understanding our own local (~ Mpc scale) environment better than we do today. (Probably we will get there by using CMB observations to steer local oscillators, much as we do in atomic clocks, H2 masers, and so forth, and use those steered oscillators to predict CMB spectrum on the (well-supported) assumption that it is that of an ideal blackbody with a certain temperature). However, we'll still record actual observations at local CMB temperature (perhaps recording what the "universal" CMB temperature would be if we were a "universal" observer). We are still stuck with the problem of position of the observed object, since we cannot directly observe the CMB temperature at an event like this flare. In other words, the coordination (or if you like, localization) problem here isn't solved by switching to a different set of coordinates. Indeed, a different set of coordinates might make things worse (which was the thrust of what you replied to, and essentially what you write yourself).

gpderetta 9y ago

Sorry for the content free reply, but that was an awesome and insightful comment.

raattgift 9y ago
Thanks, it's just that the self-described "nitpick" is one which I've seen here and there several times, and I was finally motivated to say something about it. Describing an observed event seen at observer's time X and then recording it as happening at observer's time X minus several factors (the dominant one being light-travel-time) is at best complicated and likely loses accuracy as well. Making something needlessly less simple and less accurate is kinda the opposite of clever, but for various reasons relatively clever people keep doing exactly that. (You can get very far down the rathole of how to do precise timings of events generally, and astronomers do just that with e.g. barycentric julian dates. One might ask a computer-engineering question like: suppose one is watching timestamped logs. What does one want to see exactly? The time the logger commits the log message? The time the event generating the log entry occurred (which may require adjusting for network conditions etc)? Reversing that, does one want a clock display program to adjust for conditions between the program and the retina or even primary visual cortex of the user? And what if there are multiple users looking at such a clock?) Time is hard. Let the UTC/TAI/TT/IERS people sort it out, and use their results.)