Distance for Alnilam/ε Ori?

[Monque27]

This morning I discovered that the distance to ε Ori has been revised recently, from ~1900 ly to around ~1300 ly.

I checked to see which value Stellarium had (version 25.1), and it's missing altogether. Is this intentional, or an oversight?

[henrysky]

This is intentional beacuse ε Ori has very large parallax uncertainty [1.65 +/- 0.45 mas] so you cannot just inverse parallax to get the distance. In general, you can only do inverse parallax to get distance up to 20% parallax uncertainty.

[Monque27]

Thank you, Henry.

You could still calculate a distance - I've seen the distance stated in a number of places, including SIMBAD. The uncertainty in the parallax obviously leads to uncertainty in the distance, and I can appreciate the reservations about a large uncertainty making the value less useful. But still calculable.

Is the problem that the +/- distances would be two significantly different values? Does Stellarium compute the uncertainty, or is it encoded in the database?

[henrysky]

You can always compute inverse parallax as the distance but does not mean you always should and the major issue is not large parallax uncertainty propagate to large distance uncertainty but how can one even infer unbiased distance in the case of large parallax uncertainty (the answer to this question is nothing you can really do unless you have other information like prior assumptions on distances). If you are interested, you can refer to this article (https://arxiv.org/pdf/1507.02105) that discusses the issue in-depth. SIMBAD and other places might use distance measurements that do not rely on parallax. But the distance value in Stellarium solely depends on parallax and is simply inverse parallax.

If you are looking for a short answer: the distance distribution will look normal by inverting parallax distribution with small normal uncertainty while the distance distribution you get from parallax distribution with large parallax uncertainty will not look normal. Distance from inverting parallax with large uncertainty will always be biased (by underestimating the true distance in most cases).

[Monque27]

Thanks again for an expansive answer, and the pointer to the article.