da Bokonon » 06/12/2019, 14:20
Riporto due passi dal capitolo 5 del paper in oggetto inbtitolato "Sociologia delle discussioni scientifiche".
Una paternità congiunta (in grassetto) di un risultato sotto un nome generico di un'attribuzione scientifica era impensabile fino a 50 anni fa (il lavoro era lasciato agli storici).
As one sees from Section 3 and Figure 1, there was some partial dissemination of the eigenvector-eigenvalue identity amongst some mathematical communities, to the point where it was regarded as “folklore” by several of these communities.
However, this process was unable to raise broader awareness of this identity, resulting in the remarkable phenomenon of multiple trees of references sprouting from independent roots, and only loosely interacting with each other. For instance, as discussed in the previous section, for two months after the release of our own preprint [DPTZ19], we only received a single report of another reference [FZ19] containing a form of the identity, despite some substantial online discussion and the dozens of extant papers on the identity. It was only in response to the popular science article [Wol19] that awareness of the identity finally “went viral”, leading to what was effectively an ad hoc crowdsourced effort to gather all the prior references to the identity in the literature.
[...]
It is not fully clear to us how best to attribute authorship for the eigenvectoreigenvalue identity (2). The earliest reference that contains an identity that implies (2) is by Loewner [L¨34], but the implication is not immediate, and this reference had only a modest impact on the subsequent literature. The paper of Thompson [Tho66] is the first explicit place we know of in which the identity appears, and it was propagated through citations into several further papers in the literature; but this did not prevent the identity from then being independently rediscovered several further times; furthermore, we are not able to guarantee that there is an even earlier place in the literature where some form of this identity has appeared.
We propose the name “eigenvector-eigenvalue identity” for (2) on the grounds that it is descriptive, and hopefully is a term that can be detected through search engines by researchers looking for for identities of this form.