İçeriğe geç
debya
Türkçe
Giriş yap Ücretsiz üye ol
Bluesky’da aç Bu hesabı takip et

Jens Eisert

@jenseisert.bsky.social did:plc:vf7a7kmygi3yy7l7khmoba2t

Scientist, professor of quantum physics at Freie Universität Berlin and affiliated with Helmholtz Center Berlin and the Fraunhofer Heinrich Hertz Institute. ERC Fellow.

Bluesky’a katılma tarihi:

Takipçi
1.810
Takip edilen
1.141
Post
304
Sıralamaya girdiği gün
0

Özet

Jens Eisert (@jenseisert.bsky.social), 10 Ekim 2023 tarihinde katılan bir Bluesky kullanıcısı. 1.810 takipçisi var, 1.141 hesabı takip ediyor ve 304 post paylaştı.

Sıralama geçmişi

Bu hesabın postlarının Debya günlük listesine girdiği en iyi günler.

Bu hesap henüz günlük sıralamaya girmedi.

Son beğendiği postlar

Bluesky ağındaki herkese açık beğeniler.

  • Antonio Anna Mele@antonioannamele.bsky.social

    We then realized that the same idea could be pushed further to other interesting applications, giving sample-optimal tolerant single-copy testing, optimal testing with limited quantum memory, and improved learning of states beyond stabilizers (e.g. Clifford+T states). ✨

    Postu göster
  • Antonio Anna Mele@antonioannamele.bsky.social

    Then GPT-6 came out. We asked essentially the same thing, and it gave us a surprising answer: "What you want to prove is false -- in fact, the opposite is true: there is an optimal algorithm using adaptive single-copy measurements." 💀😅 We checked the argument carefully and.. it worked!

    Postu göster
  • Antonio Anna Mele@antonioannamele.bsky.social

    So our first thought was: okay, then we just need to generalize the lower bound to adaptive strategies. We also tried prompting GPT-5.6 with something like "please prove an Omega(n^2) adaptive lower bound for this problem" -- but it couldn't do it.

    Postu göster
  • Antonio Anna Mele@antonioannamele.bsky.social

    We believed that this could not be done optimally, because a previous paper proved an Omega(n^2) lower bound. Earlier this year, however, we noticed that this proof actually applied only to non-adaptive single-copy measurements.

    Postu göster
  • Antonio Anna Mele@antonioannamele.bsky.social

    Story: It was known that one can optimally learn an n-qubit stabilizer state using O(n) two-copy Bell measurements. But what about single-copy measurements, which are much more practical?

    Postu göster

Daha fazlası için üye girişi yap

Blokladığı hesaplar

Bluesky’da bloklar herkese açık kayıtlardır.