weak but distinct divergence is revealed on
both sides of the superfluid transition, leading
to a pronounced lambda peak aroundTcwith
a considerable incrementdk∼ 3 nℏkB=m. This
weak divergence is consistent with the dy-
namic critical scaling theory of the super-
fluid transition ( 3 – 5 ), in whichk∼jtj n=^2 ≃jtj ^1 =^3.
The observed sudden rise in the sound diffusivity,
asshowninFig.4,AandC,canbealsoattrib-
uted to such a divergence. Finally, the Prandtl
number Pr (inset of Fig. 4B) is about unity
near the superfluid transition, suggesting that
the viscous damping and thermal damping are
equally important to the sound attenuation.
The obtained Pr≈1 implies that the unitary
Fermi gas can be treated as a holographic con-
formal nonrelativistic fluid ( 41 ). In liquid helium,
the investigations of critical divergence in the
thermal conductivity above thel-transition
and in the second sound attenuation below
thel-transition play a vital role in setting up the
effective theory for the critical mode across
the superfluid transition ( 3 – 5 ). For the unitary
Fermi gas, our measurements not only com-
plete the macroscopic description of its super-
fluidity but also provide a means to understand
the microscopic details of the superfluid tran-
sition in the strongly interacting regime.
Outlook
Our system offers great promise for studying
many fundamental problems in quantum many-
body systems with strong interactions. For
example, by investigating the temperature and
wave number dependence of the density re-
sponse, the transition from collisionless to
hydrodynamic behavior of the unitary Fermi
gas can be fully characterized and thus illu-
minate the establishment of hydrodynamics
in the strongly interacting regime. Moreover,
by adjusting the box-trap geometry (e.g., a
longer longitudinal length) and further opti-
mizing the system, Bragg spectroscopy with a
smallerwavenumberandahigherenergyre-
solution can be implemented. Therefore, a
systematic exploration of the quantum critical
region with improved temperature controlla-
bility can be achieved, paving the way to map
out several long-sought universal critical dy-
namic scaling functions. Our setup can also be
readily modified to realize a two-dimensional
homogeneous Fermi superfluid and thus pro-
vides an ideal platform for investigating the
second sound attenuation and related quan-
tum transport across the Berezinskii-Kosterlitz-
Thouless transition.
REFERENCES AND NOTES
- L. Tisza,Nature 141 , 913 (1938).
- L. Landau,Phys. Rev. 60 , 356–358 (1941).
- R. A. Ferrell, N. Menyhárd, H. Schmidt, F. Schwabl,
P. Szépfalusy,Phys. Rev. Lett. 18 , 891–894 (1967). - P. C. Hohenberg, A. Aharony, B. I. Halperin, E. D. Siggia,
Phys. Rev. B 13 , 2986–2996 (1976). - P. C. Hohenberg, B. I. Halperin,Rev. Mod. Phys. 49 , 435–479 (1977).
6. P. C. Hohenberg, P. C. Martin,Ann. Phys. 34 , 291–359 (1965).
7. H. Hu, P. Zou, X.-J. Liu,Phys. Rev. A 97 , 023615 (2018).
8. P. Zhang, Z. Yu,Phys. Rev. A 97 , 041601 (2018).
9. J.A.Tarvin,F.Vidal,T.J.Greytak,Phys. Rev. B 15 , 4193–4210 (1977).
10. G. Ahlers,Phys. Rev. Lett. 43 , 1417–1420 (1979).
11. R. A. Ferrell, J. K. Bhattacharjee,Phys.Rev.Lett. 51 , 487–489 (1983).
12. P. A. Lee, N. Nagaosa, X.-G. Wen,Rev. Mod. Phys. 78 , 17–85 (2006).
13. S. A. Hartnoll,Nat. Phys. 11 , 54–61 (2014).
14. S. Giorgini, L. P. Pitaevskii, S. Stringari,Rev. Mod. Phys. 80 ,
1215 – 1274 (2008).
15. H. Hu, P. D. Drummond, X.-J. Liu,Nat. Phys. 3 , 469–472 (2007).
16. T.-L. Ho, Q. Zhou,Nat. Phys. 6 , 131–134 (2009).
17. S. Nascimbène, N. Navon, K. J. Jiang, F. Chevy, C. Salomon,
Nature 463 , 1057–1060 (2010).
18. M. J. H. Ku, A. T. Sommer, L. W. Cheuk, M. W. Zwierlein,
Science 335 , 563–567 (2012).
19. M. Horikoshi, S. Nakajima, M. Ueda, T. Mukaiyama,Science
327 , 442–445 (2010).
20. C. Caoet al.,Science 331 , 58–61 (2011).
21. L. A. Sidorenkovet al.,Nature 498 , 78–81 (2013).
22. P. B. Patelet al.,Science 370 , 1222–1226 (2020).
23. B. Mukherjeeet al.,Phys. Rev. Lett. 118 , 123401 (2017).
24. L. Baird, X. Wang, S. Roof, J. E. Thomas,Phys. Rev. Lett. 123 ,
160402 (2019).
25. B. Frank, W. Zwerger, T. Enss,Phys. Rev. Res. 2 , 023301 (2020).
26. H. Hu, E. Taylor, X.-J. Liu, S. Stringari, A. Griffin,New J. Phys.
12 , 043040 (2010).
27. J. Stengeret al.,Phys. Rev. Lett. 82 , 4569–4573 (1999).
28. S. Hoinkaet al.,Nat. Phys. 13 , 943–946 (2017).
29. T. L. Yanget al.,Phys. Rev. Lett. 121 , 103001 (2018).
30. G. Zürnet al.,Phys. Rev. Lett. 110 , 135301 (2013).
31. A. L. Gaunt, T. F. Schmidutz, I. Gotlibovych, R. P. Smith,
Z. Hadzibabic,Phys. Rev. Lett. 110 , 200406 (2013).
32. See supplementary materials.
33. P. C. Hohenberg, E. D. Siggia, B. I. Halperin,Phys. Rev. B 14 ,
2865 – 2874 (1976).
34. A. D. B. Woods, R. A. Cowley,Rep. Prog. Phys. 36 , 1135–1231 (1973).
35. S. Jensen, C. N. Gilbreth, Y. Alhassid,Phys. Rev. Lett. 125 ,
043402 (2020).
36. R. J. Donnelly, C. F. Barenghi,J. Phys. Chem. Ref. Data 27 ,
1217 – 1274 (1998).
37. M. Blake,Phys. Rev. Lett. 117 , 091601 (2016).
38. D. T. Son,Phys. Rev. Lett. 98 , 020604 (2007).
39. J. A. Joseph, E. Elliott, J. E. Thomas,Phys. Rev. Lett. 115 ,
020401 (2015).
40. P. K. Kovtun, D. T. Son, A. O. Starinets,Phys. Rev. Lett. 94 ,
111601 (2005).
41. M. Rangamani, S. F. Ross, D. T. Son, E. G. Thompson,J. High
Energy Phys. 2009 , 075 (2009).
42. X. Li, X. Luo, S. Wang, K. Xie, X.-P. Liu, H. Hu, Y.-A. Chen,
X.-C. Yao, J.-W. Pan, Data for“Second sound attenuation near
quantum criticality.”Zenodo (2021); https://doi.org/10.5281/
zenodo.5767197.
ACKNOWLEDGMENTS
We thank X.-J. Liu, H. Zhai, M. K. Tey, H.-N. Dai, and Q.-J. Chen for
their critical reading of the manuscript; J. E. Thomas, M. K. Tey,
and M. W. Zwierlein for sharing their experimental data; and
J. E. Drut, G. Wlazłowski, S. Jensen, and Y. Alhassid for sharing
their quantum Monte Carlo (QMC) data.Funding:This work is
supported by the National Key R&D Program of China (grant no.
2018YFA0306501), National Natural Science Foundation of
China (grant no. 11874340), the Chinese Academy of Sciences
(CAS), the Anhui Initiative in Quantum Information Technologies,
the Shanghai Municipal Science and Technology Major Project
(grant no. 2019SHZDZX01), and the Fundamental Research
Funds for the Central Universities (under grant no.
WK2340000081).Author contributions:Y.-A.C., X.-C.Y., and
J.-W.P. conceived the research. X.Li, X.Luo., S.W., K.X., X.-P.L.,
and X.-C.Y. performed the experiment and collected the data. X.Li,
X.Luo., S.W., H.H., Y.-A.C., X.-C.Y., and J.-W.P. contributed to the data
analysis and writing of the manuscript.Competing interests:The
authors declare no competing interests.Data and materials:The data
are archived at Zenodo ( 42 ).SUPPLEMENTARY MATERIALS
science.org/doi/10.1126/science.abi4480
Supplementary Text
Figs. S1 to S10
References ( 43 – 62 )
31 March 2021; accepted 21 December 2021
10.1126/science.abi4480EMERGING COMPUTINGReconfigurable perovskite nickelate electronics for
artificial intelligence
Hai-Tian Zhang^1 *†‡, Tae Joon Park^1 *†, A. N. M. Nafiul Islam^2 †, Dat S. J. Tran^3 †, Sukriti Manna4,5†,
Qi Wang^1 †, Sandip Mondal^1 §, Haoming Yu^1 , Suvo Banik4,5, Shaobo Cheng^6 ¶, Hua Zhou^7 , Sampath Gamage^8 ,
Sayantan Mahapatra^9 , Yimei Zhu^6 , Yohannes Abate^8 , Nan Jiang^9 , Subramanian K. R. S. Sankaranarayanan4,5,
Abhronil Sengupta^2 , Christof Teuscher^10 , Shriram Ramanathan^1 *Reconfigurable devices offer the ability to program electronic circuits on demand. In this work, we
demonstrated on-demand creation of artificial neurons, synapses, and memory capacitors in post-fabricated
perovskite NdNiO 3 devices that can be simply reconfigured for a specific purpose by single-shot electric pulses.
The sensitivity of electronic properties of perovskite nickelates to the local distribution of hydrogen ions
enabled these results. With experimental data from our memory capacitors, simulation results of a reservoir
computing framework showed excellent performance for tasks such as digit recognition and classification of
electrocardiogram heartbeat activity. Using our reconfigurable artificial neurons and synapses, simulated
dynamic networks outperformed static networks for incremental learning scenarios. The ability to fashion the
building blocks of brain-inspired computers on demand opens up new directions in adaptive networks.C
ontinual learning in artificial intelligence
(AI) presents a formidable challenge.
Models are generally trained on station-
ary data distributions, and thus when
new data are presented incrementally
to a neural network, this interferes with the
previously learned knowledge, resulting inpoor performance, which is known as cata-
strophic forgetting and remains an active field
of research ( 1 , 2 ). One of the major approaches
to tackle this issue is to actively adapt the
structure of the network itself when new data
becomes available. Not only does modulating
the architecture of the network in response toSCIENCEscience.org 4 FEBRUARY 2022•VOL 375 ISSUE 6580 533
RESEARCH | RESEARCH ARTICLES