Wednesday, March 17, 2021

APS March Meeting, Day 3

As I warned yesterday, my work commitments (plus attending talks by three of my students)  mean that this set of highlights is attenuated.  Still some excellent talks, though, and if you are registered I encourage pulling up the recordings for some of these.

  • There was a talk this morning by Lukas Prochaska from TU Vienna, pertaining to this paper, where the charge fluctuations in the quantum critical heavy fermion compound YbRh2Si2 really blow up, as seen via THz optical conductivity measurements.  (Full disclosure, I'm working with these folks as well, and two of my colleagues are on that paper.)  A key advance is the ability to grow this comparatively exotic compound via molecular beam epitaxy (MBE).
  • Speaking of MBE, I strongly recommend the talks by this year's McGroddy Prize winners, Ivan Božović, Darrell Schlom, and Jim Eckstein.  These folks are pioneers of the growth of complex oxides by MBE, and it is really amazing how much good science has come out of the development of this technique and the resulting materials.  (Again, full disclosure, I've had the opportunity to collaborate with the first two.)
  • Speaking of pioneers, I also strongly endorse the Buckley Prize talk by Moty Heiblum.  It was simply a great explanation of how shot noise can be an incredibly useful tool to examine comparatively exotic physics (e.g., fractionally charged quasiparticles in the fractional quantum Hall regime; the breakup of neutral excitations in the fractional quantum Hall regime).   (Unfortunately I was not able to watch the other Buckley Prize talk today, but since Pablo Jarillo-Herrero is giving our colloquium next week, I get to see similar material soon.)
  • The talk by Prof. Xiaoxing Xi, very similar to his remarkable Harvard colloquium, should be required viewing.  Here is a link to the JASON report (pdf) about a much better way to handle scientific and security concerns re China.
  • Finally, you should watch this whole session if you want to see a great cross-section of the state-of-the-art on different quantum computing approaches (superconducting qubits, trapped ions, Si spin qubits (that I'd mentioned here), the ongoing Majorana business, and photonic quantum computing).  Very interesting.


Anonymous said...

What did you think about the Delft group saying that in the 2018 paper they had deleted points in a graph? I think this should be a scandal in and of itself. One can understand the self-biasing towards an interesting result, but deleting data and not coming clean up to 3 years later? In my mind it should be more of a scandal.

Anonymous said...

I also doubt that the problems with the Delft paper can be explained by just confirmation bias. Data appears to have been manipulated to confirm with what they wanted to show. For example the calibration error that shifted their plateau onto the expected quantized value. Overall, it just looks really bad and I feel deeply disappointed by the involved scientists.

Douglas Natelson said...

Anons, yeah, selectively leaving out data points or regions of gate voltage is definitely bad. The fact that there was an independent panel that investigated shows that it was taken seriously by Delft. I agree that the whole thing is disappointing.

Anonymous said...

I think the situation is a bit more nuanced; leaving out data happens all the time.
Either because of artifacts (spikes due to range switching in Keithley's, XRD equipment with spurious peaks, Fourier filtering out periodic vibrations in STM data etc) or (more problematic) due to known physics that is not relevant for the effect at hand. The latter could be XRD peaks due to sample containers or clips or something like that.

I do think in this case what they left out was problematic, because it is physics of the sample, and it thus describes the sample, provides a more complete picture of the sample. It can be okay to leave that out, BUT it should be noted in the paper.
That is the standard even for artifacts as vibrations in STM - you can take it out but have to let folks know.
This case it is actual physics, so certainly they have to mention that they did that.

The "bad" part is not correct language here, I think. Bad supposes an intention. And frankly, we don't know. Intentions are generally unknown, even if people think they know.

Incorrect, inappropriate actions can be called out. Intentions, bad or otherwise, remain hidden from the viewer, unless you have metaphysical powers.

I suggest to only describe the facts.

And even there I find the report of the investigative panel to be almost borderline - though they base their remarks about (absence of bad) intention on interviews with the actors here.

I do think the Delft group acted wholly inappropriately, blind, not objective enough in looking at their own data.

I also note that the excuse (read the report, paraphrased here) "these papers in Nature do not allow for the length to fully describe everything" - implying they would have described all they did (leaving out data) had they submitted their work to e.g. PRB - to be galling.
And I leave you to draw your own conclusion about glossy short format journals...