China’s dark matter space probe detects tantalizing signal | Hacker News Reader

China’s dark matter space probe detects tantalizing signal

sciencemag.org

52 pointsyuvalr111 comments

For those interested the preprint is at https://arxiv.org/pdf/1711.10981.pdf The satellite carries a very large BGO calorimeter https://arxiv.org/abs/1406.3886

I noticed this paragraph about code to do this analysis:

  "Code availability. The numerical code has been developed
  with a dedicated application to the DAMPE data analysis. Due
  to the uniqueness of the DAMPE design and the complexity in-
  volved in the data analysis, the software package has limited 
  application to the relevant community. We have opted not to
  make the code public."

I'm just an IT guy, and not familiar with how disciplines distribute source-code for this type of analysis, but is this approach common? what does "...limited application" mean here and why not release the code?

I'm a tech, not an academic, but I see it as part of the broader problem of science reproducibility. Software plays a huge huge part in modern science publishing and, in the case of data derived using proprietary software, it's expected we just trust the results - that the software is infallible.

The comment "We have opted not to make the code public." really gets my goat, but is probably less because they sampled community opinion and found it wanting but more because it's just plain inconvenient or unknown how to publish software.

Authors are not actively discouraged from publishing details about the software used (in journals in general) or including copies in the supplementing files, but the process typically has the editors (who guide the authors through the submission process) paring back the submission to just bog standard and familiar media types: image, audio, video, pdf, etc that are known quantities and can be nicely formatted in the final pdf.

My small contribution to combating this has been encouraging our editors at eLife to be alert to authors using bespoke or modified software and capturing that software at a specific revision at the time of publication (https://github.com/elifesciences-publications) and making sure it meets some basic criteria, like being openly licensed.

I'm hoping this effort will help resist general entropy and bit-rot, software moving on or getting lost, licence changes and the overall mandate of eLife to improve science publishing, which means reproducible results and software amenable to audit.

It's frustrating but normal not to release code.

It makes it easier to release further papers, with less work. Limited application means, "this is our work, all rewards should go to us", preventing others from doing any further research, so no one can "beat them to the punch", so to speak.

However, on the other side, it does also prevent checking results accurately.

Hmmmm...scientists are still debating whether the signal is dark matter or pulsar output (high energies create electron-positron pairs). I see this recent Chinese paper debating if DAMPE can even distinguish dark matter positron remnants or pulsars - Wang et al. (2017) say this - "For the pulsar case, the latest Fermi-LAT anisotropy limits have also been taken into account. Our results show that the cases of DM annihilations to τ+τ− and all charged leptons are difficult to be distinguished from the pulsar cases by the current experiments and DAMPE"

> In its first 530 days of scientific observations, DAMPE detected 1.5 million cosmic ray electrons and positrons

> "we now expect it to last 5 years"

> That will allow the satellite to record more than 10 billion cosmic ray events.

Assuming DAMPE continues detecting cosmic ray events at its current rate, I think the total number after 5 years would be a bit over 5 million. That's a few magnitudes off of 10 billion... Math fail, or do I misunderstand something?

1.5m / 530d * 5y * 365d/y = 5.165m

Your first quotation should go: > In its first 530 days of scientific observations, DAMPE detected 1.5 million cosmic ray electrons and positrons above a certain energy threshold

So presumably the total number of recorded events is more than 1.5 million.

From the paper: During these ∼530 days of operation, DAMPE has recorded more than 2.8 billion cosmic ray events, including ∼ 1.5 million CREs above 25 GeV.

Ah, makes sense now -- should've gone to the source instead of trusting the article to summarize all numbers. Thanks!

Have we ever observed annihilation? Its supposed to emit a ton of energy, right?

Matter-antimatter annihilation? Yes, absolutely. As one starting point, see: https://en.wikipedia.org/wiki/Electron–positron_annihilation

Reply on news.ycombinator.com