SLIDE 1 What to use for the default supernova burst model(s)?
DAQ Physics Performance WG October 10, 2019 It would be nice to have a robust prediction of the “mean” expected signal, as well as the an expected range of signals, in flavor, energy, time, over full interval
SLIDE 2 Issues with selecting default model:
- there are very few models based on sophisticated
supercomputer sims that go out to even ~10 secs
- there may be interesting stuff happening later
- the ones that exist vary by ~order of magnitude
in many features (flavor-time-energy)
- some variation is expected due to “truth” variations,
(progenitor mass, composition, etc.)
different assumptions in the modeling (2D vs 3D, EOS, transport, magnetic fields,...)
- we have some model databases
- but they are primarily “collections of anecdotes”
- desirable to have a frequency distribution
SLIDE 3
What we have in hand (can apply MSW oscillations under NO/IO)
Model Time dependence? Frequency distribution? Comments
Generic pinched spectrum No Can choose any parameters Livermore Yes, to ~10 s (integrated over time in Snowglobes) No, single model Deprecated, but used a lot in the past GKVM Yes, to ~10 s (integrated over time in Snowglobes) No, single model Collective effects, but single example Garching Yes, to 10 s No, single model (a bit weird, low flux) Used for most plots so far Huedepohl database Yes, to ~10 s No, collection of anecdotes Detailed sim Nakazato database Yes, to ~10 s No, collection of anecdotes Detailed sim, no explicit pinching Burrows database Only neutronization burst No, collection of anecdotes O’Connor compactness distribution No, integrated over time Yes, based on compactness distribution Strange star formation Out to ~100 s No, unknown rate Poor spectral info
SLIDE 4
“Garching model”: used for these plots
But it’s a possibly a non-standard kind of supernova
(the only complete time-dependent info available until recently)
SLIDE 5 More extensive models we’ve been working with: see Vitor Luzio’s collaboration meeting talk (he’s showing the first ~1 s, but many go to ~10 s)
SLIDE 6
(Some of these may be unreliable...)
SLIDE 9 An attempt to find a sensible probability distribution : for signals: see Crystal Burgos collaboration meeting talk
- find time-integrated spectrum/flux vs compactness
- use compactness & distance distributions to find
distribution of neutrino fluxes
But this doesn’t have time profile information...
SLIDE 10
We shouldn’t forget that weird stuff may happen:
AJ Roeth Jan 2019 collab meeting talk on strange star formation Events out to ~100 sec with specific time profile
SLIDE 11 What to use? at the moment, can’t have everything...
Low-flux, so maybe conservative
- Huedepohl database: citable, extensive,
BUT some models may be less reliable (?), and no indication of relative probabilities (spectrum vs time histograms are available)
/pnfs/dune/persistent/users/schol/histos
- We should not design anything too dependent
- n model assumptions, because there may be
surprises
- IMO if we tune DAQ/trigger such that MO
matters, it’s too tuned
