Hello again after a long time... with pretty graphs.

[benbartlett]

Hello everyone,

It's been a very long time since I've last been on here (just over three years - definitely fallen off the map for a "neutron clubber"), but I was feeling nostalgic and figured I might post a little about what I'm doing right now, for the (hopefully) most of you who remember me. I figure there's not a "welcome back" forum, and I have pretty graphs, so I'd put it in here, even though it's not solely fusor-related.

I haven't really done any work on the fusor since I won at ISEF/JSHS, but I still have it, and, though the XRT is busted, it should definitely still be plasma worthy, so an obligatory picture is attached.

Current state of my fusor and associated nostalgia...

I'm currently in Geneva working at CERN on the CMS detector in the Caltech HGCAL (High-Granularity CALorimeter) group. Before this, I was actually working in computational geophysics, of all things, working on this: http://arxiv.org/abs/1502.01421, which, after one round of revisions (not reflected in the arXiv preprint yet) will hopefully be getting published in Geophysical Research Letters soon...

What I'm currently working on is writing a jet vertex reconstruction algorithm using some of the high-precision timing data available in the upcoming HGCAL in the endcaps (hopefully to a resolution of a few hundredths of a nanosecond). It's basically a combined filtering and nonlinear minimization algorithm that will hopefully replace the current boosted decision tree (BDT) machine learning algorithms when I finish it, as it already outperforms it in resolution and should vastly outperform it in the case of high-pileup scenarios, where the latter completely breaks down. These high-pileup scenarios will be more common than not in the Phase-II upgrade going out in a few yeas.

As of yesterday, I actually got the algorithm to correctly reconstruct 99.25% of the vertices in the 1,996 simulated events I gave it down to about 1mm precision (the current machine learning algorithms would be completely unable to identify this vertex altogether, as "GammaGun" events don't show up in the tracker):

Current algorithmic error in reconstructing interaction vertices.

We can actually do some pretty cool things with the timing data that will be available in this upgrade. It will be fine enough to allow us to differentiate the actual interactions in the events, such as the animation I made from simulated data of this photon interacting with the HGCAL crystals on a 5ps time bin:



It's not very likely that we'll get the actual detector this precise, but it will still be a massive upgrade from the current version.

I quite like CERN, so I definitely see myself working with them for the foreseeable future, possibly even doing my PhD work in grad school here, though that's a little ways off. Anyway, though I haven't been particularly involved with fusor.net the last few years (other than the occasional lurking and helping some new fusioneers), I figured I'd share this stuff in case anyone finds it as interesting as I do.

Ben

[Richard Hull]

Thanks Ben. Nice to see one of our old boys doing well in the world of physics and busy on the work of the future. All the very best.

Richard Hull

[Rex Allers]

Nice post, Ben. I'm mostly a lurker here. When was your fusion achieved (~2009, 10)?

Very nice to see a (young, I assume) fusor constructor go on to joining a team for serious scientific research. Congratulations on being accepted into that realm.

On your post, the 3D gif graph with the "shadows" on the X and Y planes is a really a nice image. Was that done with some public domain package? I never saw a plot like that and I find it very intuitive and useful, with or without the animation aspect.

I hope you continue to find challenging and rewarding work to be involved with.

[benbartlett]

Hi Rexa,

I made fusion in Oct. 2009, if I remember correctly. I actually made the code for the "5D" graph (x,y,z,t,E) from scratch in python with Matplotlib, and I released a generalized, cleaned-up version of it as a relatively simply-to-use framework in GitHub repository: https://github.com/bencbartlett/Animator5D. You're free to use it if you'd like!

Best,
Ben

[Rex Allers]

Wow. Thanks for sharing your great work.