ProgressReport 5 June 2026
Overall a relatively slow month. Habits have all been in check and meditation has been slow and steady. Workouts have hit a bit of a downturn; I've been playing good basketball, but I have a bunch of lingering injuries: heel, right knee, left shoulder and right middle finger.
Reading has been good. I am really leaning into the physics of energy, this has been fun and I am really learning a lot. I will probably do a first pass through the book before sketching a draft of the course. Shakespeare has been progressing and I am about to start a run on AI.
I'm starting to hit a great rhythm with blogging. The routine: polish this week's post, then rough draft next. (Denis Overbye is now subscribed!!!)
Lots of social activities. Went over to Cat and Tony's, did a guys' dinner downtown with cigars, date nights are back, and I hung with T-Man. I'm thinking of getting a group together for a supper club. Evelyn Thompson was in town for Pheno, I saw her, but didn't manage to get to one Pheno talk! Rami is moving to Pittsburgh, will have an office in Shadyside.
My AI fever has slowed a bit, but it's still baked into my daily work. I'm not sure the daily driver /start /done, /reflect, etc. routine is working, but I am still all in for coding and starting to offload more of the stuff I really don't want to do; e.g., finishing rec letters. I will still look for areas to use it more.
I have migrated cworkers, which was in Python, to an Emacs build. I think this is just the right thing to do for all TUIs, there is so much that comes along for the ride. I started to build out tworkers, but still really need to put it through the paces. Emacs is another area that I think compounds for AI; it's unlocking a bunch of features I probably wouldn't have stumbled on otherwise. I think a good default is "try it first in Emacs".
On the work side, my big push recently has been my tenure packet. I will be glad when this is in the wind. It will be much stronger than three years ago. The research statement is there, but I need to push more on the teaching statement; Manfred reports that I got a bad student letter, so this is something that I will have to address in the teaching statement. AI of course was great here for copy editing; I also found it useful for structuring the research statement. I will try something similar for the teaching statement.
We got the DOE budget guidance and reviews back. Almost no change with respect to last year, except that LDMX got cut. With hindsight it was definitely a mistake for Valentina to go 50% CMS; she should have talked to a program manager. She will bump to the CMS grant at 100% (e.g., 2 months). Not sure this is fair… My reviews were pretty much all good, with some minor quibbles.
On the 4b side, most of my work has been on mixing. I have implemented and tested matching beyond just nearest neighbors (==rank 0). Rank 3 and 7 look nearly identical. With cross-ranked matches (e.g., (0,1) (1,0)) I can make N2 datasets that are as good as rank N. I still need to check the effective statistical uncertainties, but I doubt this will be a problem. I implemented the making of these datasets and the MvD training all in Snakemake workflows. Feels so good to have this in place. (This is another place I am using a lot of AI to tell me the Snakemake calls to make.)
I ran an SvB vs FeyNet comparision almost entirely with AI. When evaluated on Run3 data, the Run2 SvB training is worse than the Run3-Feynet training. This pushed me to retrain SvB, which I got running in Snakemake. The retrained Run-3 SvB outperformcs FeyNet. I still need to write this up cleanly.
We have also started making the push to make the mixed dataset public. It looks like this is going to be pretty straightforward. The low pT analysis still smells off.
In bbWW, Aniket presented his work. I think we still need a rough first-principles assessment of sensitivity and the potential gain.
On the HGC the new personnel is starting to crystallize: Ianna accepts and will start June 8th. Brennan has accepted and will start June 15th. Shrinidi will join in July. The MMTS is coming along; Aniket and Christian have really stepped up. We have had a fiasco with the Ti baseplates. All the recent batch has been essentially unusable; KIT is investigating.