Last week I looked inside Brookhaven National Laboratory's tandem Van de Graaff accelerator. The platform in the middle is called the terminal and can be raised to 15 million volts while surrounded by insulating SF6 gas (unusual to see an accelerator large enough to climb inside!) I wanted to see if there was room for an ultra-cold ion source, which could create nanometre-scale ion beams once accelerated by the Tandem. #physics #accelerators
#Accelerators
Basin-hopping optimisation of a particle beam transfer line with six different energies in it. #physics #accelerators Something like this might be required for the 22GeV CEBAF energy upgrade at Jefferson Lab, to join the arcs to the linac where all the beams are in the same pipe.
Why this matters to the industry
Nvidia solidifies dominance beyond GPU training
Competitive pressure shifts in AI hardware
The deal signals industry focus on inference
Talent consolidation and future architectures (LPU?)
#NVIDIA #Groq #AI #ArtificialIntelligence #Semiconductors #AIChips #AIHardware #Inference #AIInference #GPUs #ASIC #Accelerators #DataCenter #TechNews #ChipWars #tech
What Nvidia’s acquisition of Groq means for AI industry?
Why this matters to the industry
Nvidia solidifies dominance beyond GPU training
Competitive pressure shifts in AI hardware
The deal signals industry focus on inference
Talent consolidation and future architectures (LPU?)
#NVIDIA #Groq #AI #ArtificialIntelligence #Semiconductors #AIChips #AIHardware #Inference #AIInference #GPUs #ASIC #Accelerators #DataCenter #TechNews #ChipWars #Tech
Doing a first assembly and fit test of ion trap's next configuration. This time, the electron beam is going through the trap and can be switched on and off rapidly via deflection electrodes and a small hole. The blue plastic is modelling the Micro-Channel Plate that is still in storage vacuum. #physics #accelerators
Simulation of a long bunch breaking up into solitons in the BNL booster synchrotron. #accelerators #physics This required setting focussing magnets to much weaker than normal and starting with a dense, high space-charge bunch. Doing this experiment in practice could be a precursor to crystalline beams, as when Coulomb crystals are put in place of the solitons here, they heat up by less than 1 Kelvin/second.
Space charge solitons forming in a simulation of the BNL booster (with unusual magnet settings). #accelerators #physics I was trying to figure out if the stability of 3D crystalline bunches I'd found was because they were solitons and it appears to be the case.
3-day week this week. Finished a conference paper, reviewed a paper, and did some error studies of the crystalline beam in a ring. It can go 25mm off axis and bounce around, not heating up much, but more sensitive to its internal velocity distribution. #physics #accelerators
Just finished the paper to go with my COOL'25 talk on crystalline beams. There are some rings in which a Coulomb crystal can make 20000+ turns without cooling. #physics #accelerators
Energy that would normally go to waste inside powerful #particle #accelerators could be used to create valuable #medical isotopes, scientists have found.
#Physics #Cancer #sflorg
https://www.sflorg.com/2025/11/phy11242501.html
#accelerators #physics Rectangular bends are OK, horizontal tune 0 (but different path lengths). Sector bends are a complete disaster! But an in-between one gives you roughly equal path lengths and a horizontal tune of 1.
Today I presented a talk at COOL'25 workshop, including these simulations of a ring that can store a three-dimensional Coulomb crystal for 1/16 of a second, or more than 24000 turns. The crystal stays solid and bunched without cooling or RF. #accelerators #physics
Will be going to the COOL'25 workshop on beam cooling at Stony Brook University next week. Looks like a very interesting set of talks: https://indico.jacow.org/event/98/timetable/#20251027.detailed
I made a picture of the cold bunch dynamical model while writing some slides. Just equations can be a bit dry! #physics #accelerators
https://stephenbrooks.org/ap/report/2025-9/model_order1_analytic.pdf
Fixed some well-hidden simulation bugs. One was putting the magnetic gradient dB/dx where the magnetic field B should be. The other was using u^v to write vector cross product in C, but precedence means a+u^v evaluates as (a+u)^v, so I needed brackets! #physics #accelerators
Beam dynamics in #accelerators usually assumes your bunch has a finite temperature. But what if your bunch has zero temperature? This is the case for transporting ultra-low emittance bunches (Coulomb crystals) from ion traps. The usual "beta functions" don't really work, so I developed equations for the 3D fully-coupled dynamics in the note below. I'm hoping it can be used to do stability analysis (in 21 parameter space, so many ways it can go unstable!) #physics
https://stephenbrooks.org/ap/report/2025-9/model_order1_analytic.pdf
GPU vẫn là vua AI nhờ linh hoạt & hệ sinh thái mạnh; TPU/custom chip chỉ thắng ở mảng chuyên biệt, tiết kiệm điện. Doanh nghiệp chọn GPU vì dễ scale, đào tạo nhanh, chi phí tốt. #GPU #AI #DeepLearning #TPU #Accelerators #CôngNghệ #GPUvn
https://www.reddit.com/r/SaaS/comments/1o249l3/do_you_think_gpu_clusters_will_remain_dominant/
Lots of interesting talks at the Fixed Field #Accelerators (FFA'25) workshop at Imperial College London last week. #physics
Composite of all 12 beams put through our medical accelerator test section.
From left to right:
[1-6] Deuterium at 12.9, 14.3, 16, 20, 24.2, 28MeV at the BNL Tandem Van de Graaff, equivalent (after energy loss in beam windows) to 10, 15, 20, 30, 40, 50 MeV protons.
[7-12] Protons at 50, 70, 100, 150, 200, 250MeV at the BNL NASA Space Radiation Facility (NSRL).
The permanent magnets (built by SABR Enterprises, LLC.) delivered up to 1.85 Tesla of field to bend the beams on this tight radius.
More information on the design:
https://www.bnl.gov/newsroom/news.php?a=222471
Client Info
Server: https://mastodon.social
Version: 2025.07
Repository: https://github.com/cyevgeniy/lmst