I have been experimenting with Claude Code to write codes to do extensive calculations in theoretical physics, and I have become more and more convinced that LLM and AI can help scientists accelerate the progress.
This brings me to my old day dream of simulating KS gauge theory to study the instability of anti-d3-brane using lattice QCD. This does not seem to be an impossible goal, I might start doing something about this as a hobby.
Feb 27, 2026
Due to the difficult job market and the end of my postdoc at Stanford on the horizon, I started looking at AI/ML for good reasons. I have been experimenting with some AI tools, and seriously, they are getting really good in real time.
I think it might be even possible to write some code to perform string loop amplitude calculations in CY orientifolds in large volume approximations with the help of AI. Given that the technicality of the SFT calculations has been prohibitive, I eventually had to recognize that to push the boundary of what is doable and what is not, one must write a package to do calculations. But, given my limited time, it was a hard decision to write such a package. But, now, i don't think the cost of such work is too high, as long as I have the right ideas.
I will explore how to use AI for physics projects, and it makes me genuinely excited. Having a difficult time in the job market after all wasn't that bad thing, as this opened my eyes finally!
Jan 25, 2026.
Happy New Year!
The winter break this time gave me a good time to reflect on my career. Also, it seems to me that my, probably the last, academic job search ended in failure at least for now. I will start learning more employable skills very soon.
Jan 9, 2026