search

SmoQySuite / JDQMCFramework.jl

Exports suite of types and routines to simplify writing a DQMC code.
MIT License
5 stars 2 forks source link

Other DQMC efforts #2

Open carstenbauer opened 1 year ago

carstenbauer commented 1 year ago

Hi, great to see more Julia DQMC stuff! Just to make sure you haven't missed it, are you aware of the other efforts over at StableDQMC.jl, MonteCarlo.jl, and my original DQMC code in https://github.com/carstenbauer/dqmc ? Would be great to make clear how they relate to each other.

cohensbw commented 1 year ago

Your packages look really awesome and cool! Not just your DQMC ones either, a lot of them are very handy! For instance, your BinningAnalysis.jl package is very nice! And the discussion in your docs for ThreadPinning.jl regarding the threading issues with using OpenBLAS vs MKL are super helpful! I wasted a week of my life on that problem in grad school...

Thank you for reaching out! I have been slammed during the first year of my postdoc, but I was actually planning on reaching out to you myself in a few months once things calmed down a bit for me, and some projects with collaborators started getting off the ground.

Your scipost article that goes along with your StableDQMC.jl package is a really useful resource! I plan on citing that paper (and your actual dissertation) in basically every future DQMC related paper I write, as they are both very valuable resources for people trying to understand the nuts and bolts of DQMC 😅

carstenbauer commented 1 year ago

Glad you find some of my/our work useful! While I'm not directly in physics anymore, we can definitely chat more about what your plans are and if there's room for collaboration (in whatever form). (Maybe at JuliaCon at MIT in July?)

Regarding the Scipost article, I'm really happy that you find it useful! I wrote it during my PhD because I spent soo much time on this stuff and thought that future PhD students shouldn't have to suffer as much as I did. So, it's close to my heart :D Most of the papers about these stabilization techniques were rather math focused and, in some sense, "unpractical" if you understand what I mean. Ironically, my paper was almost rejected for exactly this reason (because it wasn't math heavy enough, i.e. no real proofs).

cohensbw commented 1 year ago

Your dissertation is also great!

After looking at ALF's documentation, and reading this older review article by Loh and Gubernatis (https://www.cs.ucdavis.edu/~bai/QUEST/tutorial/qmc_article.pdf) I became aware of this trick of alternating the direction of iterating over imaginary time in order to achieve linear scaling in beta. But neither resource really explained the details at all so I had to work it out for myself...

Imagine my shock then when I opened up your dissertation for the first time and discovered you describe this entire procedure in it that I had been forced to work out on my own! :joy: I was both frustrated I hadn't seen your dissertation earlier, and also grateful you wrote it. I plan on citing it moving forward because it is such useful resource for the nuts and bolts of dqmc, even above and beyond your scipost paper.

cohensbw commented 1 year ago

Unfortunately I will not be able to attend JuliaCon this year. I hope to in future years. That said, I would be more than happy to set up a time to chat via Zoom or something. As you will be in the United States on the east coast in my time zone in July for JuliaCon we can try and figure out a time to chat then. Sooner than that would also work, though finding a time that works for both of us may be a little tricky.