@sundarpichai Impressive. However, in this Hands-on video (https://t.co/u73gU4s1pC), Gemini discusses why the rubber duck floats: "It is made of a material that is less dense than water." In fact, the density of rubber is higher. Seems it is misunderstanding physics: Buoyancy <> Density.
A closer look at the Community Notes algorithm, which is supposed to make Twitter the most accurate source of information https://t.co/jvyjLA0XbH @elonmusk@CommunityNotes
@Olu_GH @HohlbeinLab Yes. I would suggest checking the CLI code here:
https://t.co/CHYzbd95g6 This shows which functions to call. As for phasor: I had a fork once but didn't implement it completely and therefore deleted it. The implementation of the core idea was quite straightforward though.
@christlet In principle, you can keep both. One thing to consider though is that both can be called via the command line and then it might not be clear whether you called the Conda or the installed version.
@christlet Hi, unfortunately, I currently only compile the one-click installer for CPU. I'll respond to your issue on GitHub with the technicalities and would be grateful for ideas or suggestions.
Our new study is out: Quantifying accessibility and incorporation of DNA origami with DNA-PAINT: Special thanks to @FSchueder Daniel and Philipp https://t.co/UcF3b1mP90
@mrpaulreynolds@kmdouglass I shifted from MATLAB to Python as well - and was at first missing the workspace explorer very much. For that I can recommend https://t.co/Ifj7ejK37R. To test little things I use Jupyter Notebooks, for everything else atom/sublime text + cmd.
@_Hrun_@mrpaulreynolds @flxrue @HohlbeinLab @christlet@HenriquesLab The current hdf5 format is optimized for speed but can do an option. What is the most commonly used format for TS? *.csv?
@mrpaulreynolds @flxrue @HohlbeinLab @christlet@_Hrun_@HenriquesLab Just updated the cmd wrapper for localize to be able to batch process folders. So far works for 2D. Details on
https://t.co/HjCehxJH6b
@flxrue @HohlbeinLab @christlet@_Hrun_@HenriquesLab@mrpaulreynolds If you have a movie that's split into many tif stacks that were created with uManager Picasso recognizes it and reconstructs it as one movie. For this only the naming needs to correct. Do you have several movies and want a molecule list for each of them?
@christlet@_Hrun_@HenriquesLab @HohlbeinLab @mrpaulreynolds Still want to do some performance optimizations + get loc precision values so that rendering is correctly, 3D is also another story, so still enough work to do. Quick test showed similar performance to MLE, speed was 15k/s (MLE), 64k/s(LQ) and 164k/s (Phasor) in parallel exec.
@HohlbeinLab @christlet@_Hrun_@HenriquesLab@mrpaulreynolds I'm going to check it out. I presume the Matlab Code of the SI is a good starting point? Looks straightforward, I guess I only need to adapt the z equation for calibration.
@christlet@_Hrun_@HenriquesLab @HohlbeinLab @mrpaulreynolds Recent versions can do 3D via astigmatism. GPUfit is implemented but not with the one-click installer, yet. Wanted to check again how to best do this.
@_Hrun_ @i_jayas @HenriquesLab @HohlbeinLab @christlet@mrpaulreynolds Great that you like it! Potentially localize has troubles for some locs with low SNRs. If you don't mind can you open an issue on GitHub / sent a dataset? I could try to write a fix.