This reminds me of a technique used in machine learning for dimensional reduction. If you have data with a very high dimensionality, you can reduce the dimensionality by projecting the data onto a random set of basis vectors. It tends to be pretty good at preserving the norms between the data points. One example would be if you have a discrete graph/network with a million points. The adjacency vector representatuon for each point would have a dimensionality of 1 million. You could then use the technique to reduce the dimensionality to only a few hundred pretty easily and generate an embedding of the graph.
Very pretty video, did not expect such math to be applied directly to such problem. Overall, calculating efficient approximations of values are beautiful in how they may be counterintuitive at first, yet absolutely logical when you think about them
Thanks! I couldn’t find problem anywhere except the main contest (Moscow Regional Qualifiers 2021), where you have to be pre-registered to upsolve, sorry, can’t help you here…
Thanks! Since it was my first project with manim all source code is such an atrocity that i deleted it once I uploaded video so I couldn't translate mistakes from that code to future projects.
This reminds me of a technique used in machine learning for dimensional reduction. If you have data with a very high dimensionality, you can reduce the dimensionality by projecting the data onto a random set of basis vectors. It tends to be pretty good at preserving the norms between the data points.
One example would be if you have a discrete graph/network with a million points. The adjacency vector representatuon for each point would have a dimensionality of 1 million. You could then use the technique to reduce the dimensionality to only a few hundred pretty easily and generate an embedding of the graph.
Very pretty video, did not expect such math to be applied directly to such problem. Overall, calculating efficient approximations of values are beautiful in how they may be counterintuitive at first, yet absolutely logical when you think about them
What a beautiful problem! And a very well-made video
Very interesting ! Thanks
noice solution
Great video!
very cool!
brings me back to my codeforces days man, super fun
4:32 optical illusion
Reminds me of the Steinhaus Longimeter
Damn! that is pretty!
УРА УРА ДАВАЙ ДАВАЙ
Really liked the video! Could you give a link to the problem? on icpc website or some online judge
Thanks! I couldn’t find problem anywhere except the main contest (Moscow Regional Qualifiers 2021), where you have to be pre-registered to upsolve, sorry, can’t help you here…
Really good video. Can you share the source code for the video?
Thanks! Since it was my first project with manim all source code is such an atrocity that i deleted it once I uploaded video so I couldn't translate mistakes from that code to future projects.
Too strong russian accent, the rest is great