Connecting AI and Physics
This section is to discuss the origins of AI, machine learning and its origins. We will talk about and more importantly post evidence to illustrate that machine learning is not anything new. The ideas and algorithms put forth are simply the same set of algorithms we had going back to Gauss. What has changed is we now have the ability to perform calculations in nanoscale time frames. This has allowed the use of older algorithms which would take to long to find solutions by hand due to the number of iterative calculations required. For those of you who are currently students will know there is analytical solutions to physics problems and there is the differential equations, both pose a solution but the differential equation must be step wise iterated. That is all that machine learning is. These few blog posts, I hope will serve as a means to remove the black box, allow you to see that it’s not the algorithms, but the compute power that is new. Neural networks work not because they are some arbitrary thing we dont know about. No, its the exact opposite, for decades we have been searching for ways to model the universe, and time and time again the simple linear, to maybe some non linear approximations always end up being the solution. The equations that describe waves appear in diffusion, you just change some of the parameters. Those familiar with the boundary value problems in undergrad physics know that you get this crazy answer that pops up in every book: X=Asin(x)+Bcos(x)+csin(x)+…..
This can be re written as a forrier series. Foureir series goes back to statisticisc. It’s just a linear approximation for complex systems. We’ve studied this in physics for decades. In these posts I will provide content others have done, that directly illustrates the math behind neural networks is exactly that of n body simulations. This isn’t unknown, it’s in the literature, but often over looked.
I hope this information can help those who are performing research to better understand what the tools are and how they can be used.