Conversation

Notices

1. Embed this notice
   niconiconi (niconiconi@mk.absturztau.be)'s status on Wednesday, 01-Feb-2023 15:11:59 JST niconiconi

   In electromagnetic simulations, all waves only exist in a small simulation box, so you need to do something when they reach the edge, this is the physical intuition of boundary condition in PDE. For E&M, people use a funny trick called the Perfectly Matched Layer. Instead of a real mathematical boundary condition, PML behaves more like a model of an anomalous material with strange nonphysical properties that magically absorbs all the E&M waves perfectly. Someone needs to write some SCP fiction about how The Factory made PML...

   • Puniko ? likes this.
   • Embed this notice
     Puniko ? (puniko@mk.absturztau.be)'s status on Wednesday, 01-Feb-2023 15:12:42 JST Puniko ?

     in reply to

     @niconiconi

   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Wednesday, 01-Feb-2023 15:17:20 JST niconiconi

     in reply to

     • Puniko ?

     @puniko It makes me wonder what kind of funny "anomalous" materials are invented in other fields of physics for computer simulations. For example, what funny things do people use in fluid dynamics?

     Puniko ? likes this.
     Puniko ? repeated this.
   • Embed this notice
     malevolent dictator for life (astrid@fedi.astrid.tech)'s status on Wednesday, 01-Feb-2023 15:27:36 JST malevolent dictator for life

     in reply to

     • Puniko ?
     @niconiconi @puniko lemme tell you about a little thing called molecular dynamics
     
     So imagine you wanted to simulate how a protein or something moves over time. Nuclei and electrons interact with each other with a bunch of quantum fuckery, but the problem with calculating that quantum fuckery is that it's computationally expensive (gotta calculate eigenvalues for stupidly big matrices) and that's not feasible for big proteins.
     
     On the other hand... you could just pretend every atom is a massed ball, and every bond is a spring, and have them apply forces to each other very, very cheaply. The worst part about all this is that it's surprisingly accurate, too.
     Puniko ? likes this.
   • Embed this notice
     Puniko ? (puniko@mk.absturztau.be)'s status on Wednesday, 01-Feb-2023 15:43:33 JST Puniko ?

     in reply to

     • malevolent dictator for life

     @niconiconi @astrid@fedi.astrid.tech TIL
     didnt know this bit

   • Embed this notice
     niconiconi (niconiconi@mk.absturztau.be)'s status on Wednesday, 01-Feb-2023 15:43:35 JST niconiconi

     in reply to

     • Puniko ?
     • malevolent dictator for life

     @astrid@fedi.astrid.tech @puniko@mk.absturztau.be Back in Maxwell's time, the concept of a "field" was nonexistent, so Maxwell's equations were also originally derived in terms of gears and springs. This time-tested trick can probably model most if not all differential equations...

   • Embed this notice
     Puniko ? (puniko@mk.absturztau.be)'s status on Wednesday, 01-Feb-2023 15:49:40 JST Puniko ?

     in reply to

     • malevolent dictator for life
     • Twinkie Pie [OXIDIZED]

     @astrid@fedi.astrid.tech @cpluspatch@fedi.cpluspatch.com @niconiconi@mk.absturztau.be tbf video game physics are not focused on being accurate, but on being believeable while being efficient to calculate

   • Embed this notice
     malevolent dictator for life (astrid@fedi.astrid.tech)'s status on Wednesday, 01-Feb-2023 15:49:43 JST malevolent dictator for life

     in reply to

     • Puniko ?
     • malevolent dictator for life
     • Twinkie Pie [OXIDIZED]
     @cpluspatch @niconiconi @puniko anyways, this MD stuff isn't even the worst of it. All the video game physics engines in the world have a fairly simple core (Euler approximate acceleration, velocities, positions)... but what do you do when two objects collide? If you just calculate things the normal way as covered in Physics 101, contacts would be extremely unstable and things would randomly fly away (think goat simulator but worse; I know because I tried making a physics engine once...) There are just so many hacks used to approximate and apply contacts and friction, it's a wonder it all works.
   • Embed this notice
     malevolent dictator for life (astrid@fedi.astrid.tech)'s status on Wednesday, 01-Feb-2023 15:49:45 JST malevolent dictator for life

     in reply to

     • Puniko ?
     • malevolent dictator for life
     • Twinkie Pie [OXIDIZED]
     @cpluspatch @niconiconi @puniko pharma companies probably have petabytes of MD simulations hoarded in their data warehouses
   • Embed this notice
     malevolent dictator for life (astrid@fedi.astrid.tech)'s status on Wednesday, 01-Feb-2023 15:49:47 JST malevolent dictator for life

     in reply to

     • Puniko ?
     • malevolent dictator for life
     • Twinkie Pie [OXIDIZED]
     @cpluspatch @niconiconi @puniko oh yeah and the classical physics approximation of molecular dynamics isn't just theoretical, biophysicists run them all the time
   • Embed this notice
     malevolent dictator for life (astrid@fedi.astrid.tech)'s status on Wednesday, 01-Feb-2023 15:49:48 JST malevolent dictator for life

     in reply to

     • Puniko ?
     • Twinkie Pie [OXIDIZED]
     @cpluspatch @puniko @niconiconi approximations and analogies are fun! Wait until I tell you about how fluids, pipes, wheels, and spring-mass systems all behave identical to electric circuits, and it's a common practice to redraw those systems as electric circuits, do electric circuit analysis, and translate it back into the original system
   • Embed this notice
     Twinkie Pie [OXIDIZED] (cpluspatch@fedi.cpluspatch.com)'s status on Wednesday, 01-Feb-2023 15:49:49 JST Twinkie Pie [OXIDIZED]

     in reply to

     • Puniko ?
     • malevolent dictator for life
     @astrid @puniko @niconiconi fuck you for telling me that