plasmatorch

Self replication is not the goal. The long term vision is an automated factory with a wide variety of processes that use AI for generative design / process feedback. So, basically, if you can envision it, it can get built.

I work here. No grey goo, but lots of robots. My mental model for the long run goal is the Feed from Diamond Age.

Used to work in photonics. The star coupler portion is a Rotman lens, the whole thing is different (an AWG). It's used to spectrally separate an input (or in this case, inputs) across different outputs.

Aside from cranking the math, here's how I think about it: in the far field of a small aperture, the electric field has spherical phase (think expanding circles), and the field distribution is the Fourier transform of the aperture. A lens is an element that adds spherical phase - a plane wave passing through a convex lens now has a spherical phase distribution. So the lens focal point is now the tiny aperture in that system, and since the math works out the same no matter which way the light is going (reciprocity), the focal point is the FT of the field at the input of the lens.

Goodman is great, Hecht and Zajac covers more fun with optics at an intro level.

IIRC, it took about 50~100 W from a power supply to beam ~2.5W to a target antenna array, but that was fairly early on, so it's likely better now.

I worked on this project a few years back, on the power transmission bits. It's pretty cool from the engineering side; the mechanics need to be very light, and unfold into a large area. The beaming part is just a large phased array. The hardest bit is likely high efficiency, cost effective PV cells that can survive in space, but this is a field that Atwater has been in a long time, and is very, very good at.