4 ms·
I'm not a physicist like the Ars writer, but I always kinda though that computation requires energy consumption and non-linear elements to interact with a signa
by marshray 8y ago
I'm not a physicist like the Ars writer, but I always kinda though that computation requires energy consumption and non-linear elements to interact with a signal. The only energy source in this experiment are the Wifi transmitters and I don't see any nonlinear elements.
The paper abstract seems to define 'computation' as performing a Fourier transform like an optical lens (I'm imagining a prism separating the wavelengths). It also says "We can exploit the linearity of the computation operation Y = GX in which the entries of Y are independent from each other."
So I don't see how this meets any useful definition of "coprocessor". Can you use it to perform a NAND operation?
- JPLeRouzic 8y agoNo it does not, it is the Ars journalist that write about coprocessor, the Arxiv article is about infering the matrix G, nothing else. They do it in a quite clever way.
- LolWolf 8y agoYou have to be careful about what you mean by "linear"—linear programs are P-complete, and all binary operations are linear over F_2 (including NAND, etc), since all functions are polynomials over finite fields and the largest polynomial over F_2 is of degree 2-1=1. There are many linear elements which can generate arbitrary circuits and hence can (in principle!) implement any finite circuit (e.g., processor). I do agree, though, that in this case it's unclear how the 'coprocessor' connection arises from the original statement given in the paper.
- marshray 8y ago"Linear in F_EM" :-)