wmayner / pyphi

A toolbox for integrated information theory.
https://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1006343
Other
368 stars 98 forks source link

where is PQR cause-effect space on fpga with φ? #33

Closed peepo closed 4 years ago

peepo commented 4 years ago

Will, I understand the mailing list might prove more suitable. #32

Where is a simple worked example of PQR using gates on fpga verilog or VHDL and demonstrating a value for φ

it would be helpful to me if both feedback and feedforward-only demonstrate same output with different φ!

apologies but I find python gets in the way of my initial understanding.

many thanks

wmayner commented 4 years ago

Briefly, there is no worked example of 𝚽 for a system implemented on a FGPA to my knowledge. You could attempt to model such a system and then use PyPhi to calculate its cause-effect structure, but I suspect it would require more than ~12 nodes, at which point the calculation begins to be intractable due to the exponential time complexity of the algorithm.

It is certainly possible to design two systems, one recurrent and one feedforward, that are functionally equivalent and have different 𝚽 values.

You're welcome to post to the mailing list (which should be accessible again) for further discussion.

peepo commented 4 years ago

apologies, I did not make this issue clearer...

PQR was intended to suggest 3 nodes, as referred to in various academic papers and the recent book by Christof Koch**. in this case or, copy, xor respectively.

it is not necessary to calculate φ on the fpga, but a simple worked calculation for 3 nodes would help.

**The wiring diagram in Fig. 13.2 on page 144 does not have input or output as I understand it, which may be intended to emphasise the Wholeness, but I am currently confused.