On Tuesday, June 10, 2025 at 10:01:38 AM UTC-5 Edward K. Ream wrote:

This paper <https://www.nature.com/articles/s41586-021-03544-w> about AlphaChip 
<https://deepmind.google/discover/blog/how-alphachip-transformed-computer-chip-design/>mentioned
 
hypergraphs <https://en.wikipedia.org/wiki/Hypergraph>.


Some highlights of the AlphaChip paper:

Training placement policies that generalize across chips is extremely 
challenging, because it requires learning to optimize the placement of all 
possible chip netlists onto all possible canvases. Chip floorplanning is 
analogous to a game with [a state space] of 1000 [factorial] (greater than 
10**2,500), whereas Go has a state space of 10**360.

The underlying problem is a high-dimensional contextual bandits problem 
<https://towardsdatascience.com/an-overview-of-contextual-bandits-53ac3aa45034/>
 [not 
to be confused with a multi-armed bandit problem 
<https://en.wikipedia.org/wiki/Multi-armed_bandit>] but, as in prior work, 
... we have chosen to reformulate it as a sequential Markov decision process 
<https://en.wikipedia.org/wiki/Markov_decision_process> (MDP), because this 
allows us to more easily incorporate the problem constraints as described 
below.

To address [the challenge of an enormous state space] we first focused on 
learning rich representations of the state space. Our intuition was that a 
policy capable of the general task of chip placement should also be able to 
encode the state associated with a new unseen chip into a meaningful signal 
at inference time. We therefore trained a neural network architecture 
capable of predicting reward on placements of new netlists, with the 
ultimate goal of using this architecture as the encoder layer of our policy.

Seems like brilliant science and mathematics to me :-)

Edward

-- 
You received this message because you are subscribed to the Google Groups 
"leo-editor" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/leo-editor/79d3c1fd-52f0-495e-8de3-eac6409c8635n%40googlegroups.com.

Reply via email to