Тёмный

Artificial Life Lightning Talk, 17-Jul-2020 

Dave Ackley
Подписаться 6 тыс.
Просмотров 1,8 тыс.
50% 1

Five minute lightning talk associated with the two-page paper "Best-effort Computing with SPOTs and Spatial Threads", available open-access at www.mitpressjournals.org/doi/...
0:00 Intro
0:11 The Glass Sandwich
1:02 Indefinite Scalability
1:34 Space-time Structuring
2:00 Challenges → Opportunities
2:44 Basic Demos
3:30 Spatial Map-Reduce
4:00 Multiscale Mobile Computing

Наука

Опубликовано:

 

2 июл 2024

Поделиться:

Ссылка:

Скачать:

Готовим ссылку...

Добавить в:

Мой плейлист
Посмотреть позже
Комментарии : 28   
@w43o21l2f
@w43o21l2f 3 года назад
The age of Smart Dust will be EPIC !
@QRebound
@QRebound 4 года назад
I always get so excited when we see demos of new structures!
@QRebound
@QRebound 4 года назад
Would also love to see you talk about how big you can make that map-reduce, and how accurate/fault tolerant it is in practice.
@DaveAckley
@DaveAckley 4 года назад
The census count is off every so often -- but because of a bug in the Emitter code, not the map-reduce or the mobility.
@dantescanline
@dantescanline 4 года назад
Fascinating! Excited to see more elemental mechanisms like this actually working as a proof of concept.
@DaveAckley
@DaveAckley 4 года назад
IKR!
@AndrewWalpole
@AndrewWalpole 4 года назад
Everything about this is so amazing! And wall structures as computational mechanisms ++
@DaveAckley
@DaveAckley 4 года назад
Good fences make!
@Adiounys
@Adiounys 3 года назад
Glad to see you are still alive :)
@DaveAckley
@DaveAckley 3 года назад
Most activity is over on the T2 Tile Project for the moment!
@dorianhill2480
@dorianhill2480 4 года назад
I notice the vertical and horizontal swap lines are taking turns to pass over the atom field. Could they be simultaneous and still robust? If I'm understanding it correctly then I really like how democratic the distribution of compute resources is. Very cool thinking Dave.
@DaveAckley
@DaveAckley 4 года назад
When the swaplines do cross, the horizontal ones have higher priority so the vertical just waits when that happens.
@dorianhill2480
@dorianhill2480 4 года назад
@@DaveAckley Thanks!
@PunmasterSTP
@PunmasterSTP Год назад
I almost thought I was watching at 1.25 or 1.5x, but then I realized it was just Dave's enthusiasm shining through. This, like all his other videos, was a treat to watch. Since I found his channel, I've been pondering the merits of efficiency vs. robustness, and I'm curious if anyone has thoughts on "conventional" software becoming more robust over time. I think of containers and Kubernetes, and being able to check on measures of health and redeploy resources strategically. Another example might be self-healing router networks. To me it seems like these types of solutions are filling the niche of the space Dave speaks about, thought I admit I'm only part of the lay audience. If anyone comes across this comment, what do you think? Are soft ALife and indefinite scalability the way forward, or could it be something else?
@abqsandy
@abqsandy 4 года назад
Hi Dave. Fascinating video (also great production - kudos to Jennifer?) /Stan Lee
@DaveAckley
@DaveAckley 4 года назад
Yeah, Juniper got the thing setup in no time flat! She's the best! juniperlovato.com/
@DaveAckley
@DaveAckley 4 года назад
And hey hi!
@otonanoC
@otonanoC 3 года назад
Dave Ackley please write a book so that i can buy and read it.
@DaveAckley
@DaveAckley 3 года назад
Thanks for the comment! It could happen. Spending too much time on the T2 Tile for now though..
@DanielSMatthews
@DanielSMatthews 4 года назад
Yeah, bring on the computiverse! Imagine the quantum foam particle and antiparticle pairs coming and going all throughout the universe, but who is to say that a particle needs to interact with its own pair, so long as then end state is balanced, this opens up the ability for information to propagate through the foam in the form of the patterns of creation and annihilation that are different from the simple 2 particle scenario. See how that works, the universe is the substrate for a 3D CA, but it is still entirely virtual. Do the maths on the data density and computational steps per unit time per unit of space volume, it is enormous.
@DaveAckley
@DaveAckley 4 года назад
MIPS per milligram!
@DanielSMatthews
@DanielSMatthews 4 года назад
@@DaveAckley Or better perhaps teraflops per nanolitre of volume of empty space, just stay away from event horizons or they will swallow up the propagating computation wave. Best place to run such code is in those great voids between the strings and membranes that galaxies tend to cluster along. Less black holes there, but you may need singularities of some form as I/O ports between the physical universe and the virtual compuverse. This may also be an explanation for why we have not detected any advanced civilisations despite SETI's efforts, they all go virtual eventually as it is the optimal computation method.
@DaveAckley
@DaveAckley 4 года назад
As long as it's not me updating the firmware!
@DanielSMatthews
@DanielSMatthews 4 года назад
@@DaveAckley Ah well that would not be as hard as it seems, you just need to know what read-write system to run on what small starting seed data for what number of steps so that the result is the much larger pattern required for your code. Assuming that CA outputs have every possible bit combination somewhere if you run them long enough. Hmm has that been formally proven as possible, I'm not sure.
@DaveAckley
@DaveAckley 4 года назад
Sounds very deterministic..
Далее
OVOZ
01:00
Просмотров 2,1 млн
Robust Local Synchronization - Research Notebook Video
37:58
Linus Torvalds On Future Of Desktop Linux
44:18
Просмотров 355 тыс.
Bill Gates Reveals Superhuman AI Prediction
57:18
Просмотров 63 тыс.
The Early History of Digital Computing
13:46
Просмотров 1,6 тыс.
Coding a Web Server in 25 Lines - Computerphile
17:49
Просмотров 325 тыс.
Introduction to the T2 Tile Project
7:00
Просмотров 3,8 тыс.
How I Got Started In Artificial Life
10:00
Просмотров 950
Will the battery emit smoke if it rotates rapidly?
0:11