NOTE: These are odds and ends for me to think about.
Old Capstone page
I believe analog is form of natural computation, suited to the types of things that we, as humans, are good at. I further believe that we are so used to throwing hardware at a problem that we sometimes overlook alternative solutions. My research seeks to identify the role of analog computing in a digital-dominated world.
Where does analog fit among alternatives as varied as neural networks, quantum computing, complex systems, artificial life and/or intelligence, and others? How do we leverage this niche? What tools are required to approximate the things we humans seem to do with such ease? And heck, what are those things anyway? What is the underlying theory of analog?
As I move through my capstone year, I hope to address some of the questions. I hope to develop the tools necessary to ask new questions. I hope my work advances the field of analog, and, perhaps, encourages others to investigate what I find to be a very promising computational paradigm.
Documentation
- Capstone Proposal
- Final paper (draft) Δ (PDF)
- Photographs of the Extended Analog Computer
- Analog XOR, the analog "Hello World" (in progress)
- Analog Hardware Abstraction Layer (in progress)
- Ideas
- Sandbox, for things I'm working on
Projects
- EAC Toolkit
- Evolution in Analog
- JEAC
- ALife Independent Study
- Old: ALife Project, interesting conjecture
- Old: Notes from Alife
External Links
Links
Genetic programming, etc.
2006.10.02
How do I classify genetic algorithms in a broader sense? I've heard them referred to as:
- Hill climbing algorithms
- Stochastic algorithms
- Heuristic algorithms
- Search algorithms (of course)
Which of these are subtypes and which are supertypes? Or are they all the same thing? What of these other terms:
- Simulated annealing
- Swarm/flocking algorithms
- Stochastic beam search
- Optimization
And so and and so forth...
Analog information models
2006.09.05
How does Shannon information apply to analog? Isn't Shannon's theory explicitly based on digital communication? How do other information models apply to analog? Prof. Mills was talking about fluid information last year. I should follow up on that.
Fully analog systems
2006.08.09
The net-EAC and uEAC are digital-analog hybrid approximations of true Rubel-Mills EACs; they are research tools. However I think some of the more interesting applications ultimately lie in the domain of purely* analog systems. Such a system should require only a conductive sheet, and inputs and outputs that are modulated to the range of the sheet's voltage gradient. As far as I can tell, the only trick is creating modular, standalone LLA units, for attachment to the sheet. Who knows, analog "programming" could, one day, be as easy a lego set (and as accessbile too).
I should ask Bryce about this.
[*] to the degree possible
The nature of computation
2006.07.24
Tonight Marla gave me a book, Computer Basics, from 1985. It is surprisingly interesting as it relates the history of computing from "the knee of the curve" of the personal computer. That is, it's a fairly complete history of computing, but without the hubris that generally comes along for the ride. It isn't laden with the attention to current events that usually dominates historical recapitulation. History, as we know, is written by the victors. Computer Basics comes so soon after the battle that the perspective is fascinating.
More importantly, it chronicles the rise of the computer as a general purpose device, not as an analog or digital device. The book talks about the motivations of Babbage, Boole, Bush, and Shannon. It talks about the origins of logic, its migration to mathematics, and the rise of information theory. Everyone seems to be talking about the importance of binary on its own merits, beyond simply a good match for electronics. True it is simpler than mechanical base-10 machines, but what is the unifying theme here? What do binary, logic, mathematics, mechanics, language and human thought, automation, and computation have to do with one another? What is the deeper thread that ties it all together?
I have a feeling the answer will say a lot about the relevance of modern analog. We have nearly 100 years of experience and theory now; I definitely feel it is worthwhile to retrace our steps. Hmmm.
Abstraction Hierarchy
| (1) | Theory |
| (2) | Hardware |
| (3) | Interface |
| Configuration | |
| Evolution | |
| (4) | User Interfaces |
Links to Consider, from Mills
- KLMs
- Question: What is the significance of "KLM" vs. "EAC"?
- 42 Questions
- Question: How does my work fit into these categories?
- Analog Simulation, a recommended, out-of-print text
- Analog/Robotics
Implementation Questions
How can I modularize the script?
- EacUtilities
- EacInterface
- privatize hash arrays
- implement interface as a jumptable
- implement up-arrow completion [?]
- EacHardware
- modularize
- rewrite flags
- EacGA
- EacLogger
Flocking: