Final projects posted here
Work on your final project!
Final exam week: Projects due. Presentations Noon-3pm Weds 12/11.
Wed 12/4: Progress report #3.
Mon 12/2: No class---work on your project.
Wed 11/27: a working session...I will be there to discuss your project.
Mon 11/25: Present project progress report #2.
Wed 11/20: Working session to discuss projects.
Mon 11/18: Present project progress report #1.
Wed 11/13: Submit project proposal.
Mon 11/4 Read Skordos and Zurek, "Maxwell's demon, rectifiers, and the second law: Computer simulation of Smoluchowski's trapdoor." pdf
Part 1 [for Monday]: implement a simple "SRC-SNK" diode demon. See diagram (png Thanks Jim Youngquist!). Compare time to equilibrium for left tank initially full (and right tank initially empty), vs right tank initially full (and left initially empty). Attempt to maximize asymmetry in time to equilibrium (perhaps by using multiple diodes in parallel).
Part 2 [extensions, for next assignment, unless you want to keep going now]:
(1) What happens if you start the system in equilibrium initially? Can your demon move the system away from equilibrium?
(2) Replace the flow of particles from SRC to SNK cells with a tank full of gas particles that drains in to an empty tank.
(3) Build a diode (i.e. a device which results in asymmetric left-to-right vs right-to-left flow rates) using just walls (no particle streams as in the diode of part 1)...think about funnel- and maze-like structures.
Weds 10/30 Read Bennett, "Thermodynamics of Computation, A Review." pdf
Mon 10/28:
1. Read Landauer, Irreversibility and heat generation in the computing process pdf
2. Improve documentation of your gates from Monday
3. Improve your gates; use other people's gates if they will help; try to connect multiple gates into more complex circuits.
(a) Can you build a half adder?
(b) Can you make a feedback connection? Try to build a sequential (non-combinational) circuit such as a flip flop.
Mon 10/21:
1. Read 2nd half of Feynman Reversible computation and thermodynamics of computing pdf
2. Use reflecting walls to implement delays of 2,3, or N (your choice) in LGA.
3. Implement a non-trivial (ideally, computation universal) gate in LGA. Consider attempting Fredkin gate, switching gate, NAND, CNOT (Controlled NOT, aka reversible XOR), CCNOT (Controlled Controlled NOT aka Toffoli gate), NOT, FANOUT, AND, "SPY".
4. (optional): Connect more than one gate. (Stretch goal)
Remember that other students will want to use the nicest designs produced!
Some documentation on how walls work (and example of unit delay) pdf
Wed 10/16:
Read 1st half (up to 5.2.3, p 160) of Feynman Reversible computation and thermodynamics of computing pdf
Think about whether it is possible to build an AND gate using the 2 particle LGA collision (and any mirrors you want). By my definition, the output of the AND gate has to be encoded at a single point in space. (It's not OK to say, "if there are particles at these TWO locations then the output of the gate is TRUE.") Send me an email before class saying either "yes it is possible to implement AND using 2 particle LGA collision" or "no it's not possible." Defend your email in class.
Mon 10/14:
(1) Implement two flow experiments using LGA (remember to pull latest code from bitbucket repo). Can you create vortices? Submit documentation via Dropbox.
(2) Billiard Ball Computation: Read Fredkin Conservative Logic
(3) Supplemental (profile of Fredkin): Flying Solo (bad html excerpt) or pdf of complete article. Want to procrastinate? Might as well read this.
Wed 10/9: LGAs: Read Hasslacher (Discrete Fluids) and Moore Lattice Gas Prediction is P-Complete
Mon 10/7: Read Atkins Pt II. Using simulator, design your own container (using a paint program or code). Repeat the pressure-volume experiment from 9/30 class (see notes) and compute the constant of "kT" (i.e. PV/N). In class we will compare values.
Wed 10/2: Read Atkins Pt I and install course software. Good idea to get started early on Monday's coding assignment.
Mon 9/30: Read Demons, Engines,... and Information Physics...
Instructor: Joshua Smith
Email: jrs at CS
Meeting place: CSE 203.
Time: MW 10:30-11:50
Office hours: CSE 556, by appointment
This course is a self-contained introduction to the Physics of Computation, with applications to energy harvesting and low power computing. In recent years, energy has become an increasingly important consideration in computer science, electrical engineering, and computer engineering. The energy efficiency of computation has been improving exponentially for decades, which is creating new opportunities (such as perpetually operating systems that harvest all the energy they need from ambient sources) and new challenges (such as removing unwanted heat from increasingly dense microelectronics).
The course surveys the physics of computation literature, in order to examine the fundamental physical limits to the efficiency of energy harvesting, and to the energy efficiency of computing. It turns out that irreversible computing operations have an intrinsic cost, while reversible operations do not. This understanding has successfully resolved the paradox of Maxwell's Demon, a hypothetical energy harvesting system that had threatened to violate the physical law forbidding perpetual motion machines. The paradox of Maxwell.s Demon arises when the amount of harvestable energy is actually zero; we will examine the so-far neglected case in which harvestable energy is small but non-zero.
Students will use a specially developed simulator to perform numerical experiments on the course topics. It is expected that some student projects will use the simulator to perform computational experiments on the design and test of structures such as Maxwell's Demon / energy harvesters, reversible computers, and cooling mechanisms. Students may also propose other projects, including practical energy harvesting experiments.
The course may appeal to students interested in energy harvesting, computer architecture, low power communication, theory, computer engineering, information theory, statistics, neural computing, and quantum computing. Prerequisites: open to all graduate students in CSE and EE. Undergraduates and others by permission of the instructor.
Feel free to email me with questions about the course.
The course work consists of
25%: reading the articles and participating in class discussions
15%: presenting one article to the class
25%: homework assignments using the simulator, and
35%: a final project, with write up and presentation.
All of the books are optional. However, given the very reasonable price, you should strongly consider getting the first one ("The Laws of Thermodynamics").
The Laws of Thermodynamics: A very short introduction, by Peter Atkins, Oxford University Press, 2010. Cost: $9.16 Buy on Amazon
Feynman Lectures on Computation, Edited by Tony Hey and Robin W. Allen. Cost: $45.14 Buy on Amazon
Maxwell's Demon 2: Entropy, Classical and Quantum Information, Computing, Edited by Leff and Rex Buy on Amazon
Figures