Difference between revisions of "CS382:Unit-descriptions"
|Line 7:||Line 7:|
* Relevant and applicable
* Relevant and applicable
* Can do something that can't see easily in the real world because too big and too slow
* Can do something that can't see easily in the real world because too big and too slow
Revision as of 11:29, 21 January 2009
Environmental / ecosystem monitoring. temperature levels / air quality. modelling specific regions vs global modelling
- Talk about pollution, global warming
- Inspired from Second Life, largescale weather-based model - has this been finished??
- Relevant and applicable
- Can do something that can't see easily in the real world because too big and too slow
- Could do something with displaying raw data from Netlogo and then work with graphing, stats, or something similar
- Could compare this to a big experiment by one of the national labs or similar - see if your results follow the same trend as the national lab's
- Kind of vague, needs more details
- Difficult to see this change in real world, difficult to validate
- Existing Netlogo model similar to this (climate change model)
- Nice to find someone else's large scale model and then replicate
Socioeconomic modelling. what happens with changes in legislation eg taxes, min wage. model of population behavior. (agent)
- Computational Sociology (wikipedia)
- brian castellani sociology and complexity web
- Center for Models of Life - CMOL: Models / Interactive Java Applets
Chemical modeling - Using resources to model molecules and looking at the kind of molecule a certain combination of atoms make. Goes into using this kind of modeling to make new drugs. Would include a lecture that harks back to an (most likely) earlier lecture about how modeling is the third leg and then go into how drug companies have saved billions of dollars by being able to get through the early phases of drug development with just models. Would have a lab to come up with a feasible possibility for a new type of drug, and what it would/could possibly treat. Need to look into the availability of a computational chem server, as it might not be for use for beyond high school.
- Molecular modelling (wikipedia) This has a whole host of potential softwares to use for this.
- Ghemical homepage This is a software listed on the wiki linked above that I have heard of and have any kind of familiarity with. Though there are certainly many others and it would be a good idea to look at all or most of them in a fair amount of depth before making any decisions as to what software to use.
- Maybe able to get the lab or software from Shodor
- Accessible and immediate feedback, understand why would want to do this in the real life
- Hits chemistry (discipline we want to include)
- Potentially too complex
- No "numbers", more abstract
- Not immediately comparable to a real world thing, hard to have a real world lab
- Walk through making a drug that already exists.
- Find similarities in a drug family
Possibly a more structured/simpler version of the parachute lab done in CS290. Obviously having them perform the parachute drops and such would not be feasible but we could have them start out with the videos and some figures and they would do the netlogo aspect of the lab for themselves. (Different sized parachutes attached to different weights - dropped them and measured them, and then tried to model it and decide how fast different weights should drop with different parachutes.)
- Simple enough that they can create their own model, not just look at other people's models
- Important that we have at least one unit where they create their own model
- Lots of math and physics
- Could create own model in Netlogo (or at least adjust a toy model already created)
- Physical drops took a very long time to do.
- Would need to be able to some programming on a very basic level if they're going to create their own.
- Part of Charlie's "prove gravity" lab?
- Maybe we shouldn't have them create their own model but rather adjust sliders, etc. or at least have most of a toy model made
- Modeling a woodwind instrument - Could demonstrate how changing the hole positions would affect pitch/timbre. The paper below outlines a technique for modeling the tonehole attempting to account for the half-covered state present in real world performance. Might need some introduction to the physics of music, general properties of waves.
- Modeling airflow in a wind tunnel - The class could look at the complexity of modeling airflow around a simulated object. How detailed and complex can a virtual object become before modeling airflow around it becomes computationally infeasible? This unit might be able to use some of the physics background introduced in the Woodwind modeling unit described earlier. The fluid dynamics aspect might add a significant learning curve, however.
- Similar to Bryan's idea;Aerodynamics of an object:in try of reducing air drag with simulating different versions and simulations of a virtual object in air tunnel; improving it's aerodynamics - in purpose of enabling faster traveling but again trying to keep it safe. Air tunnel simulation basically - how would an object react depending on the material used and its shape/structure? How much does air affect its possible movement? And the direction of the 'blowing' and the movement of the object affects the simulation result. The idea started watching F1 'cars' - which try to use lighter materials so the weight doesn't affect speed, but also to improve very crucial factor - the aerodynamics of the vehicle. Must point out that this model would be based on the physics principles-so its that is the science side of it.
Wikipedia tells it all about the aerodynamics, one should know. Some ideas about modeling could be obtained.
At the lower part of the site - it talks about visualizing the results and the whole simulation of the wind tunnel. Interesting.
This link provides multiple links (e.g. http://www.ansys.com/products/cfx.asp Ansys - fluid dynamics) of the software which might be used for modeling such situations described. Honestly, didn't have bandwidth / speed to check the links; and not even to read the wikipedia that I posted; but information is accurate.
- I also found the idea of observing the behavior, in the lunch rooms and similar objects, of different people very interesting. Basically simulating 'green' and 'red' dots, having social even together- more various 'colors' are possible depending on the circumstances. The purpose would be to try to predict/simulate and conclude on what basis people divide/group themselves in everyday actions and needs. Regardless of the place, the mentality of people shows in this situations; which has the psychological aspect to be thought in the course. More about human behavior could be learned, if all important factors are included to the simulation. It might be wrong, but I see it as firstly trying to include the statistical aspect of it; counting the different dots; and bring some statistic about it-trying to figure out and predict the chronology of the events. Very important to include all the info that we already know, (as I said) numbers, age, possible past behaviors, and similar.
- P.S.Hope it is good effort - I still am confused, I admit-but I am getting there.
- Protein Folding - From the PDB to an image of a simple folded protein. Uses existing models and simulation tools to take the description of a protein from the Protein Data Bank and generate a simulated protein from that. Running existing models, scientific data store, visualization. Requires a lecture on the underlying physics/chemistry/biology.
- Fermiproblems - Use fermiproblems to encourage students to be comfortable making estimates and discovering ways to estimate with only limited data available. Examples are available here: http://iws.ccccd.edu/mbrooks/demos/fermi_questions.htm
- A list of Problems is available here: http://www.physics.umd.edu/perg/fermi/fermi.htm
- Emphasis could be put upon using available sources to find figures as well as how to make estimates where figures are not available
- Problems could be picked which are relevant to the models which the students will later be constructing as a class.
- Students should be taught how to show their reasoning behind extrapolations and to become comfortable with doing so.
- The instructors should have worked through examples showing a complete model and demonstrating which information is necessary for a ballpark estimation and which is not.
- Assembling a phylogenetic tree - Look at models available from the bioinformatics toolbench and experiment with different ways comparing protein similarity. This could be tied into a lecture on protein folding as protein homology like indicative of phylogeny.
- Lightning - Simulating a thunderstorm, including atmosphere, charged cloud, building on earth and etc. Estimating which area will get damages.
- Fermentation food - Set up food and bacteria (e.g. milk and lactobacillus) and simulate how they get fermented and rotten.
- Hurricane tracking-- like the data that was shown to CS128, we can have them simulate the paths of hurricanes, severity, etc.
- Tsunami simluation-- similar to the above, except... with tsunamis. [edit: upon further investigation, tsunamis seem less model-able and are more a matter of using sensors and communication systems to detect them and broadcast warnings]
- If we're trying to generate interest, I think it would be good to initially give them huge models of catastrophic events that, to put it crudely, feel "awesome"; Just something to give them the ability to say "Well, WE just simulated a friggin' hurricane that tore across the east coast of the U.S." It might help plug them into the class.
- http://en.wikipedia.org/wiki/Tropical_cyclone_forecast_model (has descriptions of various models for tracking cyclones and hurricanes)
- Population dynamics, use a predator-prey model and talk about the underlying interaction between two (or more) species. Could do this with agent-based modeling and/or system dynamics modeling. We could talk quite a bit about ecology and species interactions and things that affect the stability.
- Cladistics (constructing a phylogenetic tree). Start with a common protein or gene amongst a group of organisms and construct a model for how they diverged evolutionarily based on evidence from that gene. This could also start simpler with looking at homologies and analogies and constructing a tree based on that (old way), then moving to the more cutting research with the genes. Would give an opportunity to talk about the computing power behind searching large databases of proteins or nucleotides.
- I was thinking of doing something with biology. This may be too complicated, given current technology, but if we could take a model of a cell and start playing with it, maybe tweaking a few of the parts, and see if that would have any effect on how that alters the function of the cell in the body?
- Alternately, we could do something with engineering. Develop structual models, and see what sort of crazy structures you could make that would stand on their own. There's already some rudimentary stuff to do this with, such as Bridgebuilder, but if we could possibly do something more complicated, I think that would be awesome.
- Traffic flow modeling - This could be systems dynamics or agent-based (see http://en.wikipedia.org/wiki/Traffic_flow) and part of the Measuring the World unit. The simulation would involve modeling a section of traffic to see results of uses of traffic control devices. The validation/verification part of this could be to observe an actual section of traffic here in Richmond.