Fanni, Julian, & Lisa (W5): B+ 1) Sensible overall model, nice use of pressure gradient to drive flow. nice use of 3) Demonstrated a very clear and strong understanding of hysteresis. The hysteretic results were interesting and well presented --> very good. 4) Good analysis, using ode45, fsolve, analytic solution ... but rushed presentation. Too many slides. The goal is to present a few things very well rather than many things less well. 5) You showed that the reversal was dependent on precipitation rate, which is great. But you did that by showing no reversal at a low rate and a reversal at high rate. The clear question becomes: what is the critical rate of precipitation for reversal. That question should have been answered. Kate & Amanda (T1): A- 1) Nice approach of adding noise to freshwater flux; needed more description of how noise was added (i.e., distribution etc), what relative time scale makes sense for the noise? what happens if frequency of noise varies? (you varied amplitude but not frequency) 2) Very well presented. Clear slides and nice presentation pace. 3) Interesting results regarding stochastic resonance 4) A figure that plotted the noise amplitude (or frequency) against some result (i.e., a reversal) would have been very nice. 5) Overall, great job! Vivek, Shivani, & Samir (W1): B+ 1) Nice fixed point analysis; We liked the the analysis of iterating through E-space to demonstrate that the only fixed point was always stable 2) Too many arbitrary "scenarios." There is a difference between sensitivity analysis (which you did well and called "scenarios") and running one creative scenario which represents an interesting case that is analyzed in detail (which you did not do as well as you could). Atmospheric diagram was a bit unclear. Main problem was that the model lacked some single original contribution by yourself in addition to that given in the project specification. Brendan, David & Jack (W2): A- (Model); B+ (Presentation ) 1) It is very good that you came up with your own model topic... very creative! 2) slides not good: too many words on slides, need to test presentation on a projector before giving the talk 3) The presentation was hurried... need a bit more practice 4) Nice presentation of unsteady temperature result 5) Result quite a-physical. Water boils at 373 K and 1 atm. Therefore, it is a little silly to have a solubility feedback causing the temperature to rise above 373 K. Remember that if the results are unphysical, you need to change the model, see our feedback slides. This was compensated by your using an original model, though. Jeanette, Moeko, & Stephanie (T3): A 1) Very clear presentation 2) Great model of the greenhouse opacity as a function of CO2 concentration. 3) Nice use of real data (i.e., IPCC) to calibrate greenhouse opacity equations. 4) Nice human response modeling, wonderful use of time-delay! 5) Note relevant timescales: if the timescale for your system to equilibrate is short relative to the timescale by which you are changing a parameter, then there is no reason to solve the time dependent ODEs. Rather, you could just report the equilibrium values for each parameter value. Justin Roth, YJ Yun, Luther Gatered (T4): B+ 1) Clever model of effort, we are impressed by your choice of a not trivial project to work on 2) We were not convinced that the fishing to extinction result is correct and that it results from a reasonable model formulation; It was difficult to tell from the presentation. Adam Traina, Jacki Stenson, Alexandra Stanck (T5): A- 1) Very nice model and presentation 2) Be certain to state model assumptions (i.e., no heat-flow between boxes). It is fine make such an assumption, but you must be clear about it so that your audience can judge the result. 3) The "oscillations" in your final figures were the result of numerical noise. You could have demonstrated that by doing a stability analysis. If the system was actually oscillatory, then we would have known from the stability analysis. In general, you need to test the sensitivity of such suspicious results to numerical aspects by increasing the tolerance required from matlab or decreasing the time step, and showing that the results don't change. Eddie Jou, Iain Kaplan & Chris Hartl (T6): A for project and B+ for the presentation 1) Nice presentation and description of model step by step 2) Interesting analysis to first maximize revenue, then minimize costs, and compare those to the maximized profit 3) The idea of having someone set price to max revenue ignoring cost is a bit too simple; when firms try to decentralize they give divisions (like sales) some sort of internal cost-of-goods to use in their calculations. We would also have liked to see you vary parameters such as the cost of inventory, and to provide an inutiion for the results. 4) The fonts were far too small, equations illegible, too much text on slides, no conclusion slide. Stephanie & Kevin (W4): A 1) very nice presentation was perfect; clear and convincing. 2) you did something nontrivial: the non linear case is basically intractable if we have a cost of fishing, so setting cost=0 was unavoidable when you got to this part. you made the linear case look easy, although it was non trivial. nice! Cat, Andy, & Sandro (T2): B+ 1) nice sensitivity contour plot of temperature as function of two parameters. 2) could have added a diagram to illustrate how the model was derived; variable names changed from slide to slide (\epsilon_1 and \epsilon_2 became \epsilon_2 and \epsilon_3; (the same symbol used for density for different components). 3) it was good to consider extremem cases such as no atmosphere, but your analysis could have been more careful (you write that the system could not remain stable without the atmosphere, and this is not correct) 4) It was a good idea to add a natural variability effect, although making it stochastic would make more sense, and it didn't seem to add much to the analysis. 5) too many scenarios. better do just one, and do it well. see our feedback regarding the difference between scenario and parameter sensitivity.