AMS 553 / CET 554 / CSE 529 / MBA 553                                                                                                                                                                     Fall 2008

AMS 553 / CET 554 / MBA 553    Project


Due Wednesday, December 17th.  (hard copy of Project Report due during Final Exam period)



AMS 553 / CET 554 / MBA 553 :

Use the model you developed in Assignment 5 to obtain the following steady-state performance measures :

   (i)  The proportion of NIA patients who had to wait, when they arrived, before first receiving attention from a doctor.

  (ii)  The 99th. percentile for the initial waiting times of such NIA patients who had to wait, from when they first arrive until they start receiving attention from a doctor.

 (iii)  Mean initial waiting times for the CW patients, from the instant they arrive until they first start receiving attention from a doctor.

 (iv)  The variance of the initial waiting times for the CW patients, from the instant they arrive until they first start receiving attention from a doctor.

  (v)  The mean total times each of the NIA and CW patients spend from the instant they arrive until they are finally discharged.

 (vi)  The proportion of time each of the two doctors are busy.


You should go about collecting these statistics as follows.

 (1)  Independent Replications :  Make 10 independent runs. Use the ‘transient’ phase length you determined in Assignment 5. You will still need, though, to determine an adequate length for the ‘steady state‘ phase of your runs. Calculate symmetric confidence intervals at the 99% confidence level for each of the performance measures in (i)-(v)above :  in accordance with the Bonferroni Inequality, the joint coverage of your confidence intervals will thus be higher than the 90% confidence level (more precisely, no worse than the 94% level).

 (2)  Batched Means :  Make a run with 10 batches using the same length for the transient phase as in (1) above, and with each batch equal in length to the steady-state length used there. Calculate the same performance measures.


In addition to your code (which you should submit electronically – see below), submit a Report in hard copy that describes how you went about doing your experiments; and presents the data generated, confidence interval calculations, and results, etc.. The Report should also address the following issues :

(a)  Explain and comment on your choice of length for the steady-state phase in (1) (this involves including some discussion about what you did for Assignment 5).

(b)  Compare the performance measures obtained from (1) & (2), and comment. Can you reach any conclusions about the adequacy of your choice for transient phase length ?  What about the adequacy of your batch size in (2); and/or of your steady-state phase length in (1)? Explain how you would go about correcting the discrepancies found (if any) between the performance measures obtained from (1) and from (2) .

(c)  Discuss the possibility of applying Variance Reduction Techniques (specifically, the Antithetic Variance Method) to your simulation. How would you go about doing it ?  How successful is it likely to be, and why ?


Hand-in

You should submit your source code using the  electronic hand-in  procedure provided. You should also submit a hard copy of your Project Report no later than the Final Exam period on December 17th.