Revision as of 19:38, 11 July 2016

Predicting Relative 'Cleanability' from Geometry

Mentors: Dr. Stephen Merrill and Casey O'Brien

Personal Info

My name is Anna Sisk and I am going into my senior year at Marquette University. I am majoring in Mathematics and minoring in Biological Sciences.

Goals for the Summer

• Characterize the distribution from from the CT scans of the plastic models

• Determination of simulation time between plastic and mathematical models

• Use optimizing techniques to minimize particle build up

• Define 'cleanability'

• Apply model to other geometries

Weekly Accomplishments

Week 1

• Orientation to the REU
• Lunch with Dr. Merrill and Casey to determine project and develop a preliminary outline for the summer (Tuesday)
• Two additional meeting with Casey to discuss the different tools and programs we will use, receive background references and information, and work on using Matlab (Wednesday and Friday)
• Research and reading on background information: what the problem is, why it is important, how attention was brought to it, what is being done, ect.
• Practicing different tools on Matlab
• Setting up this wiki page

Week 2

• Meeting with Casey (Monday)
  • received PDF and CT scan data
• Continuing Matlab practice
• Meeting with Casey and Dr. Merrill (Tuesday)
• Graphed and plotted the PDFs for model D
  • Calculated average of the 5 runs and graphed it
• Read about One-Dimensional Search Methods in Introduction to Methods of Optimization
• Looked into sequential and dichotomous search methods
• Meeting with Casey and Dr. Merrill (Thursday)
  • Discussed what to work on and how to make a poster
• Attended Dr. Brylow's Talk on Good Research Talks and Papers (Thursday)
  • I am now terrified of grad school
• Rescale graphs, so they accurately reflect PDFs
• Calculated and graphed standard deviation and variance of PDFs
• Started testing for a Poisson distribution

Week 3

• Finished the test for a Poisson distribution
  • The data does not appear to fit this distribution
• Working on developing a conceptual model (traffic flow)
• Meeting with Dr. Merrill and Casey (Tuesday)
  • Taking the data from Casey's model and matching the mean and variance to the data from the plastic model
  • Find out how long Casey's model needs to run so it matches the plastic model data
  • Continue working on the conceptual model
• Wrote up a summary for the conceptual model
• Dr. George Corliss' talk on using LaTeX (Thursday)
• Meeting with Dr. Merrill and Casey (Thursday)
  • Continued to outline poster and paper
    • Talked about different titles
  • Discussed conceptual model and how it can inform our thoughts on this project
  • Need to come up with a definition of 'cleanability'
  • Decided on goals for the next week and and outline for the rest of the summer
• Drafted a preliminary outline for poster
• Next steps:
  • Receive data from Casey's model and determine the correct run time for his model
  • Compute the sum of squares and plot it vs time
  • Calculate the minimum for the function
  • Fit the variance and mean of the two models
  • If we are able to validate this specific model, we can then try to apply this method to other plastic models with different geometries

Week 4

• Continued working on poster outline
• Received the simulated data from Casey
  • Plotted the data
• Meeting with Casey (Tuesday)
  • Talked about mini presentation
  • Looked and the data from the simulation and the plastic models
  • Talked about interpolating the data so we can calculate the sum of squared differences (SSD)
• Leaned how to interpolate data on MATLAB
• Figured out how to calculate SSD (taking into consideration the plastic and simulated data have different slice sizes)
  • Plotted the SSD and fitted a spline curve to it
• Meeting with Casey(Thursday)
  • Went over results from SSD
  • Simulation parameters may need to be adjusted so the simulation data fits the plastic model data better
  • Continued to talked about mini-presentation
  • Explained how to create a function on MATLAB
• Worked on slides and outline for mini presentation
• Created a MATLAB function for computing SSD

Week 5

• Worked on powerpoint and talk for mini presentation
• Meeting with Dr. Merrill and Casey (Tuesday)
  • Talked about how to give a short research talk
  • Gave update on what happened last week
  • Discussed ways to get the SSD work
• Finished writing mini-presentation talk and slides
• Received new simulation data
• Mini-Presentation
• Meeting with Dr. Merrill and Casey(Thursday)
  • Talked about presentation tips
  • Brief update on what I've accomplished this week
• Figured out how to work with new data
• Interpolated simulation data
• Calculated SSD between new simulation data and plastic data average
• Discussed results of SSD with Casey
  • Still not looking correct

Week 6

• Happy Fourth of July
• Meeting with Dr. Merrill and Casey (Tuesday)
  • Looked at SSD and data calculated last week
  • Talked about using scaling to fit the data
  • Discussed ways to improve delivering talks
• Worked on scaling plastic data
  • scaled in by 10, 15, 20
• Showed results to Casey
  • Data looks much better
• Wrote a function to determine where the SSD stop the decreasing trend
• Looked at the data scaled by 16, 17, 18, 19

Meeting with Casey and Dr. Merrill (Friday)

• • Reviewed scaled data
  • Decided to pick scaling factor by matching the area under the curve for the plastic to the area under the curve for the simulation
  • Looked at picking a run time (maybe 15 seconds?)
  • Talked about writing the research paper
• Looked at area under the curve for the plastic and simulation curves
  • Area under the curve for simulation: 20.07
  • Area under the curve for plastic: 1 (since its a PDF)
  • Thus a scaling factor of 20 best fits the data