Computational Final Project Option

Project Description

During this course you have explored multiple computational models (written in NetLogo) of materials systems. These models help explain some aspect of a materials science phenomenon, often showing how a structure or property of a material emerges from the behaviors and interactions of atoms and/or electrons. The computational modeling final project option consists of three parts:

  1. Modifying one of the NetLogo models we have used in class to model a new phenomenon (or creating one from scratch if you really want to).
  2. Running a computational experiment with your model.
  3. Writing a short paper (1250±150 words) describing your model and your findings.

Each of these parts are described in more detail below.

Writing a short paper

Even though writing the paper will probably be the last thing you do, describing it first is useful because it frames the whole project. The paper should have the structure of a standard scientific paper with the following sections:

  1. Background on what you are modeling, what you already know about it, and why it is important/interesting to understand.
  2. Research question/purpose: state a research question you will try to answer with your model and/or the purpose of your model. There are two broad options here:
    • Phenomenon-based model in which you pick a real phenomenon and try to model it
    • Exploratory model in which you change some aspect of an existing model in a way that is theoretically interesting, even if you don’t know of an existing phenomenon that has the properties of the model
  3. Methods
    • Description of the model: describe what the entities of the model are (e.g., atoms, molecules, and/or electrons), how they are modeled (e.g., as Newtonian particles), how they are initialized (e.g., initial positions), and what rules they follow (e.g., molecular dynamics, Monte Carlo, etc.). You should justify the modeling choices you make.
    • Description of the experimental procedure: what parameter(s) did you vary and what outcome did you measure?
  4. Results: Description of the outcome of your experiment. In most cases this should include at least one graph. Occasionally it will only require screenshots of structures/behaviors that emerged.
  5. Discussion of both the usefulness of the model and its limitations for answering your research question or achieving your modeling purpose. End the section with future work that could be done exploring the model and/or extending it.

Modifying a computational model

Most of the models we have used in class are intentionally made as simple as possible. This is to make it easier to understand the model and the phenomenon it is modeling. For example, the models of interatomic potentials, and molecular dynamics models from which crystal structure emerged use only a single type of atom. To model many materials science phenomena requires more complex models. In many cases, it is easiest to start with an existing model and modify it in some way rather than starting from scratch. You are free to make a model from scratch, but we recommend starting with an existing model and modifying it.

Whatever you choose to model should be mappable to at least one part of the materials science and engineering paradigm, but it doesn’t need to address all aspects. For example:

  • Processing → Structure Relationship: Many of the models we have seen can be framed as models of how/why structure emerges from a certain type of processing. Often the processing is implicit in the model. For example, if you just put a bunch of atoms together in a computational model and let them crystallize, it might not seem like you modeled a processing procedure, but really this is a type of processing that could be mapped to a real process (e.g., cooling a gas of atoms until they solidify). The diffusion models we used in class can also be framed as modeling a processing procedure that allows diffusion to occur which results in a certain concentration profile of atomic species (a structure).
  • Structure → Properties Relationship: Some of the models we have seen (or will see) can be framed as models of how properties emerge from structure. For example, we will see models of how electrical properties of materials emerge from behavior of electrons and structure of the materials they are in.
  • Properties → Performance relationship: It is unlikely that your model will address this aspect of the MSE paradigm, but if you have an idea for one, feel free to propose it.

You will turn in your NetLogo model/code as part of the project. In a comment at the top of the code tab, you should indicate which lines of code you modified/wrote.

Running a computational experiment

Usually when we make a model, we want to understand how the modeled system behaves under different conditions. This requires running computational experiments in which you vary one or more parameters of the model and measure some outcome (NetLogo has built-in tools that make running this kind of experiment very easy). Here are some example of simple experiments that could be run with models you have seen in class:

  • Modify the parameters of the interatomic potential used in a molecular dynamics or Monte Carlo model and see how some structure or property changes (e.g., melting temperature, thermal expansion coefficient, etc.)
  • Modify the boundary or initial conditions of a diffusion model to see how this affects the concentration profile after a fixed amount of time.
  • Add varying concentrations of interstitial atoms to a crystal in a molecular dynamics model and see how that changes the strength of the material.

For your project, you will need to run an experiment with your model in which you systematically vary at least one parameter of the model and measure some outcome.

Assessment/Rubric

You will be assessed on both the model itself and the paper you write.

The paper (30 pts total):

  • Background/motivation (2.5 pts)
    • Full marks: Clear background to the model and motivation for exploring this topic
    • Some marks: There is some background, but not clear why this topic is important/interesting to explore
    • Few marks: Little background or motivation for why this topic is important/interesting
  • Research question/purpose (2.5 pts)
    • Full marks: Clearly stated purpose or research question
    • Some marks: Purpose or research question is stated, but with some ambiguity
    • Few marks: Purpose or research question is very vague or not stated
  • Methods
    • Description of the model (5 pts)
      • Full marks: Model description is clear and complete. Should be complete enough that an informed reader could implement your model. This should include pseudo-code in bullets outlining the main steps of the model. Modeling choices are justified.
      • Some marks: Some parts of the model description are missing or hard to understand or modeling choices are not well justified.
      • Few marks: Model is not clearly described and modeling choices are not well justified.
    • Description of the experimental procedure (5 pts)
      • Full marks: experimental procedure is clear and complete
      • Some marks: experimental procedure is described in general, but not clear enough for the reader to reproduce
      • Few marks: experimental procedure is very vague
  • Results (7.5 pts)
    • Full marks: Clear description of the results of the experiment with at least one high quality figure to communicate the results
    • Some marks: Written results and/or the figure(s) are somewhat difficult to understand/interpret
    • Few marks: Results are stated in vague terms. Either no figure, or very difficult to understand figure
  • Discussion (7.5 pts)
    • Full marks: Incisive discussion of (1) the usefulness of the model (what we learn from it) and (2) its limitations, along with (3) potential directions for future work.
    • Some marks: Discussion leaves out one of the parts or is not very insightful for one or two parts.
    • Few marks: Discussion doesn’t not clearly address the three parts of the discussion.

The model (20 pts):

  • Code works as intended, is well written, and well commented (10 pts)
    • Full marks: The code does what it is intended to do, is well written, and has comments to explain what it does (for non self-explanatory parts)
    • Some marks: The code does what it is intended to, but is poorly written or does not have comments explaining what it does.
    • Few marks: The code does not do what it is intended to do
  • Visualization is informative and helps understand the system (5 pts)
    • Full marks: Good visualization of the modeled entities (e.g., atoms, electrons etc.) helps the model user understand the phenomenon. Good graphs/monitors to help the user understand emergent/aggregate behavior.
    • Some marks: Some aspects of the visualization are confusing or hard to interpret
    • Few marks: Visualization does not help the user understand the model
  • Experimental procedure works as intended (5 pts)
    • Full marks: The outcome measure(s) for the experiment measures what it is intended to, and the experimental procedure tests the relationship(s) between the experimental parameter(s) and outcome measure(s) as intended
    • Some marks: The outcome measure(s) does not fully measure what was intended, but the experimental procedure still tests the relationship between the experimental parameter(s) and outcome measure(s)
    • Few marks: The outcome measure(s) does not measure what was intended or the experimental procedure does not test the relationship between the experimental parameter(s) and outcome measure(s)