01 Sept 22 - Notes: What is a Model?
Contents
01 Sept 22 - Notes: What is a Model?#
These is the summary of the PHY 415 class period on 01 Sept 22. If I missed anything, send me an message on Slack and I will add it here. -DC
Modeling Discussion#
We discussed models and modeling extensively today to start to form a functional definition of both. We talked about what the general idea of models and modeling can encompass. Some of the classes major ideas were:
A model takes BIG things and makes them smaller things (approximations and assumptions)
A model is only as good as it fits nature/reality/experiment
A model is not useful unless it does something we intended it
The best models destroy themselves (i.e., are pushed to their limits and need to be replaced/redeveloped)
Mathematical Models in Physics#
We narrows our discussion to mathematical models, the focus of this class, and talked about the variety of models the class has used. We had a couple general ideas. A mathematical model (in physics) is:
an equation that has a physics idea behind it
the order that you use different mathematical methods to investigate a physics problem
In this discussion, the class brought up several models they had experienced in the past; many of them were from mechanics contexts. And a couple were outside of physics models.
Physics
Hooke’s Law
Differential equations that describe physical systems
example: the Schrodinger equation
Taylor expansions
Newton’s Laws
Kepler’s Laws
Newton’s Law of Gravitation
Outside of Physics
Zipf’s Law (linguistics)
Benford’s Law (random numbers)
Features of a good mathematical model#
We then discussed the features that make up a good model in physics. Ideas included aspects of the model and the modeling process as well as social aspects of engaging in science. The list below is from the class’s discussion, where the highlighted words are starting to form the details of our class’s project rubric.
A good model:
depends on its intent and how well it does what it we intend it to do
e.g., a model of a galaxy is not useful for modeling the solar system
predicts or explains something about a physical system - predictability
gives everyone the same results with the same input - reproducibility
balances simplicity and applicability - parsimonius
cannot be biased towards the data (experiment, etc.)
avoid manipulation of the model to fit data
overfitting data
recognizes limitations and presents its assumptions
uses input from outside sources (other theories, data, experiments, etc) to confirm/evaluate - validation
can be adjusted for different conditions - investigatory
is testable in some way
needs to have its results and processes communicated clearly