IMUC 2009

Here are my talks at the International Mathematica User Conference 2009.

Inverting the SIR Model to Study Origins of Novel H1N1

The wide spread of Novel H1N1 provides an invaluable lesson for data analysis and modeling of a pandemic. A study based on analyzing case and mortality numbers using the SIR, or Kermack-McKendrick, model to provide an estimate of how long an outbreak has lasted in a population is discussed. The presentation used map images and geographic data from Wolfram|Alpha.
  • The SIR Model (aka Kermack-McKendrick Model) is a set of 3 coupled ordinary differential equations that track the three major groups of an epidemic (Susceptible, Infected, and Recovered)
  • Question: Can the SIR model be used to extrapolate backwards and estimate the origin of a pandemic?
  • Hypothesis: Given data about a pandemic's spread in individual countries, the SIR Model can be used to solve for the amount of time the pandemic has been present in the country, yielding an estimation as to when and therefore where the pandemic started.
  • Data from 132 countries as published by the World Health Organization on July 6th, 2009 was used to to test this hypothesis
  • Mathematica was used to solve for how long the epidemic has been present in each country by trying to match up the number of case numbers or number of deaths to the SIR Model
What is the value of the research?
  • It is plausible to use the SIR model to extrapolate backwards and forwards, making it an option to investigate pandemics when data is still limited; only one-day snapshot of case & mortality number are needed. 
  • Can be used to track new pandemics when only numbers are available before pathological studies are possible.
  • Can also be used to analyze historic pandemics when data sets do not contain details about time.



Using 3D Projection to Visualize Multi-Dimensional Objects

Single-Variable calculus is commonly taught from a graphic standpoint, using the concepts of area and tangent lines to describe the world of integrals and derivatives.  However, when we move to multivariable calculus, most students ask the question: what does a 4D surface look like?  Without this idea, it is difficult to draw the connection between techniques learned in elementary calculus and techniques in the 4th dimension and beyond.  We certainly cannot pull out a hypersphere from our textbooks.  Two methods of visualization using Mathematica are proposed.

It is plausible to describe a 3D object using 2D images by plotting the "shadows" that are cast on 2D planes (subspaces).  We can use the same idea to visualize a higher-dimension object by plotting its projections on a systematic set of 3D subspaces.  In order to do this, a projection matrix can be making a spanning matrix of the target 3D subspace, then multiplying it by its pseudoinverse.  Hence, we can do a linear transformation of any multi-dimensional object into a plottable 3D surface.
  • Technique 1: Projecting on a Systematic Set of Coordinate Axes
    This method generates a list of 3-D subspaces for the higher dimensions by coming up with all the different combinations of coordinate axes of the higher dimension, by taking 3 at a time.  This is more intuitive for us.
    Visualizing a 5D hypersphere is presented.
  • Technique 2: Projecting onto the Null Space of a Selected Point
    The second one allows the user to choose a point on a 4D surface.  The surface is then projected onto the null space (aka perpendicular space) of that point.  The null space of a point is defined as the set of all vectors that will linearly transform the point into the null vector.
    Visualizing a 4D hypersphere is presented.
Afterthought:
  • When we take the 3D projection of a hypersphere, we do not always get spheres.  On the other hand, if we have a 3D sphere and project it on 2D planes, we always get a circle.  However, when we project on the subspace consisting of the xyz-axes, we do end up with the original unit sphere.  Also, projecting a hypersphere on the null space of a point on a sphere produces a sphere-like object that collapses on itself becasue of the nature of the null space.
  • Although this is only a crude approximation of 4D objects, these techniques can help wrap our mind around such an abstract idea.  This shows that the various dimensions can be connected to each other somehow, and are not as foreign as assumed.

To view the Mathematica notebook files you need Mathematica Player (free) or a copy of Mathematica (discounted price for students, click here).

Sign in  |  Terms  |  Report Abuse  |  Print page  |  Powered by Google Sites