Beruflich Dokumente
Kultur Dokumente
1/23/2004
Prof. T. Dobbins
Outline
Scientific Research Method
Development of Mathematical Models
Applying these methods to completing an MS Thesis or a
Ph.D. Thesis
False
True
4. Predictions 3. Testing
If the experiments prove the hypothesis to be true, it becomes a theory or law of nature. If the experiments prove the
hypothesis to be false, the hypothesis must be rejected or modified. The scientific method used properly should give us
predictive power (to understand phenomena which have not been tested).
Suggestions (no one can tell you HOW to do this, you have to develop this skill):
1. Couple archival journal reading with your in-class theory.
2. Couple archival journal reading with your common sense or intuition about the
way things should be in the system that you are studying.
3. Follow a logical reasoning about what you are reading with scientific and/or
mathematical basis (use drawings and diagrams to help your understanding).
4. Write up your thoughts and opinions either in report format (or in a way which
you can follow within one notebook).
Experimental Testing
to be assigned a standing or evaluation on the basis of tests 2 : to apply a test
Suggestions:
You may begin thinking about predictions as soon as you have a hypothesis,
however, if your hypothesis is proven false, your prediction will also fail!
Use mathematical model to test your prediction.
Mathematical Models
Empirical vs. Deterministic
Empirical Deterministic
Based on experimental Based on first principles approach!
observation (e.g. least squares (e.g. Monte Carlo calculations,
curve fitting, coventor MEMS Density Functional Theory,
Molecular Dynamics, Computational
calculations, XAFS FEFF8.0 Fluid Dynamics)
Analysis)
Examples,
Test: EXAFS Absorption spectroscopy data is in hand. Prediction 1: If Ti2+, Ti3+ and Ti4+ gain equivalent oxidation
states after entering the structure, then the same should be true for Co2+, Co3+, and Co4+ (or any other metal
dopant).
Model: We use FEFF8.0, WINXAFS to fit the XAFS equation to measured data --- varying types of neighboring species
(e.g. neighbors may be 2 Na atoms and 1 Al atom or 3 Al atoms and 1 Na atom). bond length and # of bonds
(coordination number) until we have a good fit. We did not know what environment our Ti atom was in prior to
model fitting, now, we have modeled our system to match our data. Our model is then made up of the XAFS
equation, bond lengths, and # of bonds.
Keys to Implementation in your Research:
All thesis experiments should have empirical models (at least).
If do not have a mathematical model--- only data, then we need to find an equation
for which we can perform least squares fitting your data.
My suggestion: If you run into an interesting hypothesis which can be tested via
deterministic models (first principles approach), meet with IfM modeling faculty
(Mainerdi, Paun, Hegab or DeRosa) to discuss the hypothesis and modeling.
Further Suggestion:
Know what has been done by others. Do not reinvent the wheel
unnecessarily!!!
Do not strive to solve all problems in the field, look for small advances in
your field.