deletions | additions       diff --git a/introduction.tex b/introduction.tex  index e273a63..5912df7 100644  --- a/introduction.tex  +++ b/introduction.tex 
...
However, there is another large class of materials called strongly-correlated compounds, whose distinguishing phenomenological feature is sensitivity to external perturbation, which makes them technologically useful. Small changes in pressure, temperature or chemical doping often drives large changes in electronic or structural behavior. For example, changing the temperature by only several degrees Kelvin can drive a transition between a metallic and insulating state, behavior not observed in weakly-correlated compounds. In addition to metal-insulator transitions, these compounds display unusual magnetic properties, high-temperature superconductivity and strange-metal behavior.  The basic feature of correlated materials is their electrons cannot be described as non-interacting particles, posing special challenges for theory. particles.  Often, this occurs because the material contains atoms with partially-filled $d$ or $f$ orbitals. The electrons occupying these orbitals retain a strong atomic-like character to their behavior, which when combined with the band-like behavior of while  the remaining electrons, makes electrons form bands; their interplay poses special challenges  for a challenging problem. theory.  Consequently current implementations of DFT cannot describe their properties accurately. This led to the development of extensions to DFT such as LDA+U, and entirely more sophisticated approaches such as dynamical mean field theory (DMFT) and the GW approximation. For understanding the electronic properties of a material given a crystal structure, DMFT and GW perform remarkably well. However, only LDA+U can currently scale to produce total energies for simulations involving thousands of compounds. Thus we adopt a hybrid workflow correlated materials, one where structural prediction is performed using LDA+U and then, once the final structure has been obtain, detailed analysis is performed using DMFT or GW.  TODO: Define the problem: What IS design of correlated materials?  Describe the intersection of materials design with correlated materials. Also describe the need for large computable databases. % Only LDA+U can produce energies at scale.  % Materials design also necessarily involves handling and organizing large bodies of data since the process of check => machine learning.  To be precise, we can phrase the question of materials design concretely as follows: given a chemical system, determine the crystal structures and electronic properties of all stable compounds formed by the constituent elements. For example, if the chemical system of interest is Li-Fe-P-O, determine all binaries, ternaries and quaternaries and compute their properties (turns out LiFePO$_4$ is a promising battery material). This problem involves coordinating many moving pieces, including structural prediction, determination of thermodynamic stability against competing phases and computation of electronic properties. In this article, we seek to summarize outstanding challenges in the area, especially as it pertains to correlated materials, and propose strategies to solve them.