Sample Response for Panel 1

Important questions:

1. How can band structure calculations be used to deduce
	A. simpler atomistic models to use for multi-scale modeling 
		(both length scale and time scale)
	B. transport and optical properties for complex materials
	C. structure and interaction of defects 
 	D. microstructure of polycrystalline and alloy material, 
	E. effective models and its parameters, including dissipation, 
		for describing material on micron level and higher
	F. time evolution of defects, microstructure, etc. on
		microsecond to millisecond time scale
	G. models for understanding nonlinear and time-dependent 
		mechanical processes/behavior of real materials

2. How can band structures calculation be used to deduce effective 
   many-body models and the parameters of those models:
	A. magnetic materials, for example, extending LDA+U to
		realistic model
	B. superconductors, especially HTSC
	C. semiconductor heterostructures, transport and optical
		properties
	D. polymer materials, structure and properties
        E. proteins, especially folding and structure
        F. other phase transitions such as metal-insulator


Benefits/needed-resources/barriers/time-line/needed-expertise:

1.A. Models for multi-scale modeling. 

Be:	A suite of simpler, well-tested models -- for a wide range of
	multi-atom systems  -- that quickly reproduce the structural
	and dynamic results of laborious "first-principles" band
	structure calculations is essential to multi-scale modeling.
	They would be widely adopted by simulators in physical and
	engineering sciences.

N-R:	Extensive testing on a range of similar structures corrected
	computed by first-principle methods are required to determine
	the parameters; a protocol for this testing procedure will
	require interdisciplinary discussion and agreement.

Ba:	The principal barriers:  needed advances in model building and
	few professional rewards for such efforts.  Development of new
	ideas for hyper molecular dynamics will make microsecond
	and millisecond time scales possible.

T-L:	Target date: 2010; even for silicon, there are no universally
	accepted tight-binding, effective medium or pair-like potentials.   
        2003: development of criteria for NRL tight-binding potential.  
        2005: development of pair-potential for two-component systems; 
        2007: testing of effective medium approach potentials.  
        2010: alloy tight-binding potentials.

N-E:	Ongoing developments during that period will stimulate further
	experiments: for example, high-pressure phases and more
	detailed phase diagrams.


2.A Magnetic many-body models.  

Be:	Adding orbital degrees of freedom to the Hubbard model, for
	example, is necessary to explain magnetic behavior in real
	systems.  The large group of many-body theorists could be
	make important contributions.  Real predictive power could
	stimulate the industrial development of light-weight, 
	flexible magnets.

N-R.	Extension of such methods as FLEX to realistic wavefunctions
	in solids.  Probably this requires the development of wide
	accepted basis sets such as proved important in the 
	development of quantum chemistry.

Ba:	Synergistic strategies between electronic structure theorist 
	and many-body theory to develop a road map leading to 
	realistic models.
	
T-L:	Target date: 2010.  Preliminary efforts on the simplest system, 
	iron, suggest that single-element magnetic materials could
	be in done in five years.  Alloys and complex magnets will 
	need a decade.

N-E:	Spin-dependent and orbital-dependent measurements are needed
	to provide intermediate data to calibrate methods.  Spallation
	and synchrotron sources measurements are essential.

3. General computational advances needed.

Assume that scalable computing will steadily advance: specifically,  
compilers and preprocessors/optimizers that are effective across 
different platforms; shared memory architecture.

4. Cross-cutting computational science issues.

There are no effective tools for debugging and monitoring parallel code
development and use.  Visualization of the tools is also necessary.

Essential needs for utilizing scalable hardware.
	parallel sparse matrix-matrix routines 
	parallel sparse manipulation of multi-index objects 
	parallel fast-Fourier routines for irregularly gridded objects.