modified on 8 February 2012 at 15:55 ••• 36,326 views

BES proposal concept

From NWChem

Jump to: navigation, search

Contents

Fundamental Scientific Issues

Target: Charge formation & transport and chemical reactions at complex interfaces

We need to link them to the science areas!!!

Energy storage materials

How does it link to the fundamental scientific issue (we need to have some paragraphs here)

Catalysis for renewable energy

How does it link to the fundamental scientific issue (we need to have some paragraphs here)

Photovoltaics

How does it link to the fundamental scientific issue (we need to have some paragraphs here)



We need to link them to experiments!!!

Free electron laser

How does this probe the fundamental science issue How do the methods capture the essential physics and chemistry of experiments on relevant systems (need some paragraphs here)

x-ray / neutron scattering

How does this probe the fundamental science issue How do the methods capture the essential physics and chemistry of experiments on relevant systems (need some paragraphs here)

electron microscopy

How does this probe the fundamental science issue How do the methods capture the essential physics and chemistry of experiments on relevant systems (need some paragraphs here)

Spectroscopic imaging in electron microscopy

Mapping excitations like plasmons, excitons in nanomaterials. Can electron microscopy be used to complement spectroscopic techniques ?

What theories and/or methods we want to develop (short paragraphs here)

- Predictive and scalable methods - Coupling theory and experiment

Karol

Niri/Ken/others

Initial set of ideas

Scalable implementations of real-time dynamics (FE/FD approaches)

Combining and embedding real-time electronic dynamics with classical Maxwell approaches, which is necessary to study phenomena like dielectric breakdown, materials damage in strong local fields (or electron micropscopy)

Non-adiabatic effects

Dissipation and decoherence effects

TDDFT for strongly correlated systems out of equilibrium

 - Development of new exchange-correlation functionals (motivated using high-level theory CC, QMC)
 - Applicable to local transformations in strong fields, strongly correlated systems
 - Study the electronic dynamics of these systems

Can the fragment approach (that Marat has been working on) be used to decompose a system into blocks to study excited state dynamics ?

 - Can we get the lowest excitations with this scheme ?
 - Can we break up the system into "blocks" to capture various regions in different excited states simultaneously ?
 - Can we propagate these blocks separately to study a pathway ?

Rational design of emergent properties in nanomaterials

 - Can we use these principles to design materials via a bottom-up approach ?
 - Can we use properties like electronic coupling and charge transfer to motivate the design of nanomaterials ?

Eric

Marat

Huub

The main challenges I see are that we need relatively simple approaches that can represent the physics with sufficient accuracy. In addition we need these approaches to be scalable so that we can treat systems that a sufficiently realistic. Obviously there are some unanswered questions here.

Starting from the call we need a framework of some kind that can encapsulate the kind of processes that we need to be able represent. The framework should provide patterns or models for what we can call "measurements". A measurement is a standard procedure for getting some result that is clearly defined and well understood. For example, running an NMR spectrum in solution is a measurement rather than an experiment. For the target science areas I think we need to consider what theoretical measurements are necessary or useful, and find a way to provide them to a user in a relatively straightforward way. Also good would be how to know when a standard approach cannot be used.

Another thing is to tackle inverse problems. Marat talked about this when he mentioned that experimentalists can measure things but they don't know what they are measuring. In our methods we usually approach this problem by hypothysing a particular structure or process and running some calculations to see if the results match the experiment. Ideally one would want to go the other way. For excited states this has already been done with something called "optimal control" where they calculate a laser pulse that will drive a particular chemical transformation. For other types of problems there will be similar inversions to consider.

We need more accurate energy expressions. The problem here is that no amount of compute power can fix the inadequacy of the energy expression used. Hence we need to start from energy expressions that are at least qualitatively correct for the physics/chemistry we want to describe. There are essentially two levels that one needs. One is an 1-electron theory for relatively quick but not so accurate work, the other is a many-body level theory for when the details of the electron correlation are important. I am sure that Karol has the second one covered. The first one however, runs into some well known deficiencies of the traditional Kohn-Sham DFT approach. To fix these some kind of density matrix approach is needed.

Edo

Bert

To enable major scientific discoveries in shorter time frames, the integration of multidisciplinary research is essential. Coupling data obtained from various complex experiments and simulations has the potential to deliver new unexpected insights, and drive the next set of simulations and experiments. Key to the effective integration, mining, and visualization of diverse data sources is to work within standardized and semantically rich data formats. A good example of standardization is the Crystallographic Information Framework, which includes the widely used Crystallographic Information File (CIF) format and the Extensible Markup Language (XML), maintained by the International Union of Crystallography (IUCr). Within computational chemistry various software projects have utilized XML to generate some level of output data, though no standard format has been developed. We recently adopted the Chemical Markup Language (CML), developed by Murray-Rust and Rzepa. CML is the de facto XML for chemistry standard and has been widely adopted. However, specific language dictionaries and conventions for computational chemistry (as well as forvarious experimental capabilities) still need to be defined. In partnership with open-source software development teams our goal is to define the computational chemistry language standard and champion its adoption throughout the community.

Partners

  • Truhlar for chemical reactions at surfaces
  • Who for transport?
  • Schenter for?
  • Mundy for?


Concept papers are limited to 1-page (singly spaced using 12-point times new roman font and margins no less than 0.75”), and therefore, require concise and focused writing. From practical and successful experience of what communicates well, the following approach is recommended in development of the 1-page concept paper.

Title

Principal Investigator (one)

List Key Researchers and Affiliations (inside and outside PNNL)

Technical Area of Interest (brief identification of technical area; see announcement for representative technical areas)

Statement of Fundamental Scientific Issues That Can Be Addressed with New Theory and Computational Tools and Approaches (this section addresses why the tools are needed)

The proposed center will develop x, y, and z. Development will be done using the NWChem open-source framework as a starting point. Currently hosting an open-source software repository that is accessible to the general user community.


Proposed Technical Approach (the section addresses what theory and modeling capabilities will be developed)

Why Research Team is the Client’s Team of Choice (Why the unique combination of talent on your team can accomplish this research)

Funding mechanism: Center

Estimated Total Budget: $2,000,000