This seems reasonable. Lets approve. Note that this is a strategic LDRD project, now prio 2. Ray On 10/1/15, 1:44 PM, "[email protected] on behalf of [email protected]" <[email protected] on behalf of [email protected]> wrote:
Hello,
A yearly allocation for the LCRC cluster has been requested with the following updated information:
Submitter/PI: Carolyn Phillips Project Name: Metadynamics_LC Division: MCS Project title: Using Parameter Free Metadynamics to calculate LC elastic constants Associated funding: Internal ANL LDRD 2015-172-N0 Other Systems: Science: We will be exploring and expanding the state-of-the-art metadynamics methods in the context of measuring the mechanical properties of materials, and specifically initially, the elastic properties of liquid crystals Project description: We will be using a parameter-free metadynamics method for an atomistic model of a liquid crystal 4-n-pentyl-4’-cyanobiphenyl (5CB), as well as the homologous series (6CB,7CB,8CB).
Each nCB molecule will be simulated via a coarse-grained united-atom approximation in a classical N-body Molecular Dynamics code. We will be using NAMD + PLUMED2. Simulations will be used to measure the elastic constants of each liquid crystal system using metadynamics.
NAMD is a well-parallelized stable MD code with a history of use by other users on Blues. Equilibrium simulations of the correct scale of the model described above parallelized across on 4x16-core nodes required ~300 MB of memory and ~36 hours to implement 10 million time steps. However most simulations will also be paired with PLUMED2, a parallelized version of PLUMED1.3 that can be used to implement metadynamics. Metadynamics is not known to perform well in parallel, due to a relatively high frequency of communication across all nodes, so I do not expect to use more than 1-4 nodes for individual "walkers" (simulation implementations). Instead production run simulations will likely use multiple (~20) "walkers" each deployed on 1-4 nodes, thus requiring ~80 nodes at a time. We will be evaluating trade-offs in length of simulation, number of nodes per implementation, and number of walkers to achieve convergence. Thus a better approximation of the allocation required to complete this project will result from the initial scoping stage of this project.
Previous MD + metadynamics simulations have been performed on a different model of the system using a different MD code on Midway at UChicago. Small runs using NAMD on Blues and Fusion have already been issued to gain familiarity.
Expected number of Project Members: 2 Industry partnership: Project URL: Current FY Hours Used: undetermined amount
New FY Requested allocation: 150000 Q1: 25000 Q2: 25000 Q3: 50000 Q4: 50000
Justification: Storage requirements:
Thank You,
The LCRC Accounts System