Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: David Bross Applicant's institution: ANL Applicant's division: CSE Project Name: spectra_mine Project title: Data mining physical chemistry literature for spectroscopic constants Associated funding: DoE BES Other Systems: My group is currently operating two computer linux clusters and have access to a third. The oldest cluster consists of 30 dual-CPU, 4-core 64-bit AMD Opteron nodes and 12 dual-CPU, 6-core 64-bit AMD Opteron nodes. These nodes have either 32GB or 64GB RAM per node and local disk capacity between 500GB and 1.2TB per node. The newer cluster consists of 11 dual-CPU 10-core Intel (Xeon E5-2680 v.2) nodes, 5 quartic-CPU 8-core AMD (Opteron 6278) nodes, 5 dual-CPU 8-core Intel (Xeon E5-2680 v. 1) nodes, 3 dual-CPU 6-core Intel (Xeon X5690) nodes, 8 quartic-CPU 12-core AMD (Opteron 6180 SE) nodes, and 16 single-CPU 12-core AMD (Opteron 6174) nodes. The memory for these nodes ranges between 66 and 396 GB RAM per node and the local disk capacity ranges from 1 to 1.6 TB per node. The communications for all of these nodes is 1 Gb/s ethernet. In addition to the two clusters we operate in house, we recently acquired 30 new nodes which have been installed on the Bl ues cluster in the Laboratory Computing Resource Center. These consist of 15 dual-CPU 16-core Intel (Xeon E5-2698 v.3) nodes and 15 dual-CPU 6-core Intel (Xeon E5-2620 v.3) nodes. Each node has 132 GB RAM and local 2 TB hard drive. Infiniband interconnect is used for communication. Science: There is an extensive body of physical chemistry literature containing extensive information that has intricate interdependencies on the physical constants, the definition of units, the physical model used to interpret it, and the assumptions that underlie the physical model. This makes the task of mining the literature extremely complicated, especially since the physical model and its underlying assumptions frequently generates a series of sequential dependences that traces back through other published works and usually this information is not readily available. Ideally the information should be stored as close to the actually measured or determined values as possible, enabling assumptions to depend on the rest of the available dataset at time of processing, rather than at time of input, and this will involve expanding the database efforts into virtually all physical chemistry disciplines (e.g. creating a database including kinetics and spectroscopi c data rather than the equilibrium constants and partition functions that depend on those spectroscopic constants). Natural Language Processing and machine learning classification techniques show promise as an approach to utilizing the available literature, and I will use these as an avenue of mining the extant literature. We initially begin this by attempting to mine the existing literature for spectroscopic constants. Project description: The project will use the python chemdataextractor package, primarily intending to use parallelism through the mpi4py implantation of Alvaro Vazquez from the Argonne Leadership Computing Facility. The task is primarily embarrassingly parallel, as each article must be mined before it can be used in a broader sense, and the project will probably involve modifications of the algorithms used to mine the articles and produce desired outputs. The LCRC allocation will be used to run initial test applications that can be developed later into a larger program that may have larger computing requirements. Industry partnership: Project URL: Requested allocation: 105000 Q1: 0 Q2: 35000 Q3: 35000 Q4: 35000 Justification: Storage requirements: The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System