AutoLigand AutoDock, Rodney Harris
- Tool for "rational" ligand design
- computes affinity
GpuPy, Benjamin Eitzen, WashU
- strengths
- currently improving faster than traditional cpus
- weaknesses
- single precision
- data must be copied to gpu
- programmability
- newer gpus can execute programs in "high level" languages Cg GLSL HLSL
- programs are called shaders
- how it works
- GPU executes program once for each pixel that is drawn to screen
- overrrides default numpy functions, can be transparent
- some functions are slower on gpu (like simple add or multiply)
- some get faster with larger N (pow, arccosh)
- some are just faster (cos, sin)
- the more complex you can pass over the faster, correlation
- not hooked up yet, but have edge detection filter
- possible future work,
- autoconfig
- simulation of double with singles hal.ccssd.cnrs.fr/ccsd-00021443
- htpp://eecs.wsu.edu/~eitzenb/gpupy
Boost Graph Library, Douglas Gregor
path finding
connectivity & clustering
network flow
coloring
layout & visualization
motivation
- python is easier to teach
- dynamic features of python
- usabiliy (e.g. c++ error messages)
Task: network tolerance
from boost.graph import * g = read_graphviz("network.dot")
bicomponent = g.add_edge_property(name='label', type='integer) art_points = biconnected_components(g, bicomponent)
- for v in g.vertices:
if v in art_points:
graph types in python
- Graph
- Digraph
- Graph exploration interface
vertex and edge properties
- attach additional model domain knowledge
- computer net, vertex(host, ip_addr), edge(bandwidth, latency)
Properties in the BGL
- property map
- implicity, just access property
- explicity, allows more specifity, 2 params optional weight = g.add_edge_property(name='weight', type='float')
- using property maps
- similar to dictionaries (weight[e], =0.75
- pass to algorithms dijkstra_shortest_path(g, start_vertex, weight)
- has visitor model to do something when you hit interesting points in the algorithm. (e.g. debugging, new algorithms)
summary of research issues with/Fast multipole Algorithm, Idesbald van den Bosch
- ship signature reduction
- lots of details about problem and need for optimization
- gmsh (mesh generation)
Object Oriented Finite Elements, Andrew Reid
- OOF2,
- rapid construction of meshes adapted to irregular microstructural geometriesp
Parallel PDE solvers in python, Bill Spotz
- trilinos project
- provide central repository for sandia's solver
- http://softwrae.sandia.gov/trilinos
Python imaging tools for reconstructing magnetic resonance images, make trumpis
PyRoot, Wim Lavrijsen (pronounced more like vim)
- http://cern.ch/wlav/pyroot/
- ROOT root.cern.ch
- reflection for C++ (reflex, http://cern.ch/seal-reflex)
- they combine reflection based meta-information to generate python wrappers on the fly.
- also can call python from C++/CINT
- very impressive looking
QScimpl, Eric Dobbs
- personal, dobbse.net
- Tech-X Corporation, txcorp (scientific physics stuff)
- mostly decided to advertise the advantage of simplicity
Genesis SciFlo, Scientific Knowledge Creation on the Grid, Brian Wilson
- Problem, terabytes of data (years) distributed at several data centers
- iEarth
- showed an updating svg diagram showing the execution of their workflow.
- distributed across many machines via web services
- data access protocol, retrieve chunks of data over web.
- used xpath to extract results from amazon soap results
- software shipped
- lxml (XPath 1.0)
- twisted, pyldap, openldap,
- sleepycat dblxml has xquery and xpath
- dojo AJAX library
- openDAP, extract cdf, hdf file
- globas v4 grid
- globus v4 is supposed to be soap webservice
- WSRF (web services resource framework)
- UDDI universal description discovery and integration, find service
- grid workflow,
- rest vs soap, each have advantages/disadvantages
- UDDI4py from IBM, SOAPpy
- recommends more XML microformats for "mash ups"
- publish science algorithms as reusable web services
- www.opendap.org
Python Web & Grid Service Tools, Keith R Jackson
- Globus toolkit
- decided to simplify system by standardized into webservices
- helps organize coordinated usage of resources
- his mac crashed
- SOAPpy is officially unsported
- ZSI is hopefully the "new" better way, though they're more complicated than SOAPpy
- XML schema scary complicated
- Mapping tool from WSDL to python (wsdl2py)
- using m2crypto (they thought it was being supported, unlike pySSL)
- LIGO used it to ship 50 terabytes of data through pyglobus
- WSRF supposed to define way to hand stateful connections
- lifecycle, things will die at some point, though can be refreshed.
- WS-N another webservices spec
- seems to actually work, java, C, python call all talk.
- PyCLST
- GridFTP, implement extra features to handle transfer over highperformance network
- Visual programming environment, useful for viewing workflow
- Userinterface
- planner, condor, dagman
- exec manager (CBEI)
- exec context
- physical resources
- pyGlobus, http://dsd.lbl.gov/gtg/projects/pyGlobus/
- ZSI, http://pywebsvcs.sourceforge.net/zsi.html
- pyGridWare, http://dsd.lbl.gov/gtg/projects/pyGridWare
- ViCE, http://dsd.lbl.gov/gtg/projects/vice/
- PythonCLServiceTool, http://dsd.lbl.gov/gtg/projects/PythonCLServiceTrool
- everything BSD license
- eclipse has a nice WSDL interface generator
- python to wsdl would require static type annotation.
- webservices, text to do control, then binary for shipping data
- amazon, uses a very efficient xml binary encoding
Seeing through the MIST, Tripp Lilley
- interesting point about maliable UIs
- how do you let a user, know theyre changing their world, and revert.
- super early project, not even screenshots