experipy: Automating Script Generation

experipy is a framework for writing and running Computational Science experiments. It provides facilities for describing an experiment as a shell script, and mechanisms for then running it. Experiments can be run locally and also submitted to a cluster’s job queuing system as a PBS script.

from experipy.exp       import Experiment
from experipy.grammar   import Executable

echo = Executable("echo",
    ["Hello World", "> test.out"],
    outputs=["test.out"]
)

exp = Experiment(echo, expname="test", destdir="results")
exp.run()

The intention of experipy is to act as the core of a researcher’s scripting framework. In the author’s research group, projects often involved running dozens of benchmarks with hundreds of configurations in parallel across a cluster, so experipy was designed to ease the design and scripting of new experiments and configurations.

Installation

experipy is available on PyPI:

pip install experipy

Or, you can find it on Github at https://github.com/Elemnir/experipy.

Contents:

• 1. experipy.grammar - Composing experiments
  • 1.1. Element objects
  • 1.2. Executable objects
  • 1.3. Wrapper objects
  • 1.4. Pipeline objects
  • 1.5. Group objects
  • 1.6. Block objects
• 2. experipy.exp - The Experiment Runner
  • 2.1. An Example
  • 2.2. Experiment objects
  • 2.3. The Exp Namespace
• 3. experipy.system - System tools in the grammar
• 4. experipy.config - Configuration utilities
  • 4.1. Namespace objects
• 5. experipy.metrics - Results Parsing
  • 5.1. Metric objects

Indices and tables

• Index
• Module Index
• Search Page