1
0
Fork
You've already forked importance_sampling_study
0
Elementary importance sampling code for a certain network routing problem
  • Python 100%
Find a file
2022年08月22日 09:53:39 +02:00
doc licensing amend; doc 2022年08月22日 09:53:39 +02:00
other_tests Initial commit 2021年09月30日 14:49:23 +02:00
test_out Initial commit 2021年09月30日 14:49:23 +02:00
GraphExt.py Initial commit 2021年09月30日 14:49:23 +02:00
InsertionSorter.py Initial commit 2021年09月30日 14:49:23 +02:00
README.md licensing amend; doc 2022年08月22日 09:53:39 +02:00
ShortestPathIS.py Initial commit 2021年09月30日 14:49:23 +02:00
ShortestPathIS_run1_CMC_linear.py Initial commit 2021年09月30日 14:49:23 +02:00
ShortestPathIS_run2_CMC_vs_ISMC_linear.py Initial commit 2021年09月30日 14:49:23 +02:00
ShortestPathIS_run2b_CMC_vs_ISMC_linear_as_table.py Initial commit 2021年09月30日 14:49:23 +02:00
ShortestPathIS_run3_ISMC_Rubinstein_example.py Initial commit 2021年09月30日 14:49:23 +02:00
ShortestPathIS_test1.py Initial commit 2021年09月30日 14:49:23 +02:00
ShortestPathIS_test2_CMC.py Initial commit 2021年09月30日 14:49:23 +02:00

Importance sampling example

This repository demonstrates the strength of importance sampling for estimating multi-dimensional integrals, here in particular a probability of a rare event (= \int 1_{\omega \in rare event} \d P).

This is in comparison with "naive" random sampling, and in comparison with analytical bounds on the same probability.

The example problem used here is of estimating the probability of obtaining a shortest path length (in a given fixed graph) in excess of a certain prespecified threshold \gamma. (applications in network planning)

This problem setting is from [1], introductory section (page 31). The graph used is

Graph for example

shortest path from S to T. Edges carry weights specified collectively in a vector vec_u.

License

Specific for files, see single files. Documentation files (*.txt) are CC BY-SA 4.0 licensed.

References

[1] Reuven Y. Rubinstein, Dirk P. Kroese: "The Cross-Entropy method". Springer, 2004.

[2] https://github.com/ahojukka5/Dijkstra/blob/master/dijkstra/graph.py