... | ... | @@ -2,14 +2,16 @@ |
|
|
|
|
|
`salza` is a practical implementation of two universal algorithmic information measures on sequences based on LZ77[^1]<sup>,</sup>[^2] and relative LZ[^3]<sup>,</sup>[^4] coding.
|
|
|
|
|
|
Its rationale are described in this [preprint](https://cloud.uvolante.org/index.php/s/bJrf25qXQKEFxmZ/download). We thank the anonymous reviewers of ISIT'19 for spotting an inconsistency in the DPI ([ISIT'19 slides](https://cloud.uvolante.org/index.php/s/7oSeZKy9GKAMKKg/download)). This version is an update to the one submitted to IT, reflecting the current implementation (faster and slightly more accurate) and illustrating universal causal inference.
|
|
|
Its rationale are described in this [preprint](https://cloud.uvolante.org/index.php/s/AiGAcapjgiygcfH/download). We thank the anonymous reviewers of ISIT'19 for spotting an inconsistency in the DPI (for reference, here are the [ISIT'19 slides](https://cloud.uvolante.org/index.php/s/7oSeZKy9GKAMKKg/download)).
|
|
|
|
|
|
`salza` comes with built-in computation of a universal, normalized semi-distance (much in the spirit of Cilibrasi _et al._ work[^5]) and an implementation of causality inference using the (stable) PC algorithm[^6].
|
|
|
|
|
|
In case it is needed, [`drpt`](https://forge.uvolante.org/code/drpt/wikis) will convert a (`salza`) semi-distance matrix into a true distance matrix.
|
|
|
`salza` comes with built-in computation of two universal, normalized semi-distances (one in the spirit of dissimilarity[^3], the other in the spirit of similarity[^5]), and an implementation of causality inference using the (stable) PC algorithm[^6].
|
|
|
|
|
|
`salza` is multithreaded[^7], too.
|
|
|
|
|
|
# Companion software
|
|
|
|
|
|
In case it is needed, [`drpt`](https://forge.uvolante.org/code/drpt/wikis) will convert a (`salza`) semi-distance matrix into a true distance matrix.
|
|
|
|
|
|
# Licensing information
|
|
|
|
|
|
`salza` is released as is, without any warranty, under a dual licensing scheme.
|
... | ... | @@ -45,6 +47,14 @@ Once `oops` is compiled and installed, clone the `git` tree: |
|
|
git clone https://forge.uvolante.org/code/salza.git
|
|
|
```
|
|
|
|
|
|
# Reproducing the figures in the preprint
|
|
|
|
|
|
## Getting the datasets
|
|
|
|
|
|
## Reproducing the asymmetry heatmap
|
|
|
|
|
|
## Reproducing the classification table
|
|
|
|
|
|
# References
|
|
|
|
|
|
[^1]: Jacob Ziv and Abraham Lempel, _"A Universal Algorithm for Sequential Data Compression"_, IEEE Transactions on Information Theory, vol. 23, No. 3, pp. 337--343, May 1977.
|
... | ... | |