| [ Information ] [ Publications ] [Signal processing codes] [ Signal & Image Links ] | |
| [ Main blog: A fortunate hive ] [ Blog: Information CLAde ] [ Personal links ] | |
| [ SIVA Conferences ] [ Other conference links ] [ Journal rankings ] | |
| [ Tutorial on 2D wavelets ] [ WITS: Where is the starlet? ] | |
If you cannot find anything more, look for something else (Bridget Fountain) |
|
|
|
|
With the steady increase of scientific data generated by simulation processes and by measurement, researchers are looking for lossy compression methods to handle a floating data hard to compress in a lossless approach. Besides, the compression is no longer limited to handle the outputs to optimize their storage, transmission, but could be applied at different stages of the simulation workflow to optimize the process constrained in HPC by the memory, and the bandwidth of the existing computation tools. It turns out the lossy compression could eliminate a part of the noise from scientific data, coming from a precision willingness, and having for consequence to induce a false precision unnecessary to the scientific usage. This study illustrates the preceding remark and the potential of a compression on a property used as input for a flow simulation currently made in Geosciences. The approach based on discrete wavelet transformation and zerotree coding is compared with promising compression tools in the scientific field: SZ, ZFP, by use of objective metrics and subjective validations assessed by simulation results. Finally there is obvious in our case that some preprocessing as compandor could considerably improve the compression performances.