Impala is a fully parallelized algorithm for multiple point statistics simulation which is both fast and compact in terms of RAM requirement.
Current version (Impala 2.1) has the following features:
- Uses a hybrid tree-list method allowing to optimize the storage and retrieving of the statistics computed on the training image,
- Fully parallelized (under the OpenMP and MPI protocols),
- Proposes new methods for conditioning to existing hard data,
- Can handle incomplete training images, opening the way to image reconstruction,
- Offers a wide choice of simulation path strategies,
- Proposes an Automatic Template Reduction (ATR), to optimize the template size according to the multigrid levels,
- Embeds a novel algorithm to account for secondary attribute,
- Can account for non-stationarity, using a-priori expert information,
- Offers the possibility to respect local or global target proportions.
Non-exclusive licences for the distribution Impala have already been granted to the following software houses: