Features#
Differential Dynamic Microscopy#
FastDDM can perform the analysis of image sequences using Differential Dynamic Microscopy.
It provides means to load images on disk in various formats, namely the ones available from the
scikit-image <https://scikit-image.org/> package and the Nikon nd2 files.
It efficiently computes then image structure function and the azimuthal average and provides tools
for the fit and analysis.
Python package#
FastDDM is a Python package and is designed to interoperate with other packages in the scientific Python ecosystem and to be extendable in user scripts. To enable interoperability, all operations provide access to useful computed quantities as properties in native Python types or numpy arrays where appropriate.
CPU and GPU devices#
FastDDM can execute the computation of the image structure function on CPUs or GPUs. Typical analysis run more efficiently on GPUs for image sequences containing more than a few hundreds images. The provided binaries support NVIDIA GPUs. All other operations are executed on the CPU using standar Python libraries.
Autotuned kernel parameters#
FastDDM automatically tunes the kernel parameters to improve performance when executing on a GPU device. It also tunes the computational parameters to optimize the memory consumption (using all the RAM available) when executing on a CPU (using the C++ core) or on a GPU. The functions provide the same output regardless of the parameter (within floating point precision).
The optimal parameters can depend on the number and the size of the images in the sequence and they weakly depend on other parameters such as the bit depth.
Mixed precision#
FastDDM can perform the computation of the image structure function with mixed floating point precision. If the single precision library is installed, all calculations are performed in double precision and the data are then transferred from and to the GPU in single precision. In the case of calculations performed on the CPU, all mathematical operations requiring high precision are performed in double precision mode.