[QST] KalmanFilter and ParticleFIlter CuPy and cuSignal application opportunities

[apiszcz]

• Stonesoup is a python framework for state estimation and tracking using Kalman and Particle filters.
• Typically stonesoup measurements are iterated over for evaluation.
• Is there any possibility that CuPy or cuSignal projects would be interested in review of Stonesoup to provide or implement/prototype recommendations for GPU acceleration options?

The measures.py file contains the Mahalanbois class that consumes a significant time in my profiling of the framework. https://github.com/dstl/Stone-Soup/blob/main/stonesoup/measures.py

[awthomp]

Hey @apiszcz -- Thanks for the question, and I'm always happy to chat about the future direction of cuSignal. Do you have any specific functions from Stonesoup that you'd like to see GPU accelerated? I know you mentioned that measures takes a lot of time; what specifically would you like to see us accelerate first?

You might also want to browse the documentation of another RAPIDS library, cuml, as it supports various distance metrics: https://docs.rapids.ai/api/cuml/stable/api.html#metrics-regression-classification-and-distance

[apiszcz]

I reviewed those metrics, they are great, however I did not locate, Mahalanobis or SquaredGaussianHellinger,

[github-actions[bot]]

This issue has been labeled inactive-30d due to no recent activity in the past 30 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed. This issue will be labeled inactive-90d if there is no activity in the next 60 days.

[github-actions[bot]]

This issue has been labeled inactive-90d due to no recent activity in the past 90 days. Please close this issue if no further response or action is needed. Otherwise, please respond with a comment indicating any updates or changes to the original issue and/or confirm this issue still needs to be addressed.