torch-discounted-cumsum 1.1.0

Fast discounted cumulative sums in PyTorch

This package implements an efficient parallel algorithm for the computation of discounted cumulative sums with differentiable bindings to PyTorch. The cumsum operation is frequently seen in data science domains concerned with time series, including Reinforcement Learning (RL).

The traditional sequential algorithm performs the computation of the output elements in a loop. For an input of size N, it requires O(N) operations and takes O(N) time steps to complete.

The proposed parallel algorithm requires a total of O(N log N) operations, but takes only O(log N) time, which is a considerable trade-off in many applications involving large inputs.

Features of the parallel algorithm:

• Speed logarithmic in the input size
• Better numerical precision than sequential algorithms

Features of the package:

• CPU: sequential algorithm in C++
• GPU: parallel algorithm in CUDA
• Gradients computation for input and gamma
• Batch support for input and gamma
• Both left and right directions of summation supported
• PyTorch bindings

Find more details and the most up-to-date information on the project webpage: https://www.github.com/toshas/torch-discounted-cumsum