In this work, we have introduced a new approach for distributed matrix computations using the structure of convolutional codes. Using this approach we have developd a numerically robust and efficient (fast) schemes for straggler mitigation in distributed matrix-vector and matrix-matrix multiplication.
You can see details in our paper here https://arxiv.org/abs/1907.08064 or https://ieeexplore.ieee.org/abstract/document/9478901. If you are using these codes for distributed matrix-matrix or matrix-vector muliplication, please cite us-
Anindya Bijoy Das, Aditya Ramamoorthy, and Namrata Vaswani, “Efficient and robustdistributed matrix computations via convolutional coding,” in IEEE Transactions on Information Theory, vol. 67, no. 9, pp. 6266-6282, Sept. 2021.