Good day,

My name is Denis Shchepakin. I have a background in mathematics and software 
development, and I currently work as an Applied Scientist at Amazon. The Otava 
project caught my attention, and this thread in particular looks like a good 
opportunity for me to contribute.

To better understand the request, I reviewed a few relevant papers (notably 
Matteson & James 2013 and Fleming et al. 2023). Based on that, I’d like to 
propose the following plan in two phases:

1. Phase 1 - Core reimplementation.
Re-implement the core e-divisive solution from Matteson & James (2013) directly 
within Otava. This would effectively remove the dependency on the 
signal_processing_algorithms repository without requiring changes to existing 
code, apart from imports.

2. Phase 2 - Incremental optimizations.
Once the baseline implementation is in place, extend Otava with optimizations. 
For example, the current EDivisive class computes pairwise distances in O(n^2). 
When a new (n+1)’th point arrives, we can avoid reconstructing the entire 
structure for O((n+1)^2) by only computing the O(n) new distances involving 
that point.
I’d be happy to start with Phase 1 and share progress for review. Please let me 
know if this direction aligns with the project’s priorities.

Best regards,
Denis

On 2025/09/22 18:36:03 Henrik Ingo

Reply via email to