3 ms·

I continued learning about NetworkX, and when it comes to issues with scaling and the need for persistence when working on applications in production, Memgraph

I continued learning about NetworkX, and when it comes to issues with scaling and the need for persistence when working on applications in production, Memgraph saves the day. You can see the previous discussion at https://news.ycombinator.com/item?id=33463472 https://news.ycombinator.com/item?id=33463472.

mikkom 4y ago
Are you working for memgraph perhaps? It certainly seems so based on your posts. https://news.ycombinator.com/submitted?id=katelatte https://news.ycombinator.com/submitted?id=katelatte Interesting how you got so many upvotes for this content that basically seems a lot like advertisement for memgraphs algorithm.
- katelatte 4y ago
  Yes, I work for Memgraph, I am a developer there and I wrote this, and all of the previously published articles. I was comparing NetworkX to Memgraph algorithms, since that was the point of the whole article. I am mostly using Python in my day-to-day job and I love what they did with NetworkX. This article was influenced by many people who use NetworkX and are a part of Memgraph community. I just wanted to see how much of a difference does the underlying C++ implementation of Memgraph makes. Since I work with Python tools and Memgraph every day, and talk with a bunch of people working on graph analytics, it makes sense to compare by myself and get the facts right.
  - rkwz 4y ago
    Thanks for the context. Offtopic, does Memgraph have something similar to NetworkX's connected components [1]? Wondering what's the performance difference between both for different sizes of graphs. [1] https://networkx.org/documentation/stable/reference/algorithms/generated/networkx.algorithms.components.connected_components.html https://networkx.org/documentation/stable/reference/algorith...
    - katelatte 4y ago
      We do have our own implementation of weakly connected components [1]. Currently, we only have NetworkX strongly connected components algorithm [2] as a part of the nxalg module (set of procedures) in MAGE (our graph algorithms library). I did not compare it yet, let me know if you do! We definitely need to create official benchmarks. Lot of work! [1] https://memgraph.com/docs/mage/query-modules/cpp/weakly-connected-components https://memgraph.com/docs/mage/query-modules/cpp/weakly-conn... [2] https://memgraph.com/docs/mage/query-modules/python/nxalg#strongly_connected_components https://memgraph.com/docs/mage/query-modules/python/nxalg#st...
    - katelatte 4y ago
      I forgot to mention: we do have biconnected components algorithm [1], and since all biconnected graphs are strongly connected, it can be useful. [1] https://memgraph.com/docs/mage/query-modules/cpp/biconnected-components https://memgraph.com/docs/mage/query-modules/cpp/biconnected...

mikkom 4y ago

Are you working for memgraph perhaps? It certainly seems so based on your posts. https://news.ycombinator.com/submitted?id=katelatte https://news.ycombinator.com/submitted?id=katelatte Interesting how you got so many upvotes for this content that basically seems a lot like advertisement for memgraphs algorithm.

katelatte 4y ago
Yes, I work for Memgraph, I am a developer there and I wrote this, and all of the previously published articles. I was comparing NetworkX to Memgraph algorithms, since that was the point of the whole article. I am mostly using Python in my day-to-day job and I love what they did with NetworkX. This article was influenced by many people who use NetworkX and are a part of Memgraph community. I just wanted to see how much of a difference does the underlying C++ implementation of Memgraph makes. Since I work with Python tools and Memgraph every day, and talk with a bunch of people working on graph analytics, it makes sense to compare by myself and get the facts right.
- rkwz 4y ago
  Thanks for the context. Offtopic, does Memgraph have something similar to NetworkX's connected components [1]? Wondering what's the performance difference between both for different sizes of graphs. [1] https://networkx.org/documentation/stable/reference/algorithms/generated/networkx.algorithms.components.connected_components.html https://networkx.org/documentation/stable/reference/algorith...
  - katelatte 4y ago
    We do have our own implementation of weakly connected components [1]. Currently, we only have NetworkX strongly connected components algorithm [2] as a part of the nxalg module (set of procedures) in MAGE (our graph algorithms library). I did not compare it yet, let me know if you do! We definitely need to create official benchmarks. Lot of work! [1] https://memgraph.com/docs/mage/query-modules/cpp/weakly-connected-components https://memgraph.com/docs/mage/query-modules/cpp/weakly-conn... [2] https://memgraph.com/docs/mage/query-modules/python/nxalg#strongly_connected_components https://memgraph.com/docs/mage/query-modules/python/nxalg#st...
  - katelatte 4y ago
    I forgot to mention: we do have biconnected components algorithm [1], and since all biconnected graphs are strongly connected, it can be useful. [1] https://memgraph.com/docs/mage/query-modules/cpp/biconnected-components https://memgraph.com/docs/mage/query-modules/cpp/biconnected...