C++ Logo


Advanced search

Re: [SG19] P1709R3: Graph Library paper (for June 2020)

From: Phil Ratzloff <Phil.Ratzloff_at_[hidden]>
Date: Fri, 6 Mar 2020 17:45:45 +0000
The purpose of the library is for general-purpose graph algorithms.

While graph data structures are an important role for neural networks, I believe that is outside the scope of what we are trying to accomplish at this time. I’ll bring it up in our next SG19 meeting next week to make sure I’m aligned with others’ goals.

(Sorry for the late reply. I was out for a long break and am just getting caught up on me email.)

From: Patrik Huber <patrikhuber_at_[hidden]>
Sent: Thursday, February 13, 2020 7:30 PM
To: Phil Ratzloff <pratzl_at_[hidden]>; Phil Ratzloff <Phil.Ratzloff_at_[hidden]>
Cc: sg19_at_[hidden]g
Subject: Re: [SG19] P1709R3: Graph Library paper (for June 2020)


I'm just wondering, can you implement ONNX (https://onnx.ai/<https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fonnx.ai%2F&data=02%7C01%7CPhil.Ratzloff%40sas.com%7C7e903fdb0e0f48e2498208d7b0e4ff81%7Cb1c14d5c362545b3a4309552373a0c2f%7C0%7C0%7C637172369905109997&sdata=TeB8%2FA8js45N%2BWbJECXMInA4n6dEaIpWU7PUKyCLbGA%3D&reserved=0>) with this? That would be a proof of concept that this proposal is useful for neural networks, and you might discover whether you need to improve anything to make this proposal useful in practice.

PS: Frankfurt is spelled without an "ü", I am guessing that's not on purpose ;-).

Best wishes,

On Tue, 14 Jan 2020 at 00:22, Phil Ratzloff via SG19 <sg19_at_[hidden]<mailto:sg19_at_[hidden]>> wrote:
The R2 version, P1709R2: Graph Library<https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F1swsePTH_Ci_BNn_40tn0uX1e2pK916rPoYDNhdOnQd4%2Fedit%3Fusp%3Dsharing&data=02%7C01%7CPhil.Ratzloff%40sas.com%7C7e903fdb0e0f48e2498208d7b0e4ff81%7Cb1c14d5c362545b3a4309552373a0c2f%7C0%7C0%7C637172369905109997&sdata=tW%2BKN26pZQJfCDGsuEa2xDU5MLRH9AtO1opH3j19urM%3D&reserved=0>, has been submitted for review in Prague.

P1709R3: Graph Library<https://nam02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fdocs.google.com%2Fdocument%2Fd%2F1kLHhbSTX7j0tPeTYECQFSNx3R35Mu3xO5_dyYdRy4dM%2Fedit%3Fusp%3Dsharing&data=02%7C01%7CPhil.Ratzloff%40sas.com%7C7e903fdb0e0f48e2498208d7b0e4ff81%7Cb1c14d5c362545b3a4309552373a0c2f%7C0%7C0%7C637172369905119987&sdata=2po49eC5r%2FJ7ZI4jnbj5B%2Bo3%2Bc35OCwcR6QkcFWjmq0%3D&reserved=0>, targeted for the June 2020 meeting in Varna, is the next version that is actively being worked on. Don’t expect much activity during February as I will be out of the country for a large portion of the month.

Comments or questions about specific things in the document should be added as comments to the document. More general comments/questions should be directed to the SG19 mail list, or attend one of the monthly teleconferences. You can also email me directly if you like.

Feel free to make simple corrections for grammar or clarity. Please coordinate with me if you want to make larger changes to the paper.

My goal for the next paper is to have it 95% complete, including:

  1. Full descriptions of all ranges & algorithms, including examples.
  2. Inclusion of graph data structure(s) for directed and undirected graphs based on compressed sparse row.
  3. Relaxing requirements on ranges and algorithms for the types used for key, vertex container and edge containers. (I expect this to be an ongoing activity in later papers.)

Stretch goals

  1. Have a prototype implementation for review.
  2. Support gcc & clang in linux, in addition to MSVC. The compilers & platforms supported is dependent on getting the paper ready for serious review, and support for concepts & ranges in the compilers & their libraries. I expect to focus on gcc & clang in Linux.

SG19 mailing list

Received on 2020-03-06 11:48:31