You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: CONTRIBUTING.md
+20-1Lines changed: 20 additions & 1 deletion
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -49,4 +49,23 @@ From "Software Best Practices Effective Version Control", Alex Olivas, IceCube B
49
49
* Bad commit message: `"blerg"`
50
50
51
51
Others:
52
-
* Keep backward compatibility in mind when you change code.
52
+
* Keep backward compatibility in mind when you change code.
53
+
54
+
## Experiment tracking
55
+
56
+
We're using [Weights & Biases](https://wandb.ai/) (W&B) to track the results — i.e. losses, metrics, and model artifacts — of training runs as a means to track model experimentation and streamline optimisation. To authenticate with W&B, sign up on the website and run the following in your terminal after having installed this package:
57
+
```bash
58
+
$ wandb login
59
+
```
60
+
You can use your own, personal projects on W&B, but for projects of common interest you are encouraged to join the `graphnet-team` team on W&B [here](https://wandb.ai/graphnet-team), create new projects for your specific use cases, and log your runs there. Just ask [@asogaard](https://github.com/asogaard) for an invite to the team!
61
+
62
+
If you don't want to use W&B and/or only want to log run data locally, you can run:
63
+
```bash
64
+
$ wandb offline
65
+
```
66
+
If you change you mind, it's as simple as:
67
+
```bash
68
+
$ wandb online
69
+
```
70
+
71
+
The [examples/04_training/01_train_model.py](examples/04_training/01_train_model.py) script shows how to train a model and log the results to W&B.
**GraphNeT** is an open-source Python framework aimed at providing high quality, user friendly, end-to-end functionality to perform reconstruction tasks at neutrino telescopes using graph neural networks (GNNs). GraphNeT makes it fast and easy to train complex models that can provide event reconstruction with state-of-the-art performance, for arbitrary detector configurations, with inference times that are orders of magnitude faster than traditional reconstruction techniques.
14
18
19
+
Feel free to join the [GraphNeT Slack group](https://join.slack.com/t/graphnet-team/signup)!
20
+
21
+
### Publications using GraphNeT
22
+
23
+
| Type | Title | DOI |
24
+
| --- | --- | --- |
25
+
| Paper | GraphNeT: Graph neural networks for neutrino telescope event reconstruction |[](https://joss.theoj.org/papers/eecab02fb1ecd174a5273750c1ea0baf)|
26
+
| Paper | Graph Neural Networks for low-energy event classification & reconstruction in IceCube |[](https://doi.org/10.1088/1748-0221/17/11/P11003)|
27
+
15
28
## :gear: Install
16
29
17
30
We recommend installing `graphnet` in a separate environment, e.g. using a Python virtual environment or Anaconda (see details on installation [here](https://www.anaconda.com/products/individual)). Below we prove installation instructions for different setups.
@@ -169,25 +182,6 @@ To make sure that the process of contributing is as smooth and effective as poss
169
182
In short, everyone who wants to contribute to this project is more than welcome to do so! Contributions are handled through pull requests, that should be linked to a [GitHub issue](https://github.com/graphnet-team/graphnet/issues) describing the feature to be added or bug to be fixed. Pull requests will be reviewed by the project maintainers and merged into the main branch when accepted.
170
183
171
184
172
-
## :test_tube: Experiment tracking
173
-
174
-
We're using [Weights & Biases](https://wandb.ai/) (W&B) to track the results — i.e. losses, metrics, and model artifacts — of training runs as a means to track model experimentation and streamline optimisation. To authenticate with W&B, sign up on the website and run the following in your terminal after having installed this package:
175
-
```bash
176
-
$ wandb login
177
-
```
178
-
You can use your own, personal projects on W&B, but for projects of common interest you are encouraged to join the `graphnet-team` team on W&B [here](https://wandb.ai/graphnet-team), create new projects for your specific use cases, and log your runs there. Just ask [@asogaard](https://github.com/asogaard) for an invite to the team!
179
-
180
-
If you don't want to use W&B and/or only want to log run data locally, you can run:
181
-
```bash
182
-
$ wandb offline
183
-
```
184
-
If you change you mind, it's as simple as:
185
-
```bash
186
-
$ wandb online
187
-
```
188
-
189
-
The [examples/04_training/01_train_model.py](examples/04_training/01_train_model.py) script shows how to train a model and log the results to W&B.
190
-
191
185
## :memo: License
192
186
193
187
GraphNeT has an Apache 2.0 license, as found in the [LICENSE](LICENSE) file.
Copy file name to clipboardExpand all lines: paper/paper.bib
+10-15Lines changed: 10 additions & 15 deletions
Original file line number
Diff line number
Diff line change
@@ -1,19 +1,13 @@
1
1
// GraphNeT Zenodo
2
-
@software{graphnet_zenodo:2022,
3
-
author = {Andreas Søgaard and
4
-
Rasmus F. Ørsøe and
5
-
Leon Bozianu and
6
-
Morten Holm and
7
-
Kaare Endrup Iversen and
8
-
Tim Guggenmos and
9
-
Martin Ha Minh and
10
-
Philipp Eller},
11
-
title = {GraphNeT},
12
-
month = jun,
13
-
year = 2022,
14
-
publisher = {Zenodo},
15
-
doi = {10.5281/zenodo.6720188},
16
-
url = {https://doi.org/10.5281/zenodo.6720188}
2
+
@software{Sogaard_GraphNeT_2023,
3
+
author = {Søgaard, Andreas and F. Ørsøe, Rasmus and Holm, Morten and Bozianu, Leon and Rosted, Aske and C. Petersen, Troels and Endrup Iversen, Kaare and Hermansen, Andreas and Guggenmos, Tim and Andresen, Peter and Ha Minh, Martin and Neste, Ludwig and Holmes, Moust and Pontén, Axel and Leonard DeHolton, Kayla and Eller, Philipp},
Copy file name to clipboardExpand all lines: paper/paper.md
+63-31Lines changed: 63 additions & 31 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -12,38 +12,70 @@ tags:
12
12
- neutrinos
13
13
14
14
authors:
15
-
- name: Andreas Søgaard
16
-
orcid: 0000-0002-0823-056X
17
-
affiliation: 1# "1, 2" (Multiple affiliations must be quoted)
18
-
corresponding: true
19
-
- name: Rasmus F. Ørsøe
20
-
orcid: 0000-0001-8890-4124
21
-
affiliation: 2
22
-
- name: Leon Bozianu
23
-
affiliation: 1
24
-
- name: Morten Holm
25
-
affiliation: 1
26
-
- name: Kaare Endrup Iversen
27
-
affiliation: 1
28
-
- name: Tim Guggenmos
29
-
affiliation: 2
30
-
- name: Martin Ha Minh
31
-
orcid: 0000-0001-7776-4875
32
-
affiliation: 2
33
-
- name: Philipp Eller
34
-
orcid: 0000-0001-6354-5209
35
-
affiliation: 2
36
-
- name: Troels C. Petersen
37
-
orcid: 0000-0003-0221-3037
38
-
affiliation: 1
15
+
- name: Andreas Søgaard
16
+
affiliation: 1
17
+
orcid: 0000-0002-0823-056X
18
+
corresponding: true
19
+
- name: Rasmus F. Ørsøe
20
+
affiliation: "1, 2"
21
+
orcid: 0000-0001-8890-4124
22
+
- name: Morten Holm
23
+
affiliation: 1
24
+
orcid: 0000-0003-1383-2810
25
+
- name: Leon Bozianu
26
+
affiliation: 1
27
+
orcid: 0000-0002-1243-9980
28
+
- name: Aske Rosted
29
+
affiliation: 3
30
+
orcid: 0000-0003-2410-400X
31
+
- name: Troels C. Petersen
32
+
affiliation: 1
33
+
orcid: 0000-0003-0221-3037
34
+
- name: Kaare Endrup Iversen
35
+
affiliation: 1
36
+
orcid: 0000-0001-6533-4085
37
+
- name: Andreas Hermansen
38
+
affiliation: 1
39
+
orcid: 0009-0006-1162-9770
40
+
- name: Tim Guggenmos
41
+
affiliation: 2
42
+
- name: Peter Andresen
43
+
affiliation: 1
44
+
orcid: 0009-0008-5759-0490
45
+
- name: Martin Ha Minh
46
+
affiliation: 2
47
+
orcid: 0000-0001-7776-4875
48
+
- name: Ludwig Neste
49
+
affiliation: 4
50
+
orcid: 0000-0002-4829-3469
51
+
- name: Moust Holmes
52
+
affiliation: 1
53
+
orcid: 0009-0000-8530-7041
54
+
- name: Axel Pontén
55
+
affiliation: 5
56
+
orcid: 0009-0008-2463-2930
57
+
- name: Kayla Leonard DeHolton
58
+
affiliation: 6
59
+
orcid: 0000-0002-8795-0601
60
+
- name: Philipp Eller
61
+
affiliation: 2
62
+
orcid: 0000-0001-6354-5209
39
63
40
64
affiliations:
41
-
- name: Niels Bohr Institute, University of Copenhagen, Denmark
42
-
index: 1
43
-
- name: Technical University of Munich, Germany
44
-
index: 2
45
-
46
-
date: 16 September 2022
65
+
- name: Niels Bohr Institute, University of Copenhagen, Denmark
66
+
index: 1
67
+
- name: Technical University of Munich, Germany
68
+
index: 2
69
+
- name: Chiba University, Japan
70
+
index: 3
71
+
- name: Technical University of Dortmund, Germany
72
+
index: 4
73
+
- name: Uppsala University, Sweden
74
+
index: 5
75
+
- name: Pennsylvania State University, USA
76
+
index: 6
77
+
78
+
date: 12 May 2023
47
79
48
80
bibliography: paper.bib
49
81
@@ -53,7 +85,7 @@ bibliography: paper.bib
53
85
54
86
Neutrino telescopes, such as ANTARES [@ANTARES:2011hfw], IceCube [@Aartsen:2016nxy; @DeepCore], KM3NeT [@KM3Net:2016zxf], and Baikal-GVD [@Baikal-GVD:2018isr] have the science goal of detecting neutrinos and measuring their properties and origins. Reconstruction at these experiments is concerned with classifying the type of event or estimating properties of the interaction.
55
87
56
-
`GraphNeT`[@graphnet_zenodo:2022] is an open-source Python framework aimed at providing high quality, user friendly, end-to-end functionality to perform reconstruction tasks at neutrino telescopes using graph neural networks (GNNs). `GraphNeT` makes it fast and easy to train complex models that can provide event reconstruction with state-of-the-art performance, for arbitrary detector configurations, with inference times that are orders of magnitude faster than traditional reconstruction techniques [@gnn_icecube].
88
+
`GraphNeT`[@Sogaard_GraphNeT_2023] is an open-source Python framework aimed at providing high quality, user friendly, end-to-end functionality to perform reconstruction tasks at neutrino telescopes using graph neural networks (GNNs). `GraphNeT` makes it fast and easy to train complex models that can provide event reconstruction with state-of-the-art performance, for arbitrary detector configurations, with inference times that are orders of magnitude faster than traditional reconstruction techniques [@gnn_icecube].
57
89
58
90
GNNs from `GraphNeT` are flexible enough to be applied to data from all neutrino telescopes, including future projects such as IceCube extensions [@IceCube-PINGU:2014okk; @IceCube:2016xxt; @IceCube-Gen2:2020qha] or P-ONE [@P-ONE:2020ljt]. This means that GNN-based reconstruction can be used to provide state-of-the-art performance on most reconstruction tasks in neutrino telescopes, at real-time event rates, across experiments and physics analyses, with vast potential impact for neutrino and astro-particle physics.
0 commit comments