https://github.com/guillaumegenthial/clusters

ATLAS Calorimeter Cluster Splitting (PointNet, ConvNets) - Tensorflow
https://github.com/guillaumegenthial/clusters

Last synced: about 24 hours ago
JSON representation

ATLAS Calorimeter Cluster Splitting (PointNet, ConvNets) - Tensorflow

• Host: GitHub
• URL: https://github.com/guillaumegenthial/clusters
• Owner: guillaumegenthial
• Created: 2017-02-03T01:17:31.000Z (almost 8 years ago)
• Default Branch: master
• Last Pushed: 2017-03-20T23:14:56.000Z (almost 8 years ago)
• Last Synced: 2024-11-05T15:50:29.880Z (about 2 months ago)
• Language: Python
• Homepage:
• Size: 518 KB
• Stars: 1
• Watchers: 2
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# ATLAS Calorimeter Cluster Splitting

## Modules

ROOT / Tensorflow

## Task

- Predict number of particle in ATLAS topo cluster

## Data

- 2k Monte-Carlo simulated events in a ROOT tree

- 600k extracted clusters in jets > 20 GeV and eta < 0.5

- input data:

    - simple features

    - fill np arrays with energy deposit

To construct the data from ROOT -> pickle, run 

```

python build_data.py

```

## Models

- Fully connected network on simple features

- Fully connected network on flattened np arrays

- Convolutional layers on np arrays

- rotation invariant extraction from embedding of cells

## Test

Add flag -t

```

python simple.py -t

``` 

## Run

- for simple features, with `config` being the name of the config file in `/configs/`

```

python simple.py -c config

```

- for layers features (stack of arrays for each layer)

```

python layers.py -c config

```

- for embeddings

```

python embeddings.py -c config

```