tri-amdd / camd Goto Github PK
View Code? Open in Web Editor NEWAgent-based sequential learning software for materials discovery
License: Apache License 2.0
Agent-based sequential learning software for materials discovery
License: Apache License 2.0
Dependabot couldn't authenticate with https://pypi.python.org/simple/.
You can provide authentication details in your Dependabot dashboard by clicking into the account menu (in the top right) and selecting 'Config variables'.
Right now CI is limited to one test at a time because we're storing the state of the worker, meta_agent_test in the same place. I think we should provision the tests to use a unique prefix so tests can run in parallel (e. g. with a uuid or something).
CAMD is not currently windows compatible because of the qmpy dependency. GH Actions enables a relatively straightforward windows CI that should be enabled once this is resolved.
As of right now, the campaign bears the responsibility of choosing data. This is a decision, and I believe should be the purview of the agent. Also, the current initialization logic does not guarantee that the seed will be populated (for example if all of the experiments in the initialization fail). I think we should discuss initialization logic, where it belongs, and how to implement it.
I am running Ubuntu 18.04.5 LTS, I created the python3.7 virtual environment for the installation:
mkvirtualenv -p /usr/bin/python3.7 camd
I installed pymatgen==2020.12.31
and proceeded
git clone https://github.com/TRI-AMDD/CAMD.git
cd CAMD
pip install numpy
pip install -r requirements.txt
python setup.py develop
I ran the tri hackathon 2020 tutorial here , in CAMD_tutorial.ipynb, when running the cell:
# Load the data
from hackathon.helper import load_tutorial_data
data = load_tutorial_data()
# Inspect the first five rows
data.head()
I get the error:
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
<ipython-input-38-adaecc4da224> in <module>
1 # Load the data
----> 2 from hackathon.helper import load_tutorial_data
3 data = load_tutorial_data()
4 # Inspect the first five rows
5 data.head()
~/tri-hackathon-2020/hackathon/helper.py in <module>
1 # Copyright 2020 Toyota Research Institute
2
----> 3 from camd.analysis import AnalyzerBase
4 from matminer.datasets.convenience_loaders import load_elastic_tensor
5
~/CAMD/camd/analysis.py in <module>
18 from qmpy.analysis.thermodynamics.space import PhaseSpace
19 from multiprocessing import Pool, cpu_count
---> 20 from pymatgen import Composition
21 from pymatgen.entries.computed_entries import ComputedEntry
22 from pymatgen.analysis.phase_diagram import (
ImportError: cannot import name 'Composition' from 'pymatgen' (unknown location)
Please advise.
The issue line here.
apply_gradients
expects a list, but receives an iterator (from call to zip, which returns an iterator). Should we convert to list for Python 3 compatibility?
I think we should refactor to use plotly's backend for phase diagram plotting as pymatgen does now by default.
As I was reading the manuscript, I was curious about what a comparison with Dragonfly via RayTune or Ax (latter would just be for small data experiments).
Coverage in the docker container we're using for testing doesn't seem to be working. Will investigate and update this issue.
We're using gpflow 1.5's implementation and API for the SVGP, but the api has changed significantly for their latest version. There's some info on this here, which should be able to partially guide us, but will likely take some time.
Currently, CAMD depends on a custom fork of qmpy, which was used to ensure python3 compatibility. qmpy has recently updated their github repo to be python3 compatible, but I don't think they've released yet. When they do, we should explore reintegrating the official version.
Dependabot couldn't authenticate with https://pypi.python.org/simple/.
You can provide authentication details in your Dependabot dashboard by clicking into the account menu (in the top right) and selecting 'Config variables'.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.