Deformetrica v4.2 is out !

354 views
Skip to first unread message

Deformetrica

unread,
Apr 18, 2019, 1:58:37 PM4/18/19
to Deformetrica
Hello everyone,

The Deformetrica team is proud to announce that version 4.2.0 is now officially released and ready for download on anaconda ! 🍾

A lot has happened since the last Deformetrica release. Here are a few of the major changes:
- Performance improvements to all models
- Multi-processing and multi-gpu to deterministic atlas and longitudinal atlas models
- Dependencies update (Pytorch 1.0, Pykeops 1.0, ...)
- VTK reader
- Python 3.7 compliant
- Added the MCMC-SAEM estimator along with the longitudinal atlas model
- Bug fixes

The full CHANGELOG can be found on our gitlab repository.


New Gitlab.com repository

As you might have noticed from the link above, we have migrated our repository to Gitlab.com. This implies a small change to your git remote if you have cloned the repository. 
You may update your git remotes by using the following command: 
git remote set-url origin git@gitlab.com:icm-institute/aramislab/deformetrica.git

You can check that the previous command has been effectively applied by running:
git remote-v


Try the API online !

An online Python notebook is now available on Google colab. This enables anyone to try out Deformetrica's python API directly within a online Python notebook without having to install anything on your computer ! A little bonus is that GPU acceleration is available using Nvidia Tesla K80 cards ! Keep in mind that long running instances can by shutdown by Google. This should only be used for testing.


Documentation

Our documentation has been updated. You can view the wiki here.



Any constructive feedback is welcome. 
You can continue using this dedicated Google groups for questions or submit an issues using our Gitlab issue tracker.


The Deformetrica team




Amanda Yue Pan

unread,
Jul 25, 2019, 1:51:27 PM7/25/19
to Deformetrica
Dear Deformetrica Group,

Thank you for your great software.
I am trying to use the latest version of deformetrica on a cluster. When I use a keops kernel, I always get a segmentation fault error, the torch kernel works great. I would appreciate it if you can give me some hints about what the problem could be.

Best,
Amanda

Benoit Martin

unread,
Jul 26, 2019, 3:22:47 AM7/26/19
to Deformetrica
Hello Amanda,
Thank you for using Deformetrica !
It is indeed unfortunate that you are experiencing issues while using the keops kernel.
Could you give me more information about your configuration ? 
Please keep in mind that the keops kernel requires a few extra dependencies used to compile. Could you check that CMake, cuda a C++ compiler (g++) is installed ?

Bests,
Benoit

Reply all
Reply to author
Forward
0 new messages