megbedell / wobble

precise data-driven RV fitting, now with tellurics!
https://wobble.readthedocs.io
MIT License
39 stars 16 forks source link

Running Wobble optimization gives segmentation fault #60

Open vincentvaneylen opened 5 years ago

vincentvaneylen commented 5 years ago

Trying to run a simple wobble run results in a segmentation fault (core dumped) as soon as the optimization starts. Running on a Dell XPS on Ubuntu 18.04.

Script I am running:

import wobble data = wobble.Data('51peg_e2ds.hdf5', orders=[10,20,30]) # website example data results = wobble.Results(data=data) for r in range(data.R): model = wobble.Model(data, results, r) model.add_star('star') model.add_telluric('tellurics') wobble.optimize_order(model, niter=10)

Resulting output gives some Tensorflow messages (which I am not sure are related), then segmentation fault seemingly as soon as optimization starts. Output:

WARNING:tensorflow:From /home/vincent/.local/lib/python2.7/site-packages/tensorflow/python/framework/op_def_library.py:263: colocate_with (from tensorflow.python.framework.ops) is deprecated and will be removed in a future version. Instructions for updating: Colocations handled automatically by placer. 2019-03-29 13:23:00.181542: I tensorflow/core/platform/cpu_feature_guard.cc:141] Your CPU supports instructions that this TensorFlow binary was not compiled to use: AVX2 FMA 2019-03-29 13:23:00.203764: I tensorflow/core/platform/profile_utils/cpu_utils.cc:94] CPU Frequency: 2112000000 Hz 2019-03-29 13:23:00.204216: I tensorflow/compiler/xla/service/service.cc:150] XLA service 0x556fb6bc4b20 executing computations on platform Host. Devices: 2019-03-29 13:23:00.204238: I tensorflow/compiler/xla/service/service.cc:158] StreamExecutor device (0): , optimize: iterating through 10 optimization steps... 0%| | 0/10 [00:00<?, ?it/s]Segmentation fault (core dumped)