test FAML

[xulabs]

see issue #13

can test on existing data https://github.com/xulabs/aitom/blob/master/tutorials/000_general.md

also on simulated subtomograms https://github.com/xulabs/aitom/blob/master/tutorials/003_simulate_subtomogram.py

[buptwhr]

faml works as expected on Demo dataset As to simulated subtomograms, I can't execute it due to the absence of aitom.tomominer.simulation

the same problem as #8 @xulabs

[xulabs]

@buptwhr to simulate the subtomograms please use the tomominer package pre-installed on the server.

[xulabs]

@buptwhr read FAML paper and see if you can understand the method. Then see if the output of FAML makes sense.

[buptwhr]

There seems a few bugs in faml.py:

1. I was trying to run the model for 20 iterations as the paper said. When it comes to the 12th iteration, some of the parameters(Alpha, sigma, Xi, A) became nan or very close to zero. Then the exception happened.
2. Then I set the iterations to 10. When evaluating the model, there also seems a few bugs associated with data type in these two functions. Maybe we can talk about it with more details at our weekly meeting. https://github.com/xulabs/aitom/blob/d42ea83b5bf3eb26ec2e7324e9b4e694130ae630/aitom/average/ml/faml/faml.py#L792 https://github.com/xulabs/aitom/blob/d42ea83b5bf3eb26ec2e7324e9b4e694130ae630/aitom/average/ml/faml/faml.py#L808 @xulabs

[xulabs]

@xiangruz please take a look at this bug reported by @buptwhr

[buptwhr]

AttributeError: module 'aitom.align.fast.util' has no attribute 'fast_align' @xulabs Please add the code for it.

[xulabs]

@buptwhr fixed