Cluster.py and 0-d Arrays

[cxhernandez]

Getting a weird error when I run Cluster.py at line 136:

136  ->         ptraj = np.concatenate(ptraj)
*** ValueError: 0-d arrays can't be concatenated

ptraj is definitely not empty though.

[schwancr]

Can you send he command you're running that gives you this error.

On Fri, Nov 22, 2013 at 12:25 AM, Carlos Hernández <notifications@github.com

wrote:

Getting a weird error when I run Cluster.py at line 136:

136 -> ptraj = np.concatenate(ptraj)*\ ValueError: 0-d arrays can't be concatenated

ptraj is definitely not empty though.

— Reply to this email directly or view it on GitHubhttps://github.com/SimTk/msmbuilder/issues/262 .

[schwancr]

the*

On Fri, Nov 22, 2013 at 12:27 AM, Christian Schwantes <schwancr@stanford.edu

wrote:

Can you send he command you're running that gives you this error.

On Fri, Nov 22, 2013 at 12:25 AM, Carlos Hernández < notifications@github.com> wrote:

Getting a weird error when I run Cluster.py at line 136:

136 -> ptraj = np.concatenate(ptraj)*\ ValueError: 0-d arrays can't be concatenated

ptraj is definitely not empty though.

— Reply to this email directly or view it on GitHubhttps://github.com/SimTk/msmbuilder/issues/262 .

[cxhernandez]

 /hsgs/home/mlawrenz/epd-7.3-1-rh5-x86_64/bin/python2.7 -m pdb /hsgs/home/mlawrenz/epd-7.3-1-rh5-x86_64/bin/Cluster.py -s 250 -o d3/ lprmsd -a sirtuin_Calpha_indices.dat -l p53_Calpha_indices.dat hybrid -d 0.3 

[schwancr]

What are the contents of your atom indices files?

On Fri, Nov 22, 2013 at 12:28 AM, Carlos Hernández <notifications@github.com

wrote:

/hsgs/home/mlawrenz/epd-7.3-1-rh5-x86_64/bin/python2.7 -m pdb /hsgs/home/mlawrenz/epd-7.3-1-rh5-x86_64/bin/Cluster.py -s 250 -o d3/ lprmsd -a sirtuin_Calpha_indices.dat -l p53_Calpha_indices.dat hybrid -d 0.3

— Reply to this email directly or view it on GitHubhttps://github.com/SimTk/msmbuilder/issues/262#issuecomment-29056459 .

[cxhernandez]

rows of atom indices like this:

-bash-4.1$ head p53_Calpha_indices.dat 
3904
3915
3932
3951
3973
3984
4006
4028
4035
4052
-bash-4.1$ head sirtuin_Calpha_indices.dat 
4
21
43
60
82
97
117
136
148
167

[rmcgibbo]

I know that there is code that isn't robust to index files that contain only a single element, since they get loaded as zero-dim arrays by default.

$ cat a.txt
123
rmcgibbo@vspm9  ~ [sci]
$ python -c "import numpy as np; print np.loadtxt('a.txt', dtype=int).ndim"
0

[schwancr]

Can you do this on your Cluster.py script that you're calling. I think this is due to a bug that was fixed a while ago. Are you sure you have the right version of msmbuilder installed? For instance, my TPT branch may not have this bugfix pulled in yet

On Fri, Nov 22, 2013 at 12:32 AM, Carlos Hernández <notifications@github.com

wrote:

rows of atom indices like this:

-bash-4.1$ head p53_Calpha_indices.dat 3904 3915 3932 3951 3973 3984 4006 4028 4035 4052 -bash-4.1$ head sirtuin_Calpha_indices.dat 4 21 43 60 82 97 117 136 148 167

— Reply to this email directly or view it on GitHubhttps://github.com/SimTk/msmbuilder/issues/262#issuecomment-29056620 .

[schwancr]

Do this:

cat scripts/Cluster.py | grep "if isinstance(metric, metrics.Vectorized)"

On Fri, Nov 22, 2013 at 12:39 AM, Christian Schwantes <schwancr@stanford.edu

wrote:

Can you do this on your Cluster.py script that you're calling. I think this is due to a bug that was fixed a while ago. Are you sure you have the right version of msmbuilder installed? For instance, my TPT branch may not have this bugfix pulled in yet

On Fri, Nov 22, 2013 at 12:32 AM, Carlos Hernández < notifications@github.com> wrote:

rows of atom indices like this:

-bash-4.1$ head p53_Calpha_indices.dat 3904 3915 3932 3951 3973 3984 4006 4028 4035 4052 -bash-4.1$ head sirtuin_Calpha_indices.dat 4 21 43 60 82 97 117 136 148 167

— Reply to this email directly or view it on GitHubhttps://github.com/SimTk/msmbuilder/issues/262#issuecomment-29056620 .

[cxhernandez]

No, that line's not there. Maybe, Morgan has your branch then.... I'll pull the latest version and try it now.

[rmcgibbo]

Got it. This is the issue that was fixed in #248.

[schwancr]

The newest msmbuilder version should be fine. The TPT code should get pulled in soon, too

[cxhernandez]

Yep, that works! Thanks, and sorry about that guys!

[schwancr]

No worries! glad it's working.