Skip to content

Possible broadcasting bug in version 2.1.1 #36

@CaseyMcGrath

Description

@CaseyMcGrath

I noticed that I'm getting a new broadcasting error message which I think coincides with the recent version update to PTMCMCSampler. I wrote a very simple toy model problem with only three parameters, and I do not get the following error when using version 2.0.0. However, when using 2.1.1, the following bit of code:

injection = [2, 5, 0.5]
burnin = 100
it = 10000
sampler.sample(np.asarray(injection), it, burn=burnin, thin=1, SCAMweight=1, AMweight=1, DEweight=1)

Now returns the following error message:


ValueError Traceback (most recent call last)
in
2 it = 10000
3
----> 4 sampler.sample(np.asarray(injection), it, burn=burnin, thin=1, SCAMweight=1, AMweight=1, DEweight=1) #covUpdate=500,

2 frames
/usr/local/lib/python3.7/dist-packages/PTMCMCSampler/PTMCMCSampler.py in _updateDEbuffer(self, iter, burn)
756
757 self._DEbuffer = shift_array(self._DEbuffer, -len(self._AMbuffer)) # shift DEbuffer to the left
--> 758 self._DEbuffer[-len(self._AMbuffer) :] = self._AMbuffer # add new samples to the new empty spaces
759
760 # SCAM jump

ValueError: could not broadcast input array from shape (1000,3) into shape (100,3)

Metadata

Metadata

Assignees

No one assigned

    Labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions