Candmaker doesn't centre the wide pulses (width > 2) if DM is low and native time sampling of data is large

[Sujay Mate]

Describe the bug
I am running your_candmaker on a data with ~16 ms native time-sampling and it doesn't centre the pulse if the width is > 2 and DM is low. After little bit of debugging I think there is a bug in the candmaker.py in the if block below:

your/your/candidate.py

Lines 191 to 205 in 8845caf

 nstart = int(tstart / self.native_tsamp) 
 nsamp = int((tstop - tstart) / self.native_tsamp) 
 nsamp_read = nsamp 
  
 if for_preprocessing: 
     if self.width > 2 and nsamp_read // (self.width // 2) < self.min_samp: 
         nsamp_read = self.min_samp * self.width // 2 
         nstart_read = nstart - (nsamp_read - nsamp) // 2 
     if nsamp_read < self.min_samp: 
         nsamp_read = self.min_samp 
         nstart_read = nstart - (nsamp_read - nsamp) // 2 
     else: 
         nstart_read = nstart 
 else: 
     nstart_read = nstart 

I think this happens only if the DM of the pulse is low and time sampling is large. In this case, the nsamp is less than self.min_samp. So if the width is > 2, the if block at line 196 is evaluated as true. Then the variables nsamp_read and nstart_read are redefined. In this case the if/else block on line 199 and 202 is treated as independent block and the else is evaluated to be true as nsamp_read > self.min_samp setting the nsamp_read back to nstart. This means the pulse is not centred. A simple fix is to replace if statement on line 199 with elif so the entire if/else block between 196 to 202 is treated as one and not two independent ones.

To Reproduce
Running candmaker on pulses with width > 2, low DM and large native time-sampling.

Expected behavior
I think the if statement on line 199 above should be replaced with elif. Screenshots of buggy behaviour and expected behaviour below.

Screenshots
An example of the bug. In this case, the nsample_read is reset by the else block.

If the second if is replaced with elif, then the behaviour is correct and value set by the first if block will be used as final one (which is the correct one)

Versions (please provide the versions of the the following packages):

• your: 0.6.7
• matplotlib: 3.3.2
• numpy: 1.19.2
• h5py: 2.10.0
• scikit-image: Not using
• scipy: 1.6.1
• numba: 0.51.2
• astropy: 4.0.2
• pandas: 1.1.3
• rich: 12.6.0

If someone from the devs confirm the bug, I can fix it and create a pull request.

[Devansh Agarwal]

Hi, @sujaymate please go ahead and submit a PR. Thanks.

[Sujay Mate]

Hi @devanshkv , created the PR.

[Devansh Agarwal]

Fixed with #105/