stumpy.motifs IndexError and unexpected results

[chvalovak]

Hello,
I am trying to use stupmy.motifs but I encountered a problem. With this setting of stumpy.motifs

dists, indices = stumpy.motifs(timeseries["value"], mp['mp1'], max_matches=5, max_motifs=20)

it produces 5 motifs with 5 matches and then this error occures:

IndexError: too many indices for array: array is 1-dimensional, but 2 were indexed

The array query_matches is empty (error occures in motifs.py).

I tried running stumpy.motifs with benevolent parameters, like this:

dists, indices = stumpy.motifs(timeseries["value"], mp['mp1'],max_matches=5, max_motifs=20, cutoff=np.inf, max_distance=np.inf, min_neighbors=1)

This setting did not invoke the error but it discovered only ONE motif. Any idea what can cause this behavior?

Thanks in advance for any reply!

[seanlaw]

@chvalovak Thank you for your question and welcome to the STUMPY community. Are you able to provide a small sample time series and matrix profile input that reproduces the error? Without your inputs, it is hard to know what is causing the error

[NimaSarajpoor]

@seanlaw
As you said, it is hard to say without having the inputs. So, what I am about to say is just a guess.

I think I encountered a similar issue before. And, we fixed it. See below:

stumpy/stumpy/motifs.py

Lines 126 to 134 in 7959350

	if len(query_matches) > min_neighbors:
	motif_distances.append(query_matches[:max_matches, 0])
	motif_indices.append(query_matches[:max_matches, 1])
	if len(query_matches) == 0: # pragma: no cover
	query_matches = np.array([[np.nan, candidate_idx]])
	for idx in query_matches[:, 1]:
	core.apply_exclusion_zone(P, int(idx), excl_zone, np.inf)

Lines 130-131 were added to prevent this issue. But, this if-block is not shown in the image provided by @chvalovak.

[NimaSarajpoor]

@seanlaw
As a follow up to my previous comment, I would like to add something here:

IF len(query_matches)==0, then the for-loop in lines 133-134 has only one iteration, where idx is candidate_idx. Therefore, according to the following lines of code, candidate_idx will be updated to have the value np.argmin(P[-1]).

stumpy/stumpy/motifs.py

Lines 133 to 136 in 7959350

	for idx in query_matches[:, 1]:
	core.apply_exclusion_zone(P, int(idx), excl_zone, np.inf)
	candidate_idx = np.argmin(P[-1])

Is that meaningful in such case? We may want to investigate this.

[chvalovak]

Thank you very much for your reply. Adding lines

if len(query_matches) == 0:  # pragma: no cover 
     query_matches = np.array([[np.nan, candidate_idx]]) 

helped. Motifs matching run without error.