[bug] Integer overflow when the photometry data is too long

[teristam]

Hello,

Recently I ran into a bug importing data from pyPhotometry. The default data type of np.arange is np.int32, which will lead to an integer overflow when the recoded length is longer than 35 min at 1000Hz from the following line:

code/tools/data_import.py

Line 48 in 2e95ee7

time = np.arange(analog_1.shape[0])*1000/sampling_rate # Time relative to start of recording (ms).

[ThomasAkam]

Hi teristam,

What error message do you get when you try to import the data? I don't think the bug is due to np.arange not being able to handle large enough numbers, as a 35 minute recording at 1000Hz is 2100000 samples but e.g. np.arange(100000000) does not give an error on my machine.

Thomas

[teristam]

Hi Thomas,

But the import function also convert everything in ms by multiplying 1000. The time array gives me a negative value. The following:

np.arange(1000*36*60)*1000

give me

array([ 0, 1000, 2000, ..., -2134970296, -2134969296, -2134968296]))

[ThomasAkam]

That makes sense, can you try modifying line 48 in data_import.py to read:

time = np.arange(analog_1.shape[0], dtype=np.int64)*1000/sampling_rate # Time relative to start of recording (ms). 

and let me know if this fixes the problem?