Skip to content

replace numpy datatype strings (e.g. "Float64") with types (e.g. "np.float64")

Jop Wolffs requested to merge github/fork/jdbocarsly/jdbocarsly/numpy_dtype_bugfix into develop

Created by: jdbocarsly

Tiny change to fix a bug: When attempting a CSA fit on OS X with python 3.9.1 and numpy 1.20.2, clicking the "Sim" or "Fit" buttons results in a message "Program error. Please report."

The error is:

File "ssnake/src/simFunctions.py", line 530, in zcw_angles js = np.arange(samples, dtype='Float64') / samples 
TypeError: data type 'Float64' not understood

Referring to this line: https://github.com/smeerten/ssnake/blob/9817bfcf639d1e91eb727767fe7bc9bffa75b5fe/src/simFunctions.py#L530

The issue (at least for my version of numpy) appears to be that dtype = 'Float64' is capitalized and just needs to be lowercased. Changing to dtype = 'float64' or dtype = np.float64 fixed this issue on my machine and Sim and Fit now work.

Merge request reports