vyrjana / pyimpspec Goto Github PK

Dear vyrjana,
I'm trying to accelerated processing the circuit fit when treat batch process, I adopted multiprocess by using 6 num_workers to process a dataset with 20 frequencies and the batch size is 20，the time cost of common loop is 69.77 seconds, and the multiprocess is 22.28 seconds， but it is too long for me?
Do you have some tips for multiprocess for speeding up ?
Thanks a lot!

Specifications

• pyimpspec version: 4.1.1
• Python version: 3.11.8
• OS: Windows

Description:
DRT calculations fail at certain lambda values producing the following error:
new = pyimpspec.calculate_drt(data0, method='tr-rbf',lambda_value=lam)
DRTError: Failed to perform calculations! Try tweaking the settings.

This occurs at a lambda value of ~0.0001.

This error only occurs with cvxpy (with the windows specific instructions followed)

With both cvxopt and kvxopt, erroneous DRT distributions are produced with negative values and extra oscillations.

Expected behavior
The system is a simple randle cell with a known characteristic time at 10^-1. pyDRTtools produces the correct results at various lambda values.

This issue is for keeping track of the status of support for Python 3.11. The test-python-3-11 branch contains modified workflow files and at the moment the build workflow fails. The reason is that cvxopt does not have a wheel for Python 3.11 and pip is unsuccessful when it tries to compile the package. An issue has been opened in the cvxopt repository regarding the availability of Python 3.11 wheels.

Hi, I'm trying to use the function 'dataframe_to_data_sets', trying to convert a dataframe format into dataset to make the circuit fit. I've read the API, which need to input the path where the dataframe located，Is anyway to convert the dataframe into dataset directly?
Thanks!

Desktop (please complete the following information):

• pyimpspec version: 4.1.0
• Python version: 3.11.4
• OS: Windows

Describe the bug
current implementation of pyimpspec.analysis.utility.residuals:

def _calculate_residuals(
    Z_exp: ComplexImpedances,
    Z_fit: ComplexImpedances,
) -> ComplexResiduals:
    residuals: ComplexResiduals = empty(Z_exp.shape, dtype=ComplexResidual)
    residuals.real = (Z_exp.real - Z_fit.real) / abs(Z_exp)
    residuals.imag = (Z_exp.imag - Z_fit.imag) / abs(Z_exp)
    return residuals

according to equation [1] in B.A. Boukamp, 1995, J. Electrochem. the denominator is

the vector length (absolute value) of the modeling function

so the equations should be:

residuals.real = (Z_exp.real - Z_fit.real) / abs(Z_fit)
residuals.imag = (Z_exp.imag - Z_fit.imag) / abs(Z_fit)