Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

One possible idea to speed up the C attribute of Circuit class #969

Closed
Asachoo opened this issue Oct 30, 2023 · 3 comments · Fixed by #1013
Closed

One possible idea to speed up the C attribute of Circuit class #969

Asachoo opened this issue Oct 30, 2023 · 3 comments · Fixed by #1013
Labels
Circuit Feature Request Wished Feature Improvements Improvements of existing feature

Comments

@Asachoo
Copy link
Contributor

Asachoo commented Oct 30, 2023

Although the performance of circuit computing S-parameters has been significantly improved, I still want to try whether it is possible to further enhance the calculation performance.
Related to #574 and #575
I noticed that the performance of the X property of Circuit class has been improved, so focused on the C property of Circuit class.

In the current circuit.py, there is a section of for loop code within the C property. Due to the use of the product function, the loop is executed a large number of times. Therefore, I am wondering if it is possible to accelerate it using numpy's broadcasting feature.

Existing code:

for (ntw_name, ntw_ports) in ntws_ports_reordering.items():
    # get the port re-ordering indexes (from -> to)
    ntw_ports = np.array(ntw_ports)
    # create the port permutations
    from_port = list(product(ntw_ports[:,0], repeat=2))
    to_port = list(product(ntw_ports[:,1], repeat=2))

    #print(ntw_name, from_port, to_port)
    for (_from, _to) in zip(from_port, to_port):
        #print(f'{_from} --> {_to}')
        S[:, _to[0], _to[1]] = ntws[ntw_name].s_traveling[:, _from[0], _from[1]]

return S  # shape (nb_frequency, nb_inter*nb_n, nb_inter*nb_n)

Improved code:

for (ntw_name, ntw_ports) in ntws_ports_reordering.items():
    # get the port re-ordering indexes (from -> to)
    ntw_ports = np.array(ntw_ports)
    # create the port permutations
    from_port = ntw_ports[:,0]
    to_port = ntw_ports[:,1]

    #print(ntw_name, from_port, to_port)
    for (_from, _to) in zip(from_port, to_port):
        #print(f'{_from} --> {_to}')
        S[:, _to, to_port] = ntws[ntw_name].s_traveling[:, _from, from_port]

return S  # shape (nb_frequency, nb_inter*nb_n, nb_inter*nb_n)

However, it was found that only when the network in the circuit is connected internally, there is an efficiency improvement of about 10%. Therefore, I would like to know if this acceleration method is truly effective and if it may result in unexpected errors.
@Asachoo Asachoo added the Feature Request Wished Feature label Oct 30, 2023
@jhillairet
Copy link
Member

I don't remember why the product was used. Are the tests passing with your modification?

@Asachoo
Copy link
Contributor Author

Asachoo commented Nov 6, 2023

I don't remember why the product was used. Are the tests passing with your modification?

pytest test_network.py

I'm not sure if my testing method is correct, but when I executed the code above, this was the result I obtained (I believe all tests have passed).

=========================================================================================================== test session starts ============================================================================================================
platform linux -- Python 3.8.16, pytest-7.1.2, pluggy-1.0.0
rootdir: /home/lansus/WorkFile/Git/scikit-rf, configfile: pyproject.toml
plugins: cov-4.1.0, anyio-3.6.1
collected 107 items                                                                                                                                                                                                                        

test_network.py .......s...................................................................................................                                                                                                          [100%]

---------- coverage: platform linux, python 3.8.16-final-0 -----------
Name                                                                                                 Stmts   Miss  Cover
------------------------------------------------------------------------------------------------------------------------
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/__init__.py                        67      9    87%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/calibration/__init__.py             6      0   100%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/calibration/calibration.py       2680   2413    10%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/calibration/calibrationSet.py      52     36    31%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/calibration/deembedding.py        851    777     9%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/circuit.py                        344    262    24%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/constants.py                       25      2    92%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/data/__init__.py                   64     14    78%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/frequency.py                      223     69    69%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/instances.py                      235     76    68%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/__init__.py                      5      0   100%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/citi.py                        119    100    16%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/csv.py                         326    288    12%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/general.py                     246    198    20%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/mdif.py                        194    169    13%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/metas.py                        23     23     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/io/touchstone.py                  283    103    64%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/mathFunctions.py                  211     71    66%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/__init__.py                   9      0   100%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/circularWaveguide.py         97     54    44%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/coaxial.py                   92     61    34%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/cpw.py                      124     39    69%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/definedAEpTandZ0.py          63     36    43%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/device.py                    98     98     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/distributedCircuit.py        53     29    45%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/freespace.py                 75     25    67%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/media.py                    395    205    48%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/mline.py                    223    195    13%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/media/rectangularWaveguide.py     104     56    46%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/network2.py                       462    462     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/network.py                       1978    505    74%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/networkSet.py                     379    225    41%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/notebook/__init__.py                0      0   100%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/notebook/bokeh_.py                 46     46     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/notebook/matplotlib_.py             0      0   100%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/notebook/utils.py                  18     18     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/plotting.py                       598    419    30%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/qfactor.py                        502    468     7%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/taper.py                           71     43    39%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/time.py                           128     43    66%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/tlineFunctions.py                  90     56    38%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/util.py                           257    147    43%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vectorFitting.py                  734    692     6%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/__init__.py                      1      1     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/scpi_errors.py                   8      8     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/validators.py                  153    153     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/__init__.py                  2      2     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/hp8510c_sweep_plan.py      127    127     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/hp/__init__.py               1      1     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/hp/hp8510c.py               85     85     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/keysight/__init__.py         2      2     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/keysight/fieldfox.py       122    122     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/keysight/pna.py            257    257     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/nanovna/__init__.py          1      1     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/nanovna/nanovna.py         148    148     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/nanovna_v2.py              121    121     0%
/home/lansus/Applications/anaconda3/lib/python3.8/site-packages/skrf/vi/vna/vna.py                     180    180     0%
------------------------------------------------------------------------------------------------------------------------
TOTAL                                                                                                13758   9740    29%


====================================================================================================== 106 passed, 1 skipped in 4.36s ======================================================================================================

@zolabar zolabar mentioned this issue Jan 25, 2024
Open
jhillairet added a commit to jhillairet/scikit-rf that referenced this issue Jan 27, 2024
@jhillairet
Improve Circuit Performances
a95eeb2 
Following suggested changes in scikit-rf#969
@jhillairet jhillairet mentioned this issue Jan 27, 2024
Merged
@jhillairet
Copy link
Member

Dear @Asachoo ,

Sorry for the delay! Your suggestion is perfect. I've implemented it in PR #1013. Many thanks!

@jhillairet jhillairet added Improvements Improvements of existing feature Circuit labels Jan 27, 2024
@jhillairet jhillairet linked a pull request Jan 27, 2024 that will close this issue
Optimize Circuit C function #1013
Merged
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
Circuit Feature Request Wished Feature Improvements Improvements of existing feature
Projects
None yet
Milestone
No milestone
Development

Successfully merging a pull request may close this issue.

Optimize Circuit C function jhillairet/scikit-rf
2 participants
@jhillairet @Asachoo