Execution Graph and Computation Optimization

[yuefan98]

Execution Graph

• Added support to graph-based function evaluation (Thanks @andersonjacob for the source code)
• Added more tests to cover the changes
• Added speed benchmark for the graph on documentation

Computation Optimization

• Vectorized calculation in circuit elements for fast computation

New Circuit Elements

• Added support to nonlinear RC with constant phase element: (RCDQ, RCDQn), and (RCSQ, RCSQn)
• Added more tests to cover the changes

Documentation

• Added Release Notes

General Code Improvement

• Minor code improvement for better handling
• Bumped version to v0.2
• Updated requirement.txt and setup.py

[coveralls]

Pull Request Test Coverage Report for Build 12875811319

Details

• 474 of 483 (98.14%) changed or added relevant lines in 5 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.4%) to 96.404%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
nleis/nleis_elements_pair.py	339	348	97.41%

Totals
Change from base Build 12246244250:	0.4%
Covered Lines:	1555
Relevant Lines:	1613

---

💛 - Coveralls

[yuefan98]

Hi @andersonjacob,

Thanks for your excellent code on CircuitGraph. I made small edits to it and it worked seamlessly with my code. With some benchmark results, the graph-based evaluation can at least 3X speed up the function evaluation. Surprisingly it does not introduce much performance improvement when the circuit is matrix-driven and the eval() method can be faster sometimes (not explicitly shown in the benchmark results). Without too much change to the main code, I also made graph-based evaluation an option to the eval() method rather than directly substituting it. With said said, I would like to invite you to review the changes I made so that I can address potential issues that I haven't considered yet.

Main changes to the CircuitGraph function include:

• Supports the parsing of difference element (d)
• Added __eq__ to enable the comparison of the two graphs through node and edge comparison.
• I also realized that I have make __call__ equivalent to compute_long. And their implementations are kind of duplicated. I might probably make it to return impedance directly, but for now, they work just fine.

[andersonjacob]

The if statements are really only needed to get the number to use. You could simplify it a lot with

if operator == "p":
  nnum = self.pnum
  self.pnum += 1
elif operator == "d":
  nnum = self.dnum
  self.dnum += 1

node = f"{operator}{nnum}"
self.graph.add_node(node, Z=circuit_elements[operator])
for elem in pd_elem:
  elem = self.add_series_elements(elem)
  self.graph.add_edge(elem, node)
parsing_circuit = parsing_circuit.replace(pd, node)

[yuefan98]

Thanks for spending time reviewing my implementation. I agree that this will make the code cleaner and simpler. I will make the changes accordingly.

[andersonjacob]

One thing that may make things easier/better for the long term, but not for the short term. Your code depends on impedance.py so if the graph where merged there then you could inherit from it and just override what you needed and keep the rest the same. Then you don't have to implement things twice.

[yuefan98]

One thing that may make things easier/better for the long term, but not for the short term. Your code depends on impedance.py so if the graph were merged there then you could inherit from it and just override what you needed and keep the rest the same. Then you don't have to implement things twice.

Thanks for the note! That is exactly my long-term goal for the nleis.py project. Once the nleis.py has been stably running for a well, I will make sure it inherits impedance.py as much as possible so I can either integrate it to impedance.py or make it as a simple add-on to impedance.py

[andersonjacob]

The code itself looks very similar to what I would expect. I haven't run anything so I can't verify that it works as one would expect, but I am assuming that your tests will cover most all of that.

I think it looks good.

[yuefan98]

Yes, the tests cover most of the essential use cases and make sure the graph-based calculation is equivalent to the eval()-based calculation. With all the tests passed, the code should work as expected.

Thank you so much for your contribution and I will merge the PR soon. 🚀

[yuefan98]

@all-contributors please add @andersonjacob for code and review.

[allcontributors]

@yuefan98

I've put up a pull request to add @andersonjacob! 🎉