Week 9 (27th July to 2nd August)
I began the week by finalizing a PR that integrated benchmarks across all three libraries we had targeted so far. With toqito, qutipy, and ketjl benchmarks now running in CI, we discussed on what to do ahead.
In my original proposal, I had promised benchmarking across multiple ecosystems, and naturally, MATLAB’s QETLAB
is the next candidate. But in a quick discussion with Vincent and Purva, we also considered whether benchmarking against more general-purpose quantum frameworks like Qiskit or Pennylane could be valuable. The main concern was overlap; while Qiskit and Pennylane do provide primitives like partial_trace and partial_transpose, their scope is broader and less directly comparable to toqito.
Another thread this week was improving CI results reporting. Purva pointed out that while the benchmarks were running
, the outputs weren’t visible in a clear table format, unlike toqito’s own coverage reports. This led me to experiment with ways to generate pytest-style result tables for benchmarks, at least for Python-based runs. I confirmed this is feasible for toqito and qutipy, though it’s less straightforward for ketjl and QETLAB.
Progress:
- PR #19: Add Bench 1
- Started investigating results table generation for CI output (similar to coverage reports).
-
Started planning next benchmarking targets:
QETLAB(MATLAB) — likely next since it has direct overlap with toqito.Pennylane/Qiskit— under consideration, though functional overlap may be shallow.