ingenii / byu-iqc-ingenii-clinical-trial
Publicingenii / byu-iqc-ingenii-clinical-trial
Public
aqora test or aqora upload:
solve, a placeholder for your solver.solve on the 10 trials of the challenge with the functions
in metric.py, and updates whenever you change solve.data/pbc.csv. The header of the challenge's original copy was
shifted (the column labeled Age held the prothrombin time, and so on). It is
fixed here, which changes no scores. Every scored trial samples 100 of these
patients.page_size to embed more in the static export.solve receives the normalized covariates of one trial, an array of shape
(100, 3), and returns two binary arrays group1 and group2 of length 100,
with group1[i] = 1 if patient i is in group 1 and 50 patients in each group.
The placeholder below splits the patients at random: replace it with your own
approach. moment_features(w) returns the 9 per-patient moment terms and their
weights, with , which
can help you set up a QUBO or Ising formulation.QPU submits circuits written with
qiskit, pytket or guppy to the quantum computers and emulators of aqora's
providers. In an aqora workspace it is already
authenticated as you.
See working with jobs and
reading results for what you can
do with a submitted job.
The cells below are a skeleton to build on: build_circuit returns a
placeholder circuit, to replace with the one your approach needs. Keep in mind
that a scored trial has 100 patients, more qubits than an emulator can simulate. ┌───┐ ░ ┌─┐
q_0: ┤ H ├──■───░─┤M├───
└───┘┌─┴─┐ ░ └╥┘┌─┐
q_1: ─────┤ X ├─░──╫─┤M├
└───┘ ░ ║ └╥┘
meas: 2/══════════════╩══╩═
0 1 build_circuit() as one job each time you click
it. It doesn't run when you edit the notebook
(mo.ui.run_button), so the
emulator isn't called on every change.
qpu_platform picks the platform as provider:name.
scaleway:EMU-AER-32C-256M is a public Aer emulator; see
choosing a platform for
the others.qpu_team is your team's username: jobs are attributed to it and count
against its quota
(submitting as a team).
With None, they are attributed to you.evaluate from metric.py runs solve on the 10 trials of the challenge
(100 patients each, sampled with the challenge's seed) and checks every
assignment, and score averages over them. Higher is better: a
perfect balance scores 1, and random assignments score about 0.2.