Closed pandasa123 closed 7 months ago
Proofs are a bit wider and more generalised...maybe we can look towards "given two states, design a procedure to connect" in code instead?
This repository is no longer maintained. It has been superseded by IBM Quantum Learning.
Background
Professors test students with proof questions. This typically involve:
Where X and Y are usually pseudo-code...probably closer to LaTeX. Here's an example for Turing Decidability in classical computing theory (where decidable is a language that either accepts or halts without loops):
State if the following language is decidable or undecidable
$L_{HALT \geq 10} = \{ (\langle M \rangle, x) \}$: M halts on x after running for at least 10 steps
Where the proof is as follows:
This language is undecidable. We can show this with a reduction from the Halting Problem with a language as follows:
Given a blackbox decider $B{HALT \geq 10}$ for $L{HALT \geq 10}$, we construct the following for decider $H{HALT}$ for $L{HALT}$
Analysis:
Therefore, $H$ is a decider for $L{HALT}$. This shows that $L{HALT}$ is a reduction of $L{HALT \geq 10}$. Since $L{HALT}$ is undecidable, so is $L_{HALT \geq 10}$
Goal
We should be able to support auto-grading these problems for Professors
Possible idea:
Acceptance criteria
Out of scope
Full implementation with server isn't needed yet, same with professor definition. We just need to show one provocative example, could be a design to test
Resources
No response
Tasks
No response