Explore Prompts

Create an error budget calculator that allocates total error across depth, readout, and optimizer noise. Provide equations and a worked example. End with a rule-of-thumb table by device size.

Define a benchmarking suite for variational algorithms: datasets, seed control, metrics (energy error, convergence time), and reproducibility notes. Output: CSV spec (metrics) and YAML (suite config).

Describe how to encode a directed graph into a quantum walk operator. Include oracle construction ideas, coin vs shift operators, and observables to measure. Output a step list and pitfalls to avoid.

Explain parameter-shift gradients for variational circuits with a simple example. Show the numeric finite-difference alternative and compare variance/cost. Provide a tiny test harness.

Propose how to split compute between CPU, GPU, and QPU for a variational algorithm. Provide a pipeline diagram description, and give heuristics for batching, latency hiding, and caching parameter shifts.

Design a quantum kernel or feature map for a binary classifier. Include encoding scheme, circuit depth estimation, and classical SVM pairing. Return circuit description in a textual gate list and a short comparison to a classical RBF kernel.

Create a plan to characterize noise on a NISQ device: T1/T2, readout error, crosstalk notes. Recommend mitigation: measurement error mitigation, dynamical decoupling, zero-noise extrapolation. Output as YAML plus a small experiment schedule table.

Outline a VQE workflow for estimating ground state energy of a small molecule. Specify ansatz choices (UCCSD vs hardware-efficient), mapping (Jordan–Wigner/Bravyi–Kitaev), optimizer options, and stopping criteria. Provide a minimal code sketch and a CSV of hyperparameters.

Given an unweighted graph (edge list), generate steps to formulate a Max-Cut QAOA instance: cost & mixer Hamiltonians, initial state, parameterization, and classical optimizer loop. Output Python-like pseudocode and a JSON block describing inputs/outputs. Include a note on noise-aware depth.

Compress a long prompt by >40% while preserving constraints and output format. Show a diff (removed/kept/condensed) and a final compressed version. Return as Markdown with a 3-row risk table.

Create a rubric for humans to score the model’s outputs: 5 criteria, 5-point scale each, behavioral anchors per score. Provide a one-page training note for raters and inter-rater reliability tips.

Translate the prompt’s requirements into a lightweight program skeleton with function signatures, I/O types, and edge cases. Provide implementations for input parsing and output validation. Language-agnostic pseudocode.