Batch first simulator academy

Batch-first simulator academy

Scientific preparation is cheap; expensive execution is scheduled.

Open the eight introductory notebooks, starting with credential-free Aer CPU and moving to a separately authorised, one-GPU Slurm job. The workbench remains a small JupyterHub/KubeSpawner CPU pod; its lifetime does not own a Slurm GPU allocation.

The official upstream SDKs are taught natively: Qiskit/Aer, PennyLane/Lightning, Cirq/qsim/cuQuantum, IQM fake backends, Pasqal/Pulser neutral-atom emulation, and approved physical provider SDKs. Frameworks, simulator backends, execution resources and QPUs are distinct. No default QPU credits or GPU interactive session.

• Read M5 batch-first design and acceptance.
• Read scientific environments.
• Read Slurm runner contract.
• Status: notebook/runner source only until native researcher Slurm access, GPU GRES, image build and site tests are independently observed.