Tonight I successfully rebuilt and significantly upgraded my quantum + physics simulation environment on AWS. This is a detailed technical reflection of everything accomplished.
1. Infrastructure on AWS
- Launched Amazon Linux 2023 EC2 instance
- Started with t3.micro, upgraded to t3.large (8 GB RAM) to support Genesis World
- Resized EBS volume from 8 GB → 100 GB and properly extended the XFS filesystem using growpart + xfs_growfs
- Created persistent virtual environment (ionq-env)
2. IonQ Quantum Cloud Integration
- Created and verified multiple API keys
- Installed qiskit + qiskit-ionq provider
- Successfully ran Bell state circuits on the IonQ simulator
- Confirmed access to both ionq_simulator and ionq_qpu backends
- Job History: Two simulator jobs (circuit-172 and circuit-59) completed successfully with $0.00 cost (visible in IonQ dashboard under “My Jobs”)
3. Genesis World (v0.2.1) Multi-Physics Engine
Major technical hurdles resolved:
- Installed heavy stack: Taichi 1.7.3, MuJoCo 3.2.5, PyTorch (CPU), libigl, PyVista, OpenEXR, etc.
- Fixed numerous system library dependencies (libX11, Mesa, gcc-c++, etc.)
- Resolved deep NumPy version conflicts (downgraded to 1.26.4)
- Overcame disk space exhaustion and multiple wheel compilation issues (tetgen, etc.)
- Successfully initialized Genesis with CPU backend in headless mode
Current Status:
- Genesis initializes cleanly (gs.init(backend=gs.cpu))
- Scene creation works
- Minor remaining SDF computation error on some primitives (common compatibility issue)
- Ready for rigid body, soft body, and multi-physics simulations
4. Overall Architecture
- Quantum Layer: IonQ (simulator + real hardware ready)
- Physics Layer: Genesis World (rigid, FEM, MPM, particles, soft bodies, etc.)
- Compute: Scalable AWS EC2 with 100 GB storage, headless-ready
- Environment: Persistent venv, suitable for batch, multi-machine, and hybrid experiments
This setup gives me a strong foundation for:
- Quantum-enhanced optimization for robotics
- Embodied AI and high-fidelity physics simulation
- Hybrid quantum-classical workflows
- Cloud-based experimentation and scaling
Compared to when I did something similar with ChatGPT over a year ago, this stack is far more powerful and production-oriented.