A tutorial on Quantum Approximate Optimization Algorithm (Oct 2020). Part 2: Hands-on

Part 2 of the tutorial on Combinatorial Optimization on Quantum Computers.

The slides and the Jupyter notebooks for the hands-on session can be downloaded here:

[markusbuchberger]

Wow guys, excellent tutorial, you started really on a low level, you touched all important points and brought a simple complete coded sample with just core fucntions of QISKIT. I really thank you!!!!

[suvarnadhiraj]

I think you missed explaining about rational behind choosing the initial points at 13:05

[amitracal]

Excellent video, any plan of making such a series on VQE ?

[linlin]

Thank you Rusian! Excellent work! I have a question regarding QAOA (in my own experience):
1. I noticed that the same QAOA problem (max-cut or tsp) can also be solved using VQE approach. The optimization is actually faster and the correct solution has a higher probability than those found in QAOA. To me, QAOA is a VQE with a Hamiltonian-driven ansatz, so it is a bit counter-intuitive to me.
2. I find that QAOA TSP is very difficult to have reasonably good solutions for more than 3~4 cities on qiskit. However, D-Wave has published results for even 8~10 cities using thousands of qubits. I understand their machinery is very different but I still wonder what can be improved. Would using a different optimizer (such as a global one) be better? I also see different initialization circuit besides Hadamard. Do you have any suggestions?

[anantagnihotri]

What was the logic behind choosing the initial point?