57 lines
2.5 KiB
Markdown
57 lines
2.5 KiB
Markdown
# After recovery period list
|
|
|
|
Hello me, this is a list for you to get back on track with the QD-QAS you were working on...
|
|
|
|
## Things I've done
|
|
|
|
I have made baselines for two types of optimization free QAS, namely TF-QAS and GA-QAS.
|
|
These aren't necessarily fully confirmed to be functional yet,
|
|
but seem to be doing mostly the correct things. Although the TF-QAS doesn't give me the same
|
|
results as the version for the paper.
|
|
This might still need to be fixed, but could also not matter that much. // Discuss with sebastian
|
|
|
|
## What to do next
|
|
|
|
Make meetings with Sebastian and Koen!!!!!
|
|
|
|
1. Have a system to generate and work with quantum circuits
|
|
2. Create some metrics to use for the QAS
|
|
- expressivity: how uniformly does it cover the hilbert space
|
|
- entanglement: how many correlations exist between the qubits
|
|
- noise: what is the fidelity of the final circuit
|
|
- parameter count: how many parameters are in the circuit
|
|
3. Make a genetic algorithm using (some of) these metrics
|
|
4. Create a metric to estimate the "distance" between two PQCs
|
|
5. Use this distance metric to make Quality Diversity evolutionary algorithm instead
|
|
6. Compare with the baselines and normal GA with the metrics
|
|
|
|
## What is knowledge you might want to have
|
|
|
|
Expressivity is a double edged sword, too little and you won't get the result, too much
|
|
and barren plateaus will make the optimizer suffer.
|
|
|
|
Entanglement works well if the underlying problem is entangled, but can also decrease the expressivity.
|
|
|
|
Noise is just straight up bad.
|
|
|
|
More parameters is also kinda not what you want.
|
|
|
|
## A letter to me
|
|
|
|
You made a promise to Nynke that you would tell her if there were problems.
|
|
So please do that, try not to wait until they're about to blow up, as that won't help.
|
|
Sebastian and Koen also want you to succeed so if they're harsh that isn't a personal attack
|
|
as they're just trying to help you.
|
|
In general this will mean you waited too long to ask questions to them, not that you asked too many.
|
|
|
|
And above all, remember that you are enough already. You don't need to do a PhD worth of research in
|
|
6 months. That'd be insane. YOU CAN DO THIS!!!
|
|
|
|
Peppi believes in you, Nynke believes in you, your parents believe in you, I (you) believe in you.
|
|
You can do this, and if you think you can't I will bet on it.
|
|
If it's too much torture, and you barely can get yourself to work on it. Take a step back, ask people
|
|
and don't feel too bad about it. After all, it IS hard, you don't have to be able to do it all yet.
|
|
This is what learning is all about.
|
|
|
|
Yours truly,
|
|
Noa
|