Blog

The Importance of Quantum Portability

23
June
,
2021
Yuval Boger

Remember the time you had to commit to a game console? Will it be an XBOX? A PlayStation? A Nintendo? You knew that whatever you chose, you will be locked in. XBOX games don’t work on a PlayStation and vice versa. Once you made the choice, every game you purchased made it more difficult to switch.

Then again, it was not the end of the world. Most games are available on most platforms, and you were not too worried that SONY will discontinue the PlayStation or that another vendor will drop out.

But that’s not the case with quantum computers these days. Which vendor will win and who will drop out? If my team spends months upon months creating quantum software on one computer, what would happen if I need to move it to another vendor’s machine? For some companies, this is a major concern.

After all, different quantum computers have different architectures: different connectivity schemes, different gate fidelities that might prompt you to prioritize different gate sets, different capabilities. Some have mid-circuit measurements, some don’t. How much effort will it take to rewrite the code in case we need to switch?

This reminds me of an old proverb: a clever person knows how to wiggle out of problems that a smart person would not have gotten into in the first place. What’s the solution to porting code that’s written at the gate level? It is not to code it at the gate level in the first place. 

Instead of coding at the gate level, consider creating a model of how you want the algorithm to behave, and then using a software platform to synthesize this model into gate-level code, taking into account the various hardware constraints. As long as you remain at the high-level modeling level, it is easy to port algorithms from one machine to another. Need to avoid certain types of gates because using them is not advised in the new hardware? Add these gates as a constraint. Want to trade-off circuit depth for a few more ancilla qubits? The right platform will make this easy for you, allowing you to enter quantum computing with confidence without having to commit to a particular vendor.


Remember the time you had to commit to a game console? Will it be an XBOX? A PlayStation? A Nintendo? You knew that whatever you chose, you will be locked in. XBOX games don’t work on a PlayStation and vice versa. Once you made the choice, every game you purchased made it more difficult to switch.

Then again, it was not the end of the world. Most games are available on most platforms, and you were not too worried that SONY will discontinue the PlayStation or that another vendor will drop out.

But that’s not the case with quantum computers these days. Which vendor will win and who will drop out? If my team spends months upon months creating quantum software on one computer, what would happen if I need to move it to another vendor’s machine? For some companies, this is a major concern.

After all, different quantum computers have different architectures: different connectivity schemes, different gate fidelities that might prompt you to prioritize different gate sets, different capabilities. Some have mid-circuit measurements, some don’t. How much effort will it take to rewrite the code in case we need to switch?

This reminds me of an old proverb: a clever person knows how to wiggle out of problems that a smart person would not have gotten into in the first place. What’s the solution to porting code that’s written at the gate level? It is not to code it at the gate level in the first place. 

Instead of coding at the gate level, consider creating a model of how you want the algorithm to behave, and then using a software platform to synthesize this model into gate-level code, taking into account the various hardware constraints. As long as you remain at the high-level modeling level, it is easy to port algorithms from one machine to another. Need to avoid certain types of gates because using them is not advised in the new hardware? Add these gates as a constraint. Want to trade-off circuit depth for a few more ancilla qubits? The right platform will make this easy for you, allowing you to enter quantum computing with confidence without having to commit to a particular vendor.


See Also

No items found.

Start Creating Quantum Software Without Limits

contact us