TCI for constructing MPS from an oracle

[ACE07-Sev]

Greetings dear Dr. Gray,

Hope you are doing well. I was wondering if the library currently supports TCI algorithms or variants that basically allow for the construction of an MPS from an oracle to tackle the memory overhead of storing dense tensor.

For instance, say we want to prepare a sine wavefunction over 30 qubits. Instead of allocating a $2^{30}$ array, we would sample from the distribution at different parts (which I believe are called pivots). This would be very helpful for constructing the MPS for larger qubit sizes when an oracle exists.

Thank you very much in advance.

[jcmgray]

No, sadly there is no cross interpolation functions at the moment. It would be a welcome feature. Also it is quite a contained black-box procedure, I'm sure one could use existing implementations elsewhere to produce the tensors and simply wrap them in a MatrixProductState, probably the easiest thing currently.

[ACE07-Sev]

Greetings dear Dr. Gray,

Hope you are doing well. Thank you very much for the response!

Understood. I implemented it that way for the moment, but would definitely need proper rewriting to be presentable in a PR. I'll keep this in mind for a feature request PR.