Issues with HHL algorithm implementation

Questions
,
1

Hi,

I am facing some issues with HHL implementation here: https://resources.wolframcloud.com/ExampleRepository/resources/HarrowHassidimLloyd-HHL-algorithm/

Whenever I alter the matrix (maintaining hermiticity!), the results start to differ (right after one decimal point) between the simulated quantum and classical.

For example, if I take p*A, where p=1.25, it did not produce the expected outcome.

The same issue persisted when I picked A using Abs[SymmetricMatrix[RandomVariate[GaussianOrthogonalMatrixDistribution]]]

Later, I increased the matrix dimension to 8 (in addition to changing the precision accordingly).

I would like to know if anyone else has faced similar issues and if anyone has any remarks.

Thank you in advance.
Arun

2

Thanks for your feedback. We will respond soon with an updated example.

3

Here is a clean HHL that works for any Hermitian matrix and any dimension. Please note that the precision of HHL depends on the number of phase-estimation qubits (3rd argument). Our algorithm does not need rescaling etc that imposes some conditions (e.g. eigenvalues from 0 to 1, only). Let’s know if any question or issues.

Open in Wolfram Cloud ↗