Loading classical data into a Quantum Circuit is a huge problem. Both conceptually and due to NISQ features. Especially in machine learning.
QC Ware a startup announced its Forge Data Loaders to load classical data into machine learning quantum circuits. So we are going to explain our Data Loaders in comparison to that.
Lets say you have N = 1000 numbers to load into a quantum circuit. The question is how many Qubits are you going to need in your Quantum Circuit. And how many Gates are you going to need with what circuit depth to prepare the Quantum States with the data you want to load for these Qubits.
Automatski uses 1000 Qubits for N=1000 Data. Why?
Because trying to make Super Super Dense Encoding of the data renders the quantum circuit unusable. It is like a compressed .tar.gz file which you cannot do pretty much anything with. So Automatski maintains 1 == 1 encoding. This is the most useful for Quantum Algorithms like Machine Learning instead of anything supercompressed.
Why do other companies use Super Super Dense Encoding? i.e. try to encode their N=1000 Data into 10 or 100 Qubits.
Because their NISQ Quantum Computers have 50-100 Qubits in total and they can reliably do very shallow circuit depths e.g. 50-70. Automatski’s Quantum Computers have no such restrictions. We support billion’s of qubits with infinite precision.
Automatski uses 0 Gates and 0 Circuit Depth to prepare the Loaded Data Quantum States
We do not start with |0> qubit states and use Gates to prepare the Quantum States. We simply initialize the require Qubit State directly. Thats mindboggling.
- Circuit Depth Required (For Data Loaders) = 0
- Total Gates Required = 0
Which encodings do we support?
- Basis Encoding (Single Pass over Data required)
- Amplitude Encoding (Two Passes over Data required)
Hope that helps!
See below if you want to learning more about Classical Encoding in Quantum Circuits