Misconceptions About Quantum Programming

Qubit vs Classical Bits. Given the recent hype around Quantum Computing, I have read in several places the sentence: “The power of quantum hides in the fact that N qubits can store the information equivalent content of 2^N classical bits.” ❌ This statement is false and is a common misconception! ❌ I think it is useful to clarify this once and for all. 👉 An N-qubit quantum system can exist in a superposition of 2^N basis states (see picture for N=2). However, according to Holevo’s Theorem, although N-qubits can carry a larger amount of classical information, the real amount that can be extracted/accessed from such a system is limited to N bits upon measurement (due to the collapse of the quantum superposition). This means you cannot store 2^N classical bits of information in an N-qubit system.  So what gives to QC a potential computation advantage? The exponential growth (2^N) in the number of superposed basis states allows for the exploration of many possibilities simultaneously, a phenomenon known as quantum parallelism. * Additional info: The misconception probably arises when one wants to answer the question: how many bits do I need to simulate an N-qubit quantum state on a classical computer? The answer is 2^N (or in that order). Representing an N-qubit quantum system on a classical computer requires writing in memory the system state vector, which encompasses 2^N complex numbers. Considering that each complex number typically consists of two real numbers (real and imaginary parts), this asks for 16×2^N bytes of classical information (two 8 bytes double-precision floating-point numbers). This number can grow exponentially fast, and therefore we explained why is so hard to simulate already simple quantum systems on a classical hardware. For more details: https://lnkd.in/dRYifpVU #quantum #quantumtech #emergingtech #ai

When you read that quantum computers can solve problems that classical computers could not solve, one thing is often missing: "in a reasonable time frame". Indeed, as long as we're (1) talking about classical data and (2) not looking at computation time, classical and quantum computers are strictly equivalent. As a proof, just consider the fact that quantum circuits can be run on classical computers, using emulators. Of course, this only works up to a certain number of qubits, beyond which computation time and/or memory usage become unreasonable. But this is a practical problem, not a conceptual problem. If you could get an arbitrarily powerful classical machine, you could run an arbitrarily large quantum circuit without ever needing a real quantum computer. As for the reverse, quantum computers can run any classical circuit if you just use qubits as classical bits. This would be utterly useless, because gates are muuuch slower on qubits than on classical bits, but again, it's a practical and not a conceptual problem. However, this equivalence only holds for classical data. When you're dealing with quantum data (for example, when using a quantum sensor), three things stand in the way of this equivalence: 👉 If you don't have a quantum computer, you must perform a tomography of your quantum data, meaning taking a lot of measurements. But since each measurement destroys your quantum data, it must be possible to recreate your quantum data at will. If this is not possible, then you just cannot properly process your quantum data with a classical computer. 👉 Even if you can recreate your quantum data at will, the precision of the state you recreated will be limited by the number of measurements you made. This precision can be made arbitrarily high, but this comes at the cost of an arbitrarily high number of measurements. 👉 Even if you could retrieve the exact quantum state, classical computers only compute with finite precision. Again, you can make the precision arbitrarily high, but this comes at the cost of an arbitrarily long processing time. About the last two points, to be fair, you also need to take precision into account when using a real quantum computer. 👉 If you don't use error correction, noise limits precision. 👉 If you use error correction, continuous gates such as rotations are decomposed into discrete gates. This also limits precision, but you can make precision arbitrarily high by performing an arbitrarily high number of gates. Quantum data isn't that much of a topic for quantum computers today, because loading quantum data from a quantum sensor into a quantum computer is still mostly an open problem. But it might be an interesting application of quantum computers in a few years from now. Finally, all of the above only holds for gate-based quantum computers. Analog quantum computers are a different business, but since I don't know them well enough, I'll let those who do talk about them.

What Quantum Computers CANNOT do? In recent years, the hype regarding quantum computers has been quite big. Quantum computers are supposedly going to magic boxes that can do all our tasks faster and better and decrypt everything, including all the encryption that already exists now. While quantum computers have immense potential for certain problems, there are several things that quantum computers cannot do. I will mention some of those things here so we can correctly appreciate the potential of quantum computers. 1. Quantum computers will not replace classical computers The goal behind building Quantum computers is not to replace classical computers. Instead, to make a device that can solve certain specific problems that take decades and even centuries for classical computers to do. For example, a quantum computer can find prime factors for large numbers and simulate quantum systems, but it is not practical to use it for web browsing or sending emails. For routine tasks, classical computers will be essential. 2. "Quantum computer can break all encryptions." No, it cannot Quantum computers have (theoretically) proven to be a threat to some of the present cryptographic algorithms, such as RSA. But this doesn't mean all our encryption methods are useless. In fact, scientists are developing post-quantum cryptography, which will be secure against quantum attacks. 3. Quantum computers will NOT provide one-size-fits-all solutions Quantum computers can solve some specific problems much faster than classical computers. But for complex problems like weather forecasting or for AI applications, quantum computers do not provide speed advantages. For such problems, we need to customize quantum computers to gain quantum advantages. 4. Quantum Computer does NOT provide error-free responses Qubits are extremely sensitive to external interferences. Up on interacting with the environment, it loses its quantum superpower (quantum superposition). This makes quantum computers prone to errors. Scientists are employing error correction techniques, but so far, we have not achieved error-free large-scale quantum computing. Real-world applications need quantum speed as well as quantum accuracy. Both of which are a significant challenge.  A functional Quantum computer will revolutionize some of the specific industries but will not solve all the problems. We need to understand what quantum computers CANNOT do and set realistic and clearer expectations accordingly. Stay curious, keep questioning, and set realistic expectations for quantum computers! #QuantumComputing #TechnologyLimits #ResearchLife #FutureTech