.svg)
The Patch Panel and the Quantum Computer
.jpeg)
If you are in Mountain View, California, and have a couple of free hours, visiting the Computer History Museum could be a good idea. Amongst the many interesting exhibits there, you might see analog computers and their patch panels.
In the 1940’s analog computers were quite popular. They were faster than their digital counterparts and were better suited to simulate physical phenomena such as hydraulics. At some level, a case could be made that to simulate and analyze analog systems, an analog computer is a natural fit. As late as the 1970’s many large organizations such as NASA, Lockheed and the French Atomic Energy Commission had analog computers in use. Even today, there are some analog computers being used for very specialized applications such as flight simulation.
An analog computer is made of operational amplifier modules as well as discrete passive components like resistors, capacitors, etc. To program an analog computer, the operator would work with a patch panel, similar to old telephone switchboards. One would plug a cable to connect the output of one module with the input of another. As analog computers became larger, their patch panels also grew, and a ‘spaghetti of cables’ ensued as you can see in the picture.
Does this patch panel and way of programming remind us of anything? This is not too far from how quantum computers are programmed today. No, there is no physical patch panel, but the quantum software engineer basically explicitly connects qubits and gates.
Do we want to continue programming quantum machines using virtual patch panels? Probably not. Will programs become increasingly difficult to create, debug and maintain with the increased number of qubits? Absolutely.
It’s not that qubits and gates won’t be connected individually in the future. It’s just that we would not expect the software engineer to explicitly specify how that happens. Just like a digital electronic circuit still has interconnected gates, so would a quantum circuit. But just like digital circuits are synthesized from VHDL or Verilog high-level models, we would expect that quantum circuits would also be automatically synthesized from high-level descriptions.
If you are in Mountain View, California, and have a couple of free hours, visiting the Computer History Museum could be a good idea. Amongst the many interesting exhibits there, you might see analog computers and their patch panels.
In the 1940’s analog computers were quite popular. They were faster than their digital counterparts and were better suited to simulate physical phenomena such as hydraulics. At some level, a case could be made that to simulate and analyze analog systems, an analog computer is a natural fit. As late as the 1970’s many large organizations such as NASA, Lockheed and the French Atomic Energy Commission had analog computers in use. Even today, there are some analog computers being used for very specialized applications such as flight simulation.
An analog computer is made of operational amplifier modules as well as discrete passive components like resistors, capacitors, etc. To program an analog computer, the operator would work with a patch panel, similar to old telephone switchboards. One would plug a cable to connect the output of one module with the input of another. As analog computers became larger, their patch panels also grew, and a ‘spaghetti of cables’ ensued as you can see in the picture.
Does this patch panel and way of programming remind us of anything? This is not too far from how quantum computers are programmed today. No, there is no physical patch panel, but the quantum software engineer basically explicitly connects qubits and gates.
Do we want to continue programming quantum machines using virtual patch panels? Probably not. Will programs become increasingly difficult to create, debug and maintain with the increased number of qubits? Absolutely.
It’s not that qubits and gates won’t be connected individually in the future. It’s just that we would not expect the software engineer to explicitly specify how that happens. Just like a digital electronic circuit still has interconnected gates, so would a quantum circuit. But just like digital circuits are synthesized from VHDL or Verilog high-level models, we would expect that quantum circuits would also be automatically synthesized from high-level descriptions.
