Skip to content

Latest commit

 

History

History
66 lines (51 loc) · 4.61 KB

README.md

File metadata and controls

66 lines (51 loc) · 4.61 KB
Raw

Sample code from Dave Wecker

Examples:

  • AltOutput.fs: QASM output example
  • QLSA.fsx: Source code for the LIQUi|〉 __QLSA() sample
  • QuAM.fsx : Source code for the LIQUi|〉 __QuAM() sample

AltOutput.fs

I've had some requests on how to generate circuit output for other quantum lanaguages (like QASM). This file contains a complete example. If you add it to your project (ahead of Main.fs) you will have a new command line function __DumpQASM() that will output QASM source for a Teleport circuit and a QFT.

The sections of the sample are:

  • Sample circuit definitions for Teleport and QFT
  • DumpQASM: Main routine for the example
  • appWs: append wire names to the output for a gate
  • mapName: - placeholder to remap LIQUi|〉 names to QASM names as desired
  • out: dump out the current built line and clear it
  • outGate: call the above routines to output a complete gate. The only non-obvious piece is the pfx argument that will add a c- to the beginning of the gate name for binary controlled gates
  • dump: main recursive loop that outputs the circuit. The parse is pretty straight-forward with the exception of:
    • Label gates are skipped (g.Op = GateOp.String). You could gather all these up and then output then to QASM in a seperate pass. Not needed for this simple example.
    • BitCon: Binary controlled gates need to be parsed to find what they control so that we can pass the correct name to QASM. This is only a partial parse (it's possible to have cases that this code doesn't handle)... but in most cases this will be enough
  • At the bottom of dump, we count up the total number of qubits (wires) and use that to make a qubit list for QASM.

QLSA.fsx

There's some interest in using the Quantum Linear Systems Algorithm (QLSA) code in a course and we'e been asked to release the source code for this purpose. The routine is an implementation of Harrow, Hassidim and Lloyd (HHL). I can't release the exact code (since it contains implementations of unpublished research), but I've written a "stripped down" version that does everything the example in LIQUi|〉 (__QLSA()) does.

Items to note:

  • Change the line #r @"\Liquid\bin\Liquid1.dll" to point to wherever your LIQUi|〉 install is so that IntelliSense in Visual Studio works (if you care).
  • QLSA() is the main routine for the example. You can invoke it with:
    • \Liquid\bin\Liquid.exe /s QLSA.fsx QLSA()
  • QLSA.log shows a sample run (copy of what Liquid.log will look like when you run)
  • After you run, you will have circuit drawings in QLSA*.htm (as well as `QLSA*.tex).
  • You will also have Redund.htm (and Redund.tex) which is the detailed circuit with obvious redundancies removed.
  • The output contains lines that have the text CSV on them. If you gather these up and put them in a .CSV file, you can use Excel to analyze the data.
  • QLSA.xlsx is a sample Excel file created from QLSA.log
    • Top left is the data exctracted from the log
    • Top right is a Pivot Table that allows manipulation of the data (you can turn the filter for Good on and off)
    • Bottom middle is a chart showing the data where we've filtered to show only converged (Good) results.
  • We try to converge at each value of r. If after 500 iterations we don't converge, we give up at that value (hence the GOOD/BAD flag on the output).

QuAM.fsx

We've had a request to make the source for the Quantum Associative Memory implementation available. It turns out that the required some work to expose a few of new APIs in the system (updates will be put in the reference manual). Functionally, the provided example is idential to the one that's built into the simulator and should give a good idea of how to implement other simliar applications.

See the paper by Ventura and Martinez for details of the approach.

You can compile and run the sample with: \Liquid\bin\Liquid.exe /s QuAM.fsx QuAM()

The new APIs are:

  • CSMat.Filled() which returns a sequence of indices of filled entries in a sparse matrix.
  • CVec.NonZeros() which returns all non-zero entries in a state vector that are above a specified threshold. This routine is already used by the Dump() function and was made public for general use.
  • Ket.Join() allows two ket vectors (registers) to be joined together. There are other ways around this, but when I re-wrote the QLSA() example I decided it would be good to expose this internal routine (since Ket.Split() is already exposed).