Homework 2

1. (20 points) In HW1, we had a solution to the "imply" in a 4 Qbit embedding that we tested in the spreadsheet. Notice that if the resulting embedding does not match the bipartite graph we need, you can embed q1+q3 and q2+q4 to get a symmetric solution, which is a match for the graph we want. Let's now transform this into Quantum machine instructions (QMI).

   But first, let's run some QMI code:

       cd qOp
       cd examples
       cd dw
       bash A-execute-qmi.bash
       

   Copy the file A-execute-qmi.bash to imply4-qmi.bash, then create the same embedding as in the spreadsheet for the "imply equation".
   Hints:
   • Check out embedding (slide 16), which shows the Qbit numbering in the Chimera graph. The coupler (strength) indexing counts down top-to-bottom, i.e., you get the following connections:
     • Q0000 -- C0000 -- Q0004
     • Q0000 -- C0001 -- Q0005
     • Q0000 -- C0002 -- Q0006
     • Q0000 -- C0003 -- Q0007
     • Q0001 -- C0004 -- Q0004
     • Q0001 -- C0005 -- Q0005
     • Q0001 -- C0006 -- Q0006
     • Q0001 -- C0007 -- Q0007
     • etc.
   The solution should have 3 ground states (solutions with objective=0) matching the ones in the spreadsheet, but you may have to translate from the Qbit numbering of QMI to the one in the spreadsheet. If you have more/less than 4 solutions or different ones than expected, then your weights/strengths are wrong.

   Turn in file imply4.bash (dw program) and imply4a.pdf (spreadsheet snapshot)

2. (10 points) Let's try one more embedding exercise, again for the "imply" circuit from of slide 8 in a 4-Qbit subset in the spreadsheet, but this time by embedded q1 into 3 Qbits.
   Hints:
   • Set the a-weights for q1 and q2 and the strength for b12. Notice that you already have a solution, but that's not the point. We want to learn how to embed multiple Qbits at once.
   • To facilitate a 3-way embedding, multiply all weights and strengths by 3.
   • Split q1 into q1+q3+q4, adjust the a-weights (divide by number of resulting Qbits).
   • Connect your split Qbits with the "equal" circuit (but use -4 so that it supersedes the strength of the imply coupler) and adjust the a-weights accordingly (subtract from them the [sum of new adjacent strengths] divided by number of new couplings).
   • At this point, you should be done. However, if you wanted to create an imply coupling between q2 and the other 3 Qbits (3 links), then you would create these links at the same strength given by the original single-coupling divided by the number of target links: -3/3=-1. Try this out.
   The solution should have 3 ground states matching the ground states of slide 8. Verify that the q1+q3+q4 values match for any ground state. (If not, then you've made a mistake.) Notice that the resulting embedding does not match D-Wave's bipartite graph.

   Turn in file imply4b.pdf

3. (20 points) Let's repeat this embedding for the "imply" from slide 8 for a 5-Qbit subset with q1 replicated in another 3 Qbits such that it can be embedded into D-Wave's bipartite graph. Since you cannot do this in the spreadsheet, try it on paper. Then program it in QMI.
   Hints:
   • Please q2 in the middle, then create a "ring" of the other Qbits around it (in north, south, east, west location).
   • Follow the steps from the previous exercise to (a) multiply weights/strengths, (b) split weights, (c) add equalities with weight/strength adjustments, (d) split the original (imply) strengths across the east/west connection.
   • Code it up in dw as in the 1st exercise.

   Turn in file imply5.bash (dw program) and imply5.pdf (graph showing Qbits+couplings with weights+strengths on paper)

4. (10 points) Let's create the 1-bit full adder from slide 41 on Quirk with inputs c0, a0, b0 and outputs s0 and c1.
   Hints:
   • Watch the tutorial.
   • Create the adder circuit by just dropping down X gates (marked as "cross", right below the Y gate) and adding one or two control probes (solid dot) for C-Not and Hadamard gates, respectively.
   • Put a Sample box on the inputs (left).
   • Test your circuit by verifying that inputs/outputs are correct. You can change inputs by putting one or more X gates on Qbit input lines (very left, left of the sample box), and then watch the output.
   • Label inputs and outputs. You can do this by creating a labeled gate for each symbol, i.e., click: make gate, from matrix "{{1,0},{0,1}}" circuit symbol "c0", create. Place it on the c0 input line, etc. for a0/b0/s0/c1.
   • Export your solution as a URL (copy clipboard) in the file added.txt.

   Turn in file adder.txt

5. (20 points) Combine two adders to create a 2-bit adder in Quirk with inputs c0, a0, a1, b0, b1 and outputs s0, s1, and c2. Make sure to label inputs/outputs.

   Turn in file doubleadder.txt

6. (5 points) Create a Toffoli gate in Quirk from basic gates by modifying this solution. Start with the given circuit but substitute gates as follows:
   • T is Z^1/4 (can also be creates by matrix {{1,0},{0,i}})
   • T-dagger is Z^-1/4 (or {{1,0},{0,-i}})
   • S is Z^1/2 (or {{1,0},{0,e^(i*pi/4)}})
   • add a Sample on the inputs
   • try out all input combinations to verify output

   Turn in file toffoli.txt

7. (15 points) Create a Toffoli gate in Quirk from basic gates by modifying this solution. We now try construct Toffoli variation. Start with the given circuit but substitute gates as follows:
   • T is Z^1/8
   • T-dagger is Z^-1/8
   • S is Z
   • add a Sample on the inputs
   • try to add one more gate, X^1/2, in the right place with the right wiring
   • try out all input combinations to verify output

   Turn in file toffoli2.txt

8. (0 points) Learn about superdense coding and then check out the predefined Superdense circuit in Quirk (click menu button). Have a look at how it works in action. Fun stuff to first grasp and then watch. Nothing to turn in, no points :-)