Errata

The line right after Figure 2-12 states 'These four states will be used in Example 2-4'. That isn't true. Example 2-4 uses states |0>, |1>, |+> and |->. In particular, the example does *not* use the states |+Y> and |-Y> presented in Figure 2-12.

Text discussing Example 3-6 first, correctly, mentions the probabilities are 85%/15%, but then later, incorrectly, mentions the probabilities are 70%/30% and sticks to these figures. Luckily, the *code* in the book and online sticks to the correct 85%/15% probabilities.

Figure 7-6 is missing an all-green column of circles on the very right of the figure, for state value |15>

Based on my earlier comment on the use of invQFT in this section, Fig. 8-8 should be changed to reflect that invQFT actually outputs -2 (ie, 6) on input with eigenphase 90°. Also, less seriously, for eigenphase 180° the output of invQFT should be interpreted, rather, as being -4 (which, in this case, is equivalent to 4).

Could it be that qubits 3 and 4 in the upper row should be exchanged with the qubits 3 and 4 in the lower row ?

The top diagram currently shows rotations ( denoted by pink highlight and red arrow ) for qubit states |3>, |5> and |7>
It ought to show such rotations for qubit states |1>, |3>, |5> and |7>, i.e., as the book fails to highlight the state |1>.

The error is actually on the webpage with the code execution window that runs the "QC Engine" simulator.

When you run this code, you see the circle notion showing amplitude circles, but the display is not what it should be. The amplitude circle display does not show *separate registers* for Register 1 and Register 2, as we see, for example at the top of page 52, in figure 3-19.

Because the separate circle amplitudes are not displayed as they should be, this makes it impossible to see the "phase kickback" we are looking for. In other words, as in the example displayed in figure 3-19, the phase should NOT change for register 2, while it does change for register 1.

In fact, the QC Engine also fails similarly in other places in the book, again making it hard to use these circle amplitudes to verify what happened when the code ran.
I will report further cases where found...

The sample Qiskit code, program circuit, and circle notation for Example 3-5 appear to be incorrect. There is an extra -45-degree phase change, a 'barrier' not described in the problem, and a 90-degree phase change that was probably not intended. (The left hand side of Figure 3-27 equates a constructed CPHASE circuit to a 90-degree phase shift (shown on the right hand side), but the sample circuit code actually integrates a 90-degree phase change for some reason). The Qiskit code doesn't appear to match the provided program circuit diagram, and unfortunately, the circle notation for the sample code doesn't appear to match either the desired initial or final state for the circuit.

It appears the QCEngine code and circle notation generator for Example 3-6, "Remote-controlled randomness," output inconsistent results. When the code produces a program output of "01" in binary ("1" in decimal), the circle notation shows P(|2>)=1 (with -70-degree phase angle) when it should show a result for P(|1>)=1. Also, when the program produces an output of "10" in binary ("2" in decimal), the the circle notation shows P(|1>)=1 (with -70-degree phase angle) when it should show a result for P(|2>)=1. (It is understood you have to square the length of the radius shown in the diagram to derive a state probability). So the results appear to be backward for these two states (if I understand how circle notation works). If the code is actually running correctly, please clarify your convention for state notation.

The text seems fine, although it seems a bit "condensed" and slightly difficult to work through.

The figure at the top of the page Fig 3-30 is incorrect. The circle corresponding to the 2 qubit state |0> should be given a weight less than one, namely a weight of approx. 0.837 to correspond to the probability of 70%.

As for the text, it would improve in clarity if a reference were included to send the reader back to page 43 at the bottom, where reading out one bit in a multi-qubit structure is explained. ( This concept is evidently needed only in the code example 3-6, so that the reader has had time to forget what he / she has seen )

The error is actually in the behavior of the code on the website, but I don't know where else to post it but here.

Anyway, on the code website, when one activates example 5-2. the circle display is not properly separated into qubits for register a and qubits for register b.

The displayed qubits therefore act as if registers a and b are simply one large conglomerate of qubits, so that, for example, the first qubit of register b is treated as the conglomerate qubit corresponding to the state |16>. Therefore, when the new value of register a ought to be |2> for example (adding 1 from register a with 1 from register b) the displayed qubit circles are showing |18> which corresponds to a |2> in register a and a |1> in the first qubit of register b.

Same type of trouble in the website code as mentioned before. I.e., the website code in example 5-3 when executed gives circles which do not properly represent the registers a and b. It treats the two registers as a single conglomerate, so that the first qubit of register b is represented as a qubit of hex value 16 in the conglomerate, rather than weight 1 in the register b.

Therefore, when one of the sums in question comes to |2> the circles displayed offer the answer |18> since they are weighting the first qubit of register b as |16>.

The |15> column is missing in Figure 7-6.

The order of bits seems to be different between the address and output registers in the RAM diagram. Reading top to bottom (i.e. treating vertically higher bits as more significant) the address selected is 0x10 which corresponds to the value 0x100 in RAM but the output register shows 0x001 (unless you read it bottom to top. Reading the address bottom to top doesn't help as it corresponds to 0x110 in RAM which doesn't look like the output register however you read it.

The number 3 is encoded as binary 110 (which is 6) and 5 as 111 (which is 7)

It's stated that we normalise the vector [0,1,2,3] by "dividing each component by the summed squares of each component", when it's actually the square root of this sum that we should use to normalise (3.74).

The caption of the figure says "Quantum operations to perform a multiply by 2 modulo 35" while the figure only shows the "modulo 35" part.

The division of the first row of the matrix is a little wrong,
1/0.438 is 2.283... not 2.281