Practice
Short rounds pulled from the checks at the end of each entry. In a round your first answer counts. Each one comes with the reasoning, and the round ends with a list of what to reread.
Every question, with answers
For studying instead of testing. New entries add their checks here when they publish.
Why are almost all -qubit states entangled? Software
B. A product of single-qubit states is fixed by complex numbers, but a general -qubit state needs Each qubit of a product state is one number, so qubits give , while the whole space needs . For that is 10 against 1,023, so nearly every state can't be split into separate qubits (Rieffel & Polak, §3.2, p. 38).
Is entangled? Software
C. Yes: , which is not zero, so no choice of two single-qubit states can produce it. Rieffel & Polak give this state as one of their examples of entanglement (§3.2, p. 38).
Alice applies to her qubit of and does nothing else. What is the pair's state? Software
B. . One Pauli gate on one qubit reaches all four Bell states, which is how dense coding sends two bits by sending one qubit (Rieffel & Polak, §5.3.1, pp. 81–82).
Alice and Bob share many copies of . Alice chooses which angle to measure hers at. Can Bob learn her choice by looking only at his own results? Software
D. No: the correlation shows only when the two lists are compared, which takes ordinary communication Each side alone sees a fair coin; the agreement appears only when results are brought together. That is why an EPR pair can't send a message faster than light (Rieffel & Polak, §4.4, p. 61).
In the three-angle game, quantum mechanics predicts the two results agree half the time. What is the least any set of hidden instructions can give? Software
C. Every instruction set agrees in at least 5 of the 9 equally likely settings, so no local hidden-variable theory can bring agreement down to the seen in experiments (Rieffel & Polak, §4.4.3, pp. 63–64).
What is the concurrence of ? Software
B. : strongly entangled, but less than a Bell state.
You measure in the Bell basis. What can you get? Software
B. or , each with probability , so , and each has squared amplitude (Rieffel & Polak, §3.2, p. 39).
After , what is the amplitude of ? Software
A. Each , so every one of the basis states gets .
In teleportation, what must Alice send Bob before he can recover the state? Software
A. Two classical bits: her measurement results, which tell Bob which Pauli gate to apply Alice's two measurements give two bits that pick Bob's correction: , , or . Teleportation sends a qubit with two classical bits and a shared EPR pair, and is the mirror image of dense coding (Rieffel & Polak, §5.3.2, pp. 82–84).
Two qubits are in . Looked at on its own, what state is the first qubit in? Software
C. No pure state: on its own it is as uncertain as a qubit can be This is what maximally entangled means: each qubit alone is the maximally mixed state , with all the information in the correlations (Rieffel & Polak, §10.2.1, p. 218).
Can single-qubit gates applied separately to each qubit, , turn into the product state ? Software
A. No: never changes whether a state is entangled Entanglement isn't a local property: transformations acting separately on the parts can't create or remove it. Undoing a Bell state takes a two-qubit gate, the CNOT in the entry's circuit run backwards (Rieffel & Polak, §5.2.3, p. 76).
The book's version of the three-angle game sets polarisers at , and , while the entry uses , and . Both give agreement when the settings differ. Why? Software
B. The entry's angles are on the Bloch sphere, which doubles the polariser's angles: there is at the polariser The book's rule is for polariser angles; the entry's is for Bloch angles. Angles on the Bloch sphere are twice those between the state vectors (Rieffel & Polak, §2.5.2, p. 25; §4.4.2, pp. 62–63).
Quantum gates are reversible. Which of these classical gates can't be made reversible using only its own two bits? Software
C. AND AND maps four inputs to two outputs, so information is lost. Reversibly it needs a third bit: the Toffoli gate takes to (Rieffel & Polak, §6.1.1, p. 101).
The entry's is . Rieffel & Polak's is . How are they related? Software
B. The book's is times the entry's: the same transformation up to a global phase The book uses and notes that , the entry's version, is also called the Pauli . Since the two differ by the factor , they act identically on every state, up to a global phase (Rieffel & Polak, §5.2.1, pp. 75–76).
What single gate equals , then , then ()? Software
A. sends to and to . swaps and , and the second maps them back, so the whole thing swaps and : a NOT (Rieffel & Polak, §5.2.2, p. 76).
How many gates in a row make a gate? Software
B. 4 puts on , so puts there: (Rieffel & Polak, §5.2.1, p. 75).
Two qubits start unentangled. Why do you need a two-qubit gate such as CNOT to entangle them, rather than one single-qubit gate on each? Software
C. A gate of the form acts on each qubit separately, so it can neither create nor remove entanglement is entangled exactly when is. CNOT can't be written as a tensor product of two single-qubit gates, which is why it can take to (Rieffel & Polak, §5.2.3–5.2.4, pp. 76–77).
SWAP is then then . Stop after the first two. What does become? Software
C. gives . then replaces qubit 1 with , leaving . The third CNOT finishes the swap.
Which matrix applies to the first of two qubits and leaves the second alone, in the order ? Software
B. Applying to qubit of means applying with in position . For the first of two qubits that is (Rieffel & Polak, §5.2, p. 75).
Which of these is not a valid quantum gate? Software
D. Gates must be unitary. Sums of unitaries usually aren't: has columns of length , so it would shrink states. Scaling by instead of happens to give (Rieffel & Polak, §5.1, pp. 72–73).
What do you get by applying twice to ? Software
D. : the Hadamard gate is its own inverse (Rieffel & Polak, §5.2.2, p. 76).
The entry says , CNOT and can approximate any quantum computation. Why approximate, and not exactly? Software
C. There are uncountably many unitaries, but a finite set of gates can build only countably many circuits No finite gate set generates every unitary exactly, but some finite sets approximate any of them as closely as you like, and efficiently, by the Solovay-Kitaev theorem (Rieffel & Polak, §5.2, p. 74; §5.5, p. 91).
Two single-qubit states are orthogonal, like and or and . Where do they sit on the Bloch sphere? Software
B. At opposite ends of a diameter Orthogonal states are antipodal on the Bloch sphere, so every diameter is a measurement basis. The sphere's angles are twice the angles in the inner-product picture (Rieffel & Polak, §2.5.2, p. 25).
The general state is drawn on the Bloch sphere without its . Why can that factor go? Software
A. It is a global phase: it changes no measurement probability, so it is not part of the state Vectors that differ by a factor of magnitude 1 represent the same quantum state, so the ket picture has a redundancy that the sphere doesn't: each point on the sphere's surface is exactly one state (Rieffel & Polak, §2.5.1, p. 21; §2.5.2, p. 24).
The Bloch form is , with from to . Why and not ? Software
A. So that , the opposite pole, gives , the state orthogonal to With , moving from the north pole () to the south pole () turns the vector by , from to . Angles on the sphere are double the angles between the vectors (Rieffel & Polak, §2.5.2, p. 25).
Which pair represents the same quantum state? Software
D. and multiplies the whole vector by , a global phase, so it is the same state. Changing the phase of only one amplitude is a relative phase and gives a different state (Rieffel & Polak, §2.5.1, pp. 21–22).
Following the hint, what are and for ? Software
B. , gives , and gives (or ). That is the point on the sphere (Rieffel & Polak, §2.5.2, p. 24).
A state on the Bloch sphere has . What is the probability of measuring ? Software
C. The amplitude of is , so the probability is (Rieffel & Polak, §2.3, p. 17).
You keep fixed and change only , moving the state around a circle of latitude. What happens to the probabilities of measuring and ? Software
A. Nothing: is a relative phase, and standard-basis probabilities depend only on and . still matters, though: it changes the state, and measuring in another basis, such as the Hadamard basis, shows it (Rieffel & Polak, §2.5.1, p. 22).
For which is equivalent to ? Software
A. Every multiplies the whole vector, so it is a global phase for any (Rieffel & Polak, §2.5.1, p. 21).
For , which of these is the bra ? Software
A. A bra is the conjugate transpose of its ket: the column becomes a row and every amplitude is complex-conjugated, so becomes (Rieffel & Polak, §2.2, p. 15).
For a unit vector , what is the outer product ? Software
D. The projection onto : it keeps the part of a state along and removes the rest : the inner product with scales a copy of . Operators written as are how Dirac notation describes every linear transformation, and projectors like this one describe measurement (Rieffel & Polak, §4.1–4.2, pp. 47–49).
A qubit in is measured in the standard basis and the result is . You measure it again in the same basis. What do you get? Software
A. , with certainty Measurement moves the state onto the basis vector it reports; a second measurement in the same basis returns that vector with probability 1. That is also why a single measurement can't reveal the original amplitudes (Rieffel & Polak, §2.3, p. 17).
and both give 0 and 1 half the time when measured in the standard basis. Are they the same state? Software
C. No: they differ by a relative phase, and the Hadamard basis tells them apart every time These are and . Measured in the Hadamard basis , each gives its own outcome with certainty. A relative phase is physically meaningful; a global phase is not (Rieffel & Polak, §2.5.1, p. 22; §2.3, p. 17).
A qubit in is measured in the Hadamard basis . What do you get? Software
B. , with certainty and . Whether a state looks random depends on the basis: is random in the standard basis and certain in the Hadamard basis (Rieffel & Polak, §2.3, p. 17).
Cirq measures in the standard basis, so the article applies and then before measuring in . What do those two gates do? Software
B. They turn into and into , so a standard-basis result reads as an -basis result and ; and . Applying a unitary and then measuring in the standard basis gives the same statistics as measuring in the basis it maps onto the standard one (Rieffel & Polak, §5.1, p. 72).
Which of these is not a valid qubit state? Software
B. A qubit needs . Here , so this vector is not normalised (Rieffel & Polak, §2.2, p. 15).
A qubit is . What is the probability of measuring ? Software
C. , and , so the two add to 1 as they must (Rieffel & Polak, §2.3, pp. 16–17).
Is a different state from ? Software
B. No: is a global phase, and global phases have no physical meaning Vectors that differ by a factor of magnitude 1, like , or , represent the same quantum state (Rieffel & Polak, §2.5.1, p. 21).
A qubit can be in infinitely many states. How much classical information can one measurement of it give you? Software
A. At most one bit Each measurement yields one of the two basis states and changes the qubit, so one qubit gives up at most one classical bit (Rieffel & Polak, §2.3, pp. 17–18).