RCQT ศูนย์วิจัยเทคโนโลยีควอนตัม · มหาวิทยาลัยเชียงใหม่

อภิธานศัพท์ › หมายเหตุ

อภิธานศัพท์

หมายเหตุจากเฉลยละเอียด

ก่อนอ้างตัวเลขใด เฉลยละเอียดของหนังสือไตรภาคจะระบุข้อตกลงให้ชัดเจน — บทนั้นใช้นิยามใดในสองนิยามที่มีในเอกสาร และเหตุใดค่าหนึ่งจึงต่างจากแหล่งอื่น อภิธานศัพท์ของเฉลยเก็บหมายเหตุลักษณะนี้ไว้ 294 รายการ: 263 รายการรวมไว้ในหน้านี้ และ 31 รายการอยู่ใต้คำศัพท์นั้นในรายการ A–Z

ป้ายชื่อหนังสือ CT, QS และ SM คือหนังสือไตรภาค — Cooling and Trapping of Neutral Atoms, Cold-Atom Quantum Simulation และ Atom-Based Quantum Sensing and Metrology · IQM 1, IQM 2 และ IQM S คือหนังสือ An Introduction to Quantum Mechanics เล่ม 1 เล่ม 2 และเล่มเสริม · QJS คือ The Quantum Jump Start · ·WS หมายถึงเฉลยละเอียดของเล่มนั้น ตัวเลขคือบทที่ และ § คือหัวข้อย่อย ชี้ที่ป้ายเพื่อดูรายละเอียดเต็ม

หมายเหตุแปลจากต้นฉบับภาษาอังกฤษในหนังสือ (แปลแล้ว 0 จาก 263 รายการ ส่วนที่เหลือยังแสดงเป็นภาษาอังกฤษ) · หน้าภาษาอังกฤษของหน้านี้มีถ้อยคำเดิมของหนังสือ

Cooling and Trapping of Neutral Atoms — เฉลยละเอียด · สัญลักษณ์และข้อตกลง 10

(26.4) is a statement about mean energy. nf=ni+(r-1)2(2ni+1)/4r is identically. CT·WS 26

40K scattering length. Worked Example 17.3 takes 174 a0; Worked Example 17.4 and Appendix D take the same, and 169.7 a0, a value also current in the literature, is reproduced for comparison. CT·WS 17

6Li Feshbach parameters. (17.15) is as=abg[1+Δ/(B-B0)] with abg=-1582 a0, Δ=+262.3 G; Chapter 15’s (15.16) is abg[1-Δ/(B-B0)] with -1405 a0 and -300 G. Opposite sign conventions and different parameters. CT·WS 17

k. The recoil energy of a 1064 nm lattice uses the lattice wavevector, while the Lamb–Dicke parameter of a sideband-cooling scheme uses the wavevector of the probe. Problem 6.6 turns on that distinction and is the reason it is stated here. CT·WS 6

s0 sits in Δ. The saturation parameter s0 is always per beam. What appears in Δ=1+s0+(2δ/Γ)2 depends on how the beams are counted. CT·WS 7

β counts one direction. §28.4’s t=1-2β and OD1=4β are mutually consistent only if β is the guided fraction per direction, so that β=1/2 is the lossless limit; the definition printed beside them, β=Γ1D/(Γ1D+Γ'), runs from 0 to 1. CT·WS 28

δLS feeds Section 25.11. The first printing’s thermal row was 4.40 kHz mK-1 — the leading-order (22.9) value — times kBT/U0=0.01. Its depth-uniformity row was the shifted-ωj 5.82 kHz mK-1 times the same 0.01. Both rows now use Chapter 22’s fixed-dipole shift, 4.70. All three are carried. CT·WS 25

η means in (29.8). §28.6 defines ηd as the total transmission-and-detection efficiency, but (29.8) breaks the fiber out into 10-αL/10, so ηd there is the detection chain alone. §28.5’s cavity value ηgen=0.895 is used for the source, and the chapter’s “η=0.1” is read as the detection chain. CT·WS 29

Ω and Δ. In (29.5) and (29.6) every energy is ℏ×(angular frequency), and the blockade radius is fixed by C6/Rb6=ℏΩ. The chain therefore has no free parameter beyond Rb/a and Δ/Ω: neither C6 nor the species survives. CT·WS 29

Ω is angular. Ω and V are angular frequencies throughout; the chapter quotes Ω/2π and V/h in megahertz, so the dimensionless V/ℏΩ is (V/h)/(Ω/2π). CT·WS 27, 30

Cooling and Trapping of Neutral Atoms — เฉลยละเอียด · หมายเหตุในเนื้อหา 171

Alkali differential shift has a floor. As ω→0 the corrected two-line shift tends to νhfs/νD2 per unit depth, or 371 Hz mK-1 for 87Rb. CT·WS 22

And. η cancels identically. Cooling and heating rates both carry η2Ω2; the rate-equation steady state at η=0.05 and at η=0.80 agrees to fifteen figures. η sets the speed and the speed alone. CT·WS 26

And with the curvature put back. For L≪Rc, w02→(λ/2π)√(2LRc) and. CT·WS 28

Appendix B’s envelope rule. Integrating A=∫01g″eiθs ds by parts leaves boundary terms in g″,g‴,…; the first that does not vanish is the m-th derivative of g, giving |A|∼θ-(m-2) and ΔE∼d2/T2θ2m-4∼T-2m. CT·WS 24

Aspect ratio. ωr/ωz=√2πw0/λt is quoted as a number greater than one throughout; the axial direction is the weak one. CT·WS 2, 6, 11 …

Bandwidth. Appendix D writes the Floquet window as 4t≪ℏωd≪2√sEr, so its lower edge is the full bandwidth 4t. At s=10 that is 156 Hz, which is what Worked Example 21.2 itself uses; §21.10’s 156 Hz is 4t/h. CT·WS 21

Bang-bang spectrum. ΔE=128Md2sin4(ωT/4)/ω2T4; zeros at T=2n/fr, maxima at T=(2n+1)/fr, and the first zero is the speed limit. CT·WS 24

Bang–bang transport spectrum. ∫01f″eius ds=-(4/iu)(eiu/2-1)2, whose modulus is 16sin2(u/4)/u. Zeros at u=4πn; envelope maxima at u=2π(2n+1); and the T-4 decay of ΔE follows in one line. CT·WS 26

Barrier of a bottle beam. The LG00-LG10 bottle has an axial barrier exactly e2/4=1.847 times its radial ring, and a true escape barrier — the mountain pass, computed here by flooding the (r,z) plane — lower than either, at 0.532 of the ring. CT·WS 23

Beam radius. w is the 1/e2 intensity radius throughout, so I(r)=I0e-2r2/w2 and I0=2P/πw02. Where a “diameter” is quoted it is 2w. CT·WS 13

Beam speeds. The effusive flux distribution is ∝v3e-v2/α2 with α=√(2kBT/M), the gas most probable speed. The beam most probable speed is α√(3/2)=√(3kBT/M). Both appear constantly and they differ by 22%; every expression below states which is meant. CT·WS 8

Bessel functions. Computed from the ascending series, pure standard library, and checked against the sum rule ∑n​Jn2(β)=1, which holds to twelve figures at every β used here. CT·WS 18

Bias field. Over a 200 ms hold the loss is 0.26%, the chapter’s “0.3%”. with every coefficient in hertz: The bias field is the instructive entry. CT·WS 1, 4, 8 …

Binding energy. A state with zero binding energy is a state at threshold, and threshold is where as is infinite. CT·WS 2, 11, 15 …

Blackbody shift and its two halves. Appendix D splits the strontium shift into a static -4.96×10-15 and a dynamic -3.4×10-16; Section 20.4 and Appendix A quote a total of -5.3×10-15. The three close. CT·WS 20

Blockade leakage. εleak=1/8(R/Rb)12 contains no n, no C6, no Ω and no species, which is why Worked Example 27.2’s conclusion is so robust and why the array uniformity requirement is a percent-level one. CT·WS 27

Blockade pitch window. For a lattice whose second-neighbor distance is ρ times its first, the pitch range that blockades one and not the other is exactly a factor ρ wide, i.e. 1-1/ρ fractionally. A square array gives √2 and 29.3 %; a chain gives 2 and 50 %; a triangular, honeycomb or kagome array gives √3 and 42.3 %. CT·WS 23

Blockade radius and the soft-core radius. Rb/Rc=(2Δ/Ω)1/6 identically, from (27.5) and (27.7). CT·WS 27

Blockade shift. V is always an energy divided by h, a frequency in hertz; Ω is always an angular Rabi frequency. The chapter’s dimensionless V/ℏΩ is therefore (V/h)/(Ω/2π), and reading it that way reproduces all twenty entries of Worked Example 20.3’s table. CT·WS 20

Blockade threshold. Deep in blockade the residual double excitation is Ω2/2V2, so the 10-3 criterion is V/ℏΩ=√(500)=22.36 asymptotically and 22.98 exactly. CT·WS 20

Both magic-angle slopes. d(1-3cos2θ)/dθ=3sin2θ, and at θm=arccos(1/√3) that is 2√2=2.828427 exactly; |dP2(cosβ)/dβ| at the same angle is √2 exactly. CT·WS 30

Bottle’s barrier ratio. For LG00-LG10 the on-axis intensity is (4P/πw02) ζ2/(1+ζ2)2 with ζ=z/zR, peaking at ζ=1; the radial profile at z=0 is (P/πw02)x2e-x with x=2r2/w02, peaking at x=2. Hence. CT·WS 23

Carrier infidelity on a thermal state. Expanding Ln(η2)≃1-nη2 and averaging cos2(πx/2) over the thermal spread ⟨x2⟩=η4n(n+1),. CT·WS 26

Clock cycle time. (29.7) contains Tc, which §29.8 never states. Four of its five rows imply Tc=2.0 s, that is one second of dead time at T=1 s. The fifth, at its own T=10 s, implies the same second of dead time: 11.1 s. Tc=T+1 s is used throughout. CT·WS 29

Code distance. Appendix A gives d=2ln(A/PL)/ln(pth/p)-1, and §29.2 tabulates the distance “needed” for PL=10-10. A surface-code patch has odd distance, so the continuous value is rounded up to the next odd integer — and when it lands on an odd integer, that integer already meets the target. CT·WS 29

Coil. Section 19.5’s design (R=4 cm, N=100, I=10 A, B'=0.50 T m-1) and Worked Example 19.2’s (R=5 cm, N=80, I=5.8 A, B'=0.15 T m-1) are different coils that happen to share a resistance of 188 mΩ, because 100×4=80×5 makes their wire lengths identical at 25.13 m. CT·WS 19

Coil separation. Equation (12.31) uses the full separation d, with the coils at ±d/2; (12.32) uses the half-separation. Both describe the same pair and both are maximized at a full separation of R. Both give the coefficient 0.8587 there; 0.6413 belongs to the uniform-gradient separation √3R. CT·WS 12, 19

Cold-cavity linewidth. Cold_cavity_linewidth returns the FWHM, ΔνFSR/F with F=π√R/(1-R). This matters. CT·WS 18

Collection fraction. §28.1 and Chapter 22’s (22.15) both use the isotropic (1-cosθ)/2. An atom is a dipole: the pattern-weighted fraction into the same cone is (4-3c-c3)/8 for the favorable orientation, 1.16–1.45 times larger, and [3(1-c)-(1-c3)]/4 for a dipole along the optical axis, seven times smaller at NA=0.5. CT·WS 28

Collective decay rate of a chain. Γ(q)/Γ=(λ/2d)×(propagating orders at q), from ∑j≥1​sin(jx)/j=(π-x)/2 and confirmed to eight figures by direct summation. CT·WS 28

Composite-pulse orders. Fitted over four decades: simple π, 2.000 in both channels; SK1, 4.000 in amplitude and 2.000 in detuning; BB1, 6.000 and 2.000; CORPSE, 2.000 and 4.000. All four are the chapter’s orders. CT·WS 25

Convergence threshold. Setting D∞=1 in (30.5) with a round time tr=t0+2Tmove√N gives. CT·WS 30

Coriolis ratio. The rotation-induced fractional error in a gravimeter is δg/g=2Ω⊕v⊥/g — independent of keff and of T, because both cancel between (20.15) and (20.11). CT·WS 20

Crossed deflector. Its depth matrix is rank one, so an Nx×Ny array has Nx+Ny-1 free parameters against NxNy sites: 31 against 256. A single vacancy is not merely difficult but algebraically impossible, since AiBj=0 empties a whole row or column. CT·WS 23

Crossed pair cannot be isotropic either. Adding curvatures, ωy/ωx=√2/√(1+(ωz/ωr)2), which is bounded above by √2 and reaches unity only when the single beam is already isotropic. The reference pair gives 1.389. CT·WS 22

Current gate error is. §30.8 says two-qubit errors have fallen to 5×10-3; §30.11, Appendix C and the whole of Chapter 29 use 5×10-3. It is not the budget of §27.10 at n=70 and R=5 μm, which is 3.7×10-3; it is the measured number. CT·WS 30

Cycling transition. The cycling transition is the one whose upper level cannot decay to the other ground manifold, both at λ=852.347 nm. CT·WS 1, 2, 3 …

Deflector’s Rayleigh pitch. The resolvable spots of (23.2) are separated by λf/D, and the Gaussian waist a pupil of 1/e2 diameter D produces is λf/(πD/2). The ratio is π/2 exactly: the time–bandwidth product counts spots at 1.571 waists, while Section 23.2’s crosstalk floor demands three. CT·WS 23

Density averages. ⟨n⟩=n0/23/2 and ⟨n2⟩=n02/33/2 for a thermal Gaussian cloud; ⟨n⟩=4/7n0 and ⟨n2⟩=8/21n02 for a Thomas–Fermi condensate. Keeping the four apart matters; Appendix E’s 5 μK magnetic-trap rows are thermal and take the Gaussian pair. CT·WS 15

Dephasing crossover. Δδunif/Δδtherm=(1-U) U0/kBT, free of δLS, so a global pulse is limited by the hologram rather than by the temperature exactly when U<1-kBT/U0. CT·WS 23

Design rule inverted. Solving (24.2) for N is a quadratic in √N:. CT·WS 24

Diffusion. The momentum diffusion coefficient is defined by ⟨Δp2⟩=2Dt, which is the convention of (6.23) and the one that makes dE/dt=D/M come out right. The hint to Problem 6.7 quotes the rate of growth of the variance, d⟨Δp2⟩/dt=2D, which is twice as large. CT·WS 6

Diffusion coefficient. ⟨Δp2⟩=2Dt throughout. Equation (7.14) is written for 2Dz precisely so that no factor of two can go missing, and dE/dt=D/M then follows without further care. CT·WS 7

Dipole is in. Vdd. (30.3) is written with an effective dipole and an angular factor, and then §30.4 tabulates d2/4πε0R3 with the full permanent dipole and the angular factor set to one. That combination is the fully polarized, side-by-side value. CT·WS 30

Double well. A Gaussian barrier of waist wb in a trap of frequency ω splits it only above Vb=Mω2wb2/4, and the minima then sit at. CT·WS 21

Echoed addressing gate. The sequence [π/2 addressed][π global][π/2 addressed][π global] gives the target a net 3π — a π rotation — and the spectators a net 2π, with the light-shift phase of the first half reversed by the intervening π. CT·WS 25

Effective moment. μ'=(gF'mF'-gFmF)μB, which for the 87Rb |2,2⟩→|3,3⟩ stretched transition is μB exactly. CT·WS 14

Energy that defines a heating rate. dE/dt=2ErΓsc with E=3kBT — a harmonic trap, kinetic plus potential — gives dT/dt=2ErΓsc/3kB, which is the chapter’s convention and the one used here. The ballistic convention E=3/2kBT would double every number. CT·WS 22

Exact simulation stops at fifty-six spins. A state vector at 16 bytes per amplitude fills an exabyte at N=55.8. The certification gap of §30.10 is therefore not 103 against 103-1; it is 103 against 56. CT·WS 30

Exact Zeeman energies. Where the linear approximation is not enough — the clock states of Problem 12.1(c), the magic field of 12.1(d) — the Breit–Rabi formula is used exactly, with gJ=2.002331 and the nuclear gI included. This is what fixes the 87Rb magic field at 3.2289 G without any fitted input. CT·WS 12

Exchange matrix element. ⟨1,0|dcosθ|0,0⟩=d/√3 exactly, so the rotational flip-flop carries d2/3 and not d2. CT·WS 30

Fermi gas. Entropy is conserved, and the entropy of a weakly interacting Fermi gas is a known function of T/TF, so a measurement there fixes S and hence T at unitarity. CT·WS 15, 17, 19 …

Fidelity is “today”. §29.4 reports 0.995 for rubidium and 0.998 for ytterbium; Worked Example 29.1 calls 5×10-3 “today” and Appendix C’s 2025 row says 0.998. Both are carried, and the ytterbium row is the one the chapter never computes. CT·WS 29

Field zero. The field zero is a line, and the transverse structure about it is a two-dimensional quadrupole — the same (1,-1) shape as the Ioffe bars, produced by one wire. CT·WS 2, 3, 7 …

Förster crossover radius. At Rc the van der Waals branch, the resonant branch and the Förster defect are all the same number, 710 MHz — because Rc is the only length the pair problem contains. CT·WS 27

Gas in (19.6). Conductance goes as M-1/2, so every pumping answer depends on which molecule is asked about — by a factor of 6.6 between hydrogen and rubidium. CT·WS 19

Gate the budget. The Section 27.10 table is headed “with the Jaksch sequence” and uses A=7π/4=5.4978, and its tgate column is 4π/Ω, which is the Jaksch duration. At the same Ω a Levine–Pichler gate takes 8.59/Ω, 1.46 times shorter. Both are tabulated below. CT·WS 27

Gauss and tesla. The chapter quotes every field twice, and every pair agrees; a slip in the curvature units would cost a factor of 100. Gradients: 1 T m-1=100 G cm-1. Curvatures: 1 T m-2=1 G cm-2, so 200 G cm-2=200 T m-2 — not 2. CT·WS 12

Geometry. The Doppler limit TD=ℏΓ/2kB is the three-dimensional result. A strict one-dimensional molasses gives 2/3 of it, as Problem 7.5 shows. When a temperature is quoted below without qualification it is the 3D value. CT·WS 7

Gravimeter. Worked Example 20.2 (T=94.6 ms, N=106, contrast C=0.5, one second of averaging at 2 Hz) and Worked Example 20.4 (T=160 ms, C=1, one hundred seconds at a 0.5 s cycle) are different instruments that share a species and a target. CT·WS 20

Half-polarization field. §30.4 uses E∼2B/d. Diagonalizing the rotor gives ⟨d⟩/d=1/2 at dE/hB=2.150410, so the exact field is 7.5 % above the chapter’s rule. Appendix A’s kilovolts per centimeter are reproduced by 2B/d and are checked against that rule, not against the exact one. CT·WS 30

Hand-over frequency ratio. Two equal traps double the depth and raise ω by √2, as Section 26.8 and Worked Example 26.2 both say. CT·WS 26

Heating convention. Chapter 22 deposits 2Er/3kB=0.1207 μK per scattered photon — one recoil of absorption plus one of emission, shared over three degrees of freedom. Chapter 28’s 2.1 nK and 870 μK both use it, and it is used here unchanged. CT·WS 28

Heating rate from scattered light. Each scattering event adds 2Er of kinetic energy — one absorption recoil and one spontaneous-emission recoil, added in quadrature. In three dimensions E=3/2NkBT, so dT/dt=4/3ErΓsc/kB. The Problem 18.1 hint uses dT/dt=2ErΓsc/kB, i.e. E=NkBT. CT·WS 18

Infidelity of a pulse. Section 25.4’s table — all thirty-two entries, both halves — is the gate infidelity 1-|Tr(V†U)/2|2 of the realized propagator against the intended rotation. CT·WS 25

Is angular and what. g0, κ, Γ and Ω are angular rates; the chapter quotes g0/2π and κ/2π in megahertz. κ is the field decay half-width, so the photon number decays at 2κ; Γ is the full population decay rate, so the coherence decays at Γ/2. CT·WS 28

Jaksch blockade error. ε=3π2/64(Ω/V)2, above, reproducing a decade of direct integration, at integer V/ℏΩ, to 0.6 % and pinning its ratio to the bare leakage at 3π2/8=3.7011 exactly. CT·WS 27

Lattice units. Depths are s=V0/Er in units of the recoil energy of the lattice light: for 1064 nm and 87Rb, Er/h=2.028 kHz. Band structure, Wannier functions, J and U come from ctna/lattice.py, which diagonalizes the Mathieu matrix (11.20) by Sturm-sequence bisection — no NumPy, and no library at all. CT·WS 11

Levine–Pichler construction. Three conditions and three unknowns. The |11⟩ sector must return after each pulse, so √(2Ω2+Δ2) τ=2π and its phase is ϕ11=Δτ. CT·WS 27

Levine–Pichler Rydberg area of the pair sector. Because the |11⟩ sector executes a complete generalized-Rabi loop in each pulse, ∫ΩPW dt=(Ωτ)3/2π2=4.00734 identically, and the |01⟩ area is exactly twice its first-pulse value by time-reversal symmetry. CT·WS 27

Lifetime. Section 27.2 sets τrad=1.43 ns×n*3, giving 428 μs at n=70 and τeff=145 μs at 300 K. Chapter 20’s Appendix A gives 300 μs and 127 μs, which implies τ0=1.00 ns. CT·WS 27

Light-assisted release. From (22.13), ΔE(R)=ℏ|δ|[(RC/R)3-1], so emission at RC/2 releases exactly 7ℏ|δ| and at RC/3 exactly 26ℏ|δ|. CT·WS 22

Magnetic field. The CMOT is not at the sub-Doppler floor, and cannot be: the magnetic field is still on, and the magnetic field is exactly what prevents polarization-gradient cooling from working. CT·WS 1, 2, 3 …

Majorana coefficient. The book carries two, and they differ by a factor of 11.6. Both are used below, always labeled. CT·WS 12, 13

Maximum flux. Φmax=2d/λd, and with d=λ/2 and λd=λ that is exactly 1 — so a lattice built at its own wavelength reaches precisely one flux quantum per plaquette at 180∘ and not a fraction more. That the answer is exactly one, rather than nearly one, is not a coincidence but the statement q≤2k. CT·WS 21

Mean displacement. The superseded ∼Nt(N/Nt-1)Nt/2 carries twice the correct coefficient and uses the filling fraction where the loading probability belongs. The closed form is derived below, reproduces all six rows to three percent, and is D=(1/p-1)Nt2/4. CT·WS 24

Mean Rydberg area. A is the mean over the four computational inputs of Ω×(atom-time i⁠n |r⟩). Integrating the sequences gives 7π/4 to five figures for Jaksch — the chapter’s value, confirmed — and 3.2870 for Levine–Pichler, the chapter’s 3.29 and not the 4.29 of its single-pulse area Ωτ. CT·WS 27

Mean speed in (19.1). Section 19.1’s quoted 1782 m s-1 is the mean speed √(8kBT/πM) of H2 alone at 300 K with M=2.0 u — not the mean relative speed of the Rb–H2 pair, which is 1796 m s-1. CT·WS 19

Mean square velocity in (20.1). The equation writes 3kBT/2Mc2, the three-dimensional average. Section 20.1’s quoted -4.8×10-13 for 87Sr at 300 K is the three-dimensional value; the one-dimensional value kBT/2Mc2, a factor of three smaller, is the convention a beam measurement might use. CT·WS 20

Minimum-jerk peak curvature. f″(s)=60s-180s2+120s3 has extrema at s=1/2(1∓3-1/2)=0.21132 and 0.78868, where |f″|=10√3/3=5.77350. The hint’s 5.77 is that number, and the trap-atom offset of (26.5) is 10d/√3(ωT)2 exactly. CT·WS 26

Minimum-jerk spectrum. With E=eiθ and θ=ωT,. CT·WS 24

Mirror curvature. §28.2 tabulates six cavities by (L,F,w0) and never states Rc — but (28.2) determines it: 49.9 mm, 200 μm, 150 μm, 100 μm, 50.3 mm, 497 mm. CT·WS 28

Misread fraction depends on the loss and on nothing else. Eliminating timg and τimg between t*=timgkth/Ndet and both exponentials,. CT·WS 24

Model transition. Jg=1/2→Je=3/2 unless stated. This is the transition for which ΔU=2/3|U0| is exact; for a real F=2→F'=3 alkali manifold it is not, and Problem 9.6 finds the discrepancy showing up as a fitted C=0.62 against a predicted band of 0.1–0.5. CT·WS 9

Moment. μ'=|gFmF|μB throughout, computed from (2.16), nuclear term included, rather than tabulated, so the same code covers every alkali. For 87Rb, gF(F=2)=+gJ/4+3gI/4=0.49984, which is +1/2 to three digits; the stretched state therefore has μ'=0.9997 μB, and we write μB for it as the chapter does. CT·WS 12

Momentum in recoil units. Write p=u ℏk. Then the dark-state splitting (10.6) is Δω=4u ωr — a form worth committing to memory, since ωr is the only rate a free atom possesses and u is the only variable in the problem. CT·WS 10

Motional entropy. (30.4) at n=0.05 is 0.201016, the 0.201 §30.7 prints, so three modes give 0.6030 kB, its 0.60. Every conclusion below uses 0.6030. CT·WS 30

Natural linewidth. That is what an ion has that a neutral alkali does not: a transition whose natural linewidth is a fraction of a hertz. CT·WS 1, 2, 3 …

Numerical aperture of the reference tweezer. w0=1.00 μm at 852 nm is NA=0.271 by (22.4). NA=0.5 is the imaging aperture of Section 22.13, and at NA=0.5 the diffraction-limited waist is 542 nm. CT·WS 22

Néel temperature. (17.37) gives kBTN=0.946 Jex, the cubic-lattice Heisenberg value. Section 17.17, Worked Example 17.4 and Appendix D use it too; the literature’s 0.4 Jex is used below only for comparison, always labeled. CT·WS 17

One debye at one micrometer. d2/4πε0r3=10-31 J=h×150.92 Hz exactly, because 1 D=10-18 statC cm makes the Gaussian expression d2/r3 evaluate to 10-24 erg. CT·WS 20

One inverse answers both halves of Problem 29.2. f=(1-1/g)/(1-pthP/pthE) gives the conversion fraction for a threshold gain g in part (b) and, with g read as the tolerable raw-error penalty, the break-even fraction in part (c). The ceiling is g=pthE/pthP=4 at f=1, so no penalty above four can ever be paid for. CT·WS 29

One quantum of motion. At n=1, (30.4) gives S=2ln2 per mode identically, so a three-mode atom after molasses carries exactly six bits. CT·WS 30

Oscillatory integrals. Several results here are Fourier components of a transport trajectory at a trap frequency, and a single Gauss–Legendre panel resolves only about fifty oscillations. All such integrals are paneled at six nodes per cycle. CT·WS 13

Overhead is a quadratic. With x=ln(A/PL)/ln(pth/p) and d=2x-1, nphys=2(2x-1)2-1=8x2-8x+1 exactly. The -8x+1 is not negligible: at p=5×10-3 it is 3 % of the total, and at p=10-4 it is 22 %. CT·WS 29

Phase-noise error contains no filter function. For a Lorentzian line of full width Δν the frequency noise is white, and first-order perturbation in the dressed basis gives. CT·WS 30

Photon cost of an addressing shift. Both the differential light shift and the scattering rate are proportional to I/δ2, so in the window νhfs≪|δ|≪Δfs CT·WS 25

Polarization factor. Section 19.17 gives αpol=1–2.14. For a uniformly populated ground manifold the Clebsch–Gordan sum rule gives ∑q​|⟨Fm;1q|F'm+q⟩|2=(2F'+1)/(2F+1) for every m, so. CT·WS 19

Power-law traps. U∝r3/δ, so δ=3/2 is harmonic and δ=3 is linear. The virial theorem then gives ⟨U⟩=δkBT, and with the kinetic 3/2kBT,. CT·WS 16

Prefactor is not a constant. (29.2) uses A=0.1. The exact combinatorial prefactor of a distance-d code is (d/((d+1)/2))pth-(d+1)/2, which tends to 1/√(2πd); A=0.1 is the d≈16 value. Every number below is computed with the chapter’s A=0.1, and the exact A(d) is carried alongside. CT·WS 29

Prefactor of (29.2). A distance-d code fails when more than half its elements fail, so the leading term of the exact binomial tail is (d/((d+1)/2))p(d+1)/2, and since (d/((d+1)/2))→2d√(2/πd),. CT·WS 29

Probe ceiling and the readout time. The field at the atom must stay below nc, and in an absorptive readout the atom suppresses that field by (1+2C)-1, so the empty-cavity occupation may reach nc(1+2C)2=(g0/κ)2/2 and the readout takes t=nprobeκ/g02 — 0.91 ns for the fiber cavity, three orders below the detector. CT·WS 28

Quantum volume. A square circuit of width n carries n layers of n/2 gates and passes the heavy-output test while that count stays below ln(3/2)/ε, so n*=0.9005/√ε: at ε=5×10-3 a 6100-atom array has the quantum volume of a thirteen-qubit machine. CT·WS 29

Raman error and the Raman power. επ∝|δ1-1-δ2-1| and the intensity at fixed ΩR is proportional to the reciprocal of the same bracket, so επ×I is invariant. Section 25.8’s last two columns multiply to 5.97×10-5 in every one of its six rows. CT·WS 25

Rates against frequencies. The coherence budget (21.4) relates angular quantities. Evaluating Ωeff/Γsc=Δ/Γ in cycles per second throughout returns a number 2π smaller than the scattering rate in s-1. CT·WS 21

Read-noise optimum. Substituting y=x/2 turns the read-noise condition ye2y=1+e2y into the shot-noise condition xex=2(1+ex), so the read-noise optimum is exactly half the shot-noise optimum, 1.1089 against 2.2177, with no free parameter. CT·WS 19

Readout threshold. Worked Example 25.3 prints kth=29 for its first case but uses the unrounded 28.76 in t*, and prints 46 for its second (unrounded 46.68) and uses that. ctna/tweezer.py’s convention is i⁠n⁠t(kML)+1. CT·WS 25

Recoil. Er=ℏ2k2/2M, ωr=Er/ℏ, Tr=2Er/kB, as in (6.14). For 87Rb on the D2 line these are h×3.771 kHz, 2π×3.771 kHz and 361.96 nK; the frequently quoted Er/kB=180.98 nK is half of Tr, not equal to it. CT·WS 6

Recoil heats the motion. The rate 0.556 s-1 is 2Er(780 nm)Γsc/3hfr — the Chapter 22 convention, which applies the D2 recoil at every trap wavelength. Scattering of 852 nm trap light is elastic, so the deposit is 2Er(852 nm)/3 and the rate is 0.467 s-1, Section 26.11’s value. CT·WS 26

Recoil that heats the trap. Section 22.13 fixes 2Er/3kB=0.121 μK per scattered photon, which Section 22.6’s heating column, from 808 to 1550 nm, replaces row by row with the trap-wavelength recoil. The Er in it is the 780 nm D2 recoil. CT·WS 22

Recoil velocity. The warning is worth heeding — one recoil velocity is the entire width of a good Raman knife. CT·WS 1, 4, 5 …

Reference tweezer. Every number is anchored to Chapter 22’s: 87Rb at 852 nm, w0=1.00 μm, U0/kB=1.00 mK, fr=98.45 kHz, 3.37 mW per site, at pitch d=5 μm. Section 24.4’s twelve heating entries follow from those two numbers and nothing else. CT·WS 22, 23, 24 …

Release and recapture in a harmonic trap. For a point source the recaptured fraction is the χ2 distribution function with three degrees of freedom,. CT·WS 22

Release-and-recapture potential. All twenty entries of Section 22.16’s table are reproduced to ±0.003 by U=-U0exp(-2r2/w02)/[1+(z/zR)2] — which is (22.7) without the w2(z) expansion inside the radial exponent. Equation (22.7) itself — the full potential, which Worked Example 22.3 uses — is also evaluated, and differs. CT·WS 22

Resonant dipole–dipole coefficient. (15.23) displays C3=|deg|2/2πε0∼3ℏΓ/2k3. It is the bare matrix element times the angular factor 2 of the axial Σ state, and every number in the chapter is computed from it. CT·WS 15

Rydberg dipole matrix element. Running (27.4) backwards from the chapter’s own C6=820 and its own quantum defects gives C3/h=24.13 GHz μm3 and a pair dipole of 4975 ea0, that is 1.11 n*2ea0 — squarely inside the textbook range of 1.0–1.5. The n*2 law is a consequence of the chapter’s own numbers. CT·WS 27

Rydberg gate time. 522 ns is the Levine–Pichler gate at §27.10’s n=70 optimum. The table’s tgate column is not 8.59/Ω but the Jaksch duration: a Jaksch gate at the same Ω takes 4π/Ω=764 ns. Problem 30.1’s hint carried the 338 ns of the first printing and now uses the 522 ns. CT·WS 30

Rydberg lifetime in a hybrid. 145 μs is the 300 K effective lifetime of §27.2 and 428 μs the pure radiative one. A superconducting hybrid sits at 10 mK, where the blackbody channel is dead to forty-six decimal places and 428 μs is the right number. CT·WS 30

Rydberg-Ising chain has exactly two parameters. Dividing (29.5) by ℏΩ and using C6/Rb6=ℏΩ gives Vij/ℏΩ=(Rb/a)6/|i-j|6 identically. Neither C6, nor n, nor the species survives; V2/V1=1/64 always; and the ordering period is the integer part of Rb/a plus one. CT·WS 29

Saturation. s0=I/Isat is the resonant saturation parameter and s=s0/[1+(2δ/Γ)2] the off-resonant one, so that ρeess=1/2s/(1+s). In a standing wave s0pk denotes the value at an antinode. CT·WS 6, 8, 9

Saturation and detuning. s0 is the saturation parameter per beam, δ<0, and Δ≡1+2s0+(2δ/Γ)2 is the one-dimensional two-beam denominator of (14.1). CT·WS 14

Scattering lengths. Chapter 21 uses a=100.4 a0 for 87Rb, 443 a0 for 85Rb, 33 a0 for 39K and 280 a0 for 133Cs. With those, its species table reproduces to four figures: 480, 2956, 655 and 1194 Hz. CT·WS 21

Scattering rate. At 1064 nm the three defensible answers are The scattering rate is the most sensitive because () contains the square of the bracket. CT·WS 1, 2, 3 …

Sensitivity of the overhead to the gate. dlnnphys/dlnp=2/ln(pth/p) exactly, which is 2.885 at p=5×10-3 and 0.434 at p=10-4. That single expression is the content of the hint’s “hypersensitive near threshold and insensitive far below it”, and it is what makes (ii) and (v) inevitable. CT·WS 29

Shielding barrier. For U(R)=C3/R3-C6/R6 the maximum is at Rb=(2C6/C3)1/3 and its height is. CT·WS 30

Sideband-cooling prefactor. (26.1) is nmin=(Γeff/4ω)2(1+ξ), and every number the chapter draws from it — all eight entries of Section 26.2’s table, Appendix A’s 6.4×10-4, both of Worked Example 26.1’s limits, and Problem 26.1(c)’s 7.6 kHz — is (Γeff/4ω)2 with ξ=0. CT·WS 26

Sign of the light shift. U0∝δ, so red detuning gives U0<0 and the sublevel potentials are wells. The chapter writes |U0| throughout and so do we; where a sign matters — in gray molasses, Problem 9.9 — it is stated explicitly. CT·WS 9

Sign of the resonance condition. The atom moves along +z and the slowing beam counter-propagates, so in the atom’s frame the laser is blue-shifted to ωL+kv. Resonance with the Zeeman-shifted transition ω0+μ'B/ℏ therefore requires. CT·WS 8

Simultaneity requirement. M independent exponential links are all alive after HMtlink, the expectation of the maximum of M exponentials. H10=2.92897, so ten hops need 2.93 link times and not ten. CT·WS 29

Single atom. A negative Q has no classical description, and its measurement on the fluorescence of a single atom is the cleanest demonstration that the field is quantized. CT·WS 5, 10, 19 …

Sorting model. Section 24.3’s table is reproduced entry for entry — mean displacement, atoms moved and elapsed time, all six rows — by exactly one model, and it is worth naming precisely: N sites loaded independently at probability p; the target the centered contiguous block of Nt sites; the solution the order-preserving matching of a contiguous block of occupied sites, minimized over which block. CT·WS 24

Span of an array. The. 75 μm for sixteen sites at 5 μm is the center-to-center extent (16-1)×5; as in Worked Example 23.2, Appendix D’s d√N is 16×5=80 μm for the same array. Both are computed here; the difference is one pitch, and it is the difference between counting fence posts and counting panels. CT·WS 23

Species data. 87Rb throughout unless stated. The hint to Problem 8.1 and Section 8.14 both use the isotope: the element’s atomic weight 85.468 in place of the isotope’s 86.909 would shift their velocities by 0.8%. CT·WS 8, 9, 10 …

Spin-changing fraction. Section 25.11 quotes [Δfs/(2δ1+δ2)]2; (25.4) gives 1/3(δ2-δ1)2/(δ22+2δ12) exactly. They agree to 1.5 % at 852 nm and part company near resonance, where the exact form tends to 1/3 and the quoted one does not. CT·WS 25

Square root. The stretched state |I+1/2,I+1/2⟩ has μ’≃μB at every field — its Breit–Rabi energy is linear in B to all orders, since for |mF|=I+1/2 the square root is a perfect square — and it is the deepest trap available. CT·WS 2, 5, 7 …

Squeezed vacuum at. r=ln2. cosh(ln2)=5/4, sinh(ln2)=3/4, tanh(ln2)=3/5 — a three-four-five triangle — so n=sinh2r=9/16 exactly, P(0)=1/coshr=4/5 exactly, P(2)=18/125, and every occupation probability of a two-fold position squeeze is a fraction with a power of five in the denominator. CT·WS 26

Stark convention. The Rydberg polarizability is quoted in MHz (V cm-1)-2 and the shift is δν=αE2: as §27.15 says, the α of §27.1 is the coefficient of E² itself, half the α₀=524 of a −½α₀E² shift. Section 27.3’s 26 kHz at 10 mV cm-1 from α=261, and Worked Example 27.1’s 15 mV cm-1 at α=693, both use it. CT·WS 27

Stark problem has one parameter and one pure number. Everything about the rigid-rotor polarization depends only on λ=dE/hB; ⟨d⟩/d→λ/3 at weak field, so the rotational polarizability is αrot=d2/3hB exactly, and ⟨d⟩/d=1/2 at λ=2.1504100 — a universal constant of the rigid rotor, the same for every polar molecule. CT·WS 30

Sticky complex. RRKM with one open channel says τc=2πℏρ=hρ, so the mean level spacing of the complex, 1/hρ, is exactly 1/τc. A 1 ms complex is a four-body resonance every kilohertz — 20.8 states per microkelvin — and that, not the lifetime, is the number to compare with anything else. CT·WS 30

Storage bound. With τd=OD Γ/Ωc2 and a transparency window Δω=Ωc2/Γ√(OD), the product is τd Δω=√(OD) exactly, with neither Ωc nor Γ in it. CT·WS 28

Stretched-state sensitivity. 0.70 MHz G-1 is gFμB/h with gF=1/2: the Zeeman slope within an F manifold. The |1,-1⟩→|2,-2⟩ transition of Section 25.3 changes mFgF by 3/2 and its slope is 2.10 MHz G-1. CT·WS 25

Survival is the narrow line’s. Section 24.5’s table gives it a loss of 1.0×10-4 per image; §24.7, Appendix C, Worked Example 24.3 and Problem 24.8’s hint all use 10-4, and 10-4 is shown below to be the background-gas floor of the very exposure the table describes. CT·WS 24

Syndrome-round budget. r=|lnP|/(Na|lns|) exactly, so the per-image loss a code needs is |lnP|/(rNa) — linear in the number of ancillas and in the number of rounds, with no threshold behavior anywhere. CT·WS 24

Temperature is “five microkelvin”. Worked Example 26.1’s stage-1 occupations nr=0.64 and nz=5.0 are not the values at 4.92 and 4.79 μK but those at its own stated 5 μK, 0.636 and 5.03. The stated temperature is used here. CT·WS 26

There is a hard ceiling and the fiber cavity is nowhere near it. At the diffraction-limited waist w0=λ/2, Cmax=12F/π3=0.3870 F, which is 1.94×104 at F=5×104 — seventy-three times the fiber cavity’s 264, and exactly (3.34/0.390)2. CT·WS 28

Three-body coefficient. (15.20) is L3=Cℏa4/M with C=10.0 for rubidium; Worked Example 15.4 is L3=3Cℏa4/M with “C between about 0 and 70”. Both are used below and both are labeled; they are consistent only if C(15.20)=3CWE. CT·WS 15

Threshold pair. §25.16 and §29.3 both use pthP=1 % and pthE=4 %. Chapter 25’s hint evaluates the harmonic mean at f=0.98 as 3.77; Chapter 29’s table says 3.8. The exact value is 3.7736, and both are right. CT·WS 29

Total one-dimensional displacement. With atoms at mean spacing 1/p and targets at spacing one, the j-th displacement is (1/p-1)(j-Nt/2) once the block offset is chosen optimally, so. CT·WS 24

Transmission contrast. Weak-drive linear response of the Jaynes–Cummings model gives a=ε/[(κ+iδc)+g02/(Γ/2+iδa)], so on resonance a/aempty=1/(1+2C) and the transmitted intensity is (1+2C)-2 exactly; the Liouvillian steady state in a truncated Fock space returns the same number. CT·WS 28

Trap-loss cross-sections. Only a collision transferring more than the trap depth causes loss, so σtot is depth-dependent. The value implied by Worked Example 14.2’s own table, 3.0×10-18 m2, is used here and reproduces it exactly. CT·WS 14

Trapezoid. For acceleration ±a over the outer fractions β,. CT·WS 24

Truncated-Boltzmann evaporation rate. An atom evaporates only on a small fraction of elastic collisions, f(η)=(η-4)e-η, which is 1.34×10-3 at η=8. Every ramp-duration estimate below uses it, and Problem 16.2(c) shows that no estimate without it can be right. CT·WS 16

Tunneling time. Worked Example 21.2’s “a tunneling time is 44 ms” is h/t. Worked Example 21.1’s adiabaticity criterion is ℏ/t=7 ms. Section 21.3 applies Worked Example 21.1’s criterion and quotes 44 ms. The two differ by 2π; both appear, and both are reported. CT·WS 21

Tweezer formula. (20.19) as printed keeps only the D2 line in the rotating-wave approximation. ctna/dipole.py keeps D1, D2 and the counter-rotating terms. Both are evaluated below; they differ by 28% at 815 nm, and the two-line value gives the quoted 3.3 mK. CT·WS 20

Two free-space photon numbers. §28.9’s 2.3×103 (from §22.13) is the minimum for a hundred photoelectrons at η=4.3 %; Worked Example 28.1’s 7.2×103 (from §22.15) is what a 20 ms exposure actually scatters. CT·WS 28

Two lattice orientations with exact answers. With the field along (1,1,1) of the lattice axes and normal, both nearest-neighbor bonds of a square array are exactly zero and the two diagonals are exactly ∓1 — the lattice separates into two decoupled sublattices with equal and opposite couplings. CT·WS 30

Two molecules in one trap. For both in the motional ground state the pair overlap is (2π)-3/2(ar2az)-1, so the universal loss rate is K2/(2π)3/2ar2az — 1.3×105 s-1 for NaCs in a 1 mK, 1 μm tweezer, a lifetime of 7.9 μs. CT·WS 30

Two recoil temperatures. Both are in circulation and the chapter uses both. Write Er=ℏ2k2/2M for the recoil energy; then Er/kB=0.181 μK for 87Rb, which is the figure quoted in Section 9.7. CT·WS 9

Two-photon Doppler ratio. keffco/keffcounter=(λ2+λ1)/(λ2-λ1)=1433/593=2.4165, so the co-propagating geometry costs exactly 5.840 in Doppler error. CT·WS 27

Uniform-gradient trade. The 1%-linear half-width is 0.0969R at d=R/2 and 0.3766R at d=√3R/2 — a factor of 3.89, and 59 in volume, for a 25% loss of gradient. CT·WS 19

Uniformity. (23.1) is one minus the Michelson contrast of the depth distribution. CT·WS 23

Vacuum requirement. τ>-Ntc/lnP, which is 9.4912 Ntc at P=0.9; inverted, Nmax=τ|lnP|/tc. CT·WS 24

Van der Waals coefficient. Chapter 27 uses C6/h=820 GHz μm6 for 87Rb 70S; Chapter 20’s (20.20) uses 862 for the same state. Every Chapter 27 number below is reproduced with 820, and 862 is carried alongside so the two chapters can be compared row by row. CT·WS 27

Velocity. vmp=√(2kBT/M) is the most probable speed in a gas; √(3kBT/M) is the most probable speed in an effusive beam. Problem 7.9 uses the first, Problem 6.2 used the second. CT·WS 7

Waveguide and the cavity. Equating t=1-2β with 1/(1+2C) gives. CT·WS 28

Worked Example 19.4’s “. ϱ=2.6”. The quoted N, T and ω give ϱ=27.6, and at T=Tc the same formula returns 1.20206 — ζ(3) exactly. CT·WS 19

Worked Example 22.2’s. 0.540. By symmetry P(X≥N/2)=1/2+1/2(N/N/2)2-N, which for N=100 is 0.539795. CT·WS 22

Zeeman slope of a clock qubit. dν/dB=2K2B with K2=575.15 Hz G-2, and 1 G=100 μT. At the chapter’s 1 mT=10 G bias that is 11.50 kHz G-1=115.0 Hz μT-1. This is the chapter’s 115; both it and the first printing’s 0.115 are computed throughout. CT·WS 25

Zeros. Section 24.4 and Appendix B place the zeros of the bang-bang spectrum at ωT=4πn. The transform is 16dsin2(ωT/4)/ωT2, whose zeros are at ωT=4πn. The chapter’s own tabulated column is reproduced exactly by this expression. CT·WS 24

“the band gap” means. Everywhere in Chapter 21 the phrase denotes the on-site harmonic level spacing ℏωsite=2√s Er — 14.3 kHz at s=12.39, 12.8 kHz at s=10. The true lowest band gap, from the Mathieu spectrum, is 11.03 kHz and 9.27 kHz: smaller by 23% and 28%. CT·WS 21

Cold-Atom Quantum Simulation — เฉลยละเอียด · สัญลักษณ์และข้อตกลง 53

Angular units. Ω and V/ℏ are angular frequencies; Ω/2π and V/h are ordinary frequencies in hertz. The dimensionless ratio written V/Ω always means V/ℏΩ, which is the same number as (V/h)/(Ω/2π). Equation (11.9) is written with ℏΩ and Appendix C is tabulated with h×(Ω/2π); these are the same quantity and the appendix is consistent with the equation. QS·WS 11

Charges, links and violation vectors. Matter sites are n=0…N-1 with the staggered charge Qn=(σ̂nz+(-1)n)/2, so an even site carries Q∈{0,+1} and an odd site Q∈{-1,0}. Links carry Sn,n+1z=±1/2, with the open-boundary convention S-1,0z=SN-1,Nz=-1/2 that makes the bare vacuum gauge invariant. A violation vector g=(g0,…,gN-1) is the list of eigenvalues of the Ĝn in a given unphysical configuration; the physical sector is g=0. QS·WS 19

Effective quantum numbers. n*=n-δnlj throughout, with δ evaluated at the n in question from the two-term Rydberg–Ritz expansion, not at the series limit. The difference matters only in the sixth digit for δ0 but it is the whole of δ2, which Problem 11.1 measures. QS·WS 11

Energies and frequencies. A tunneling amplitude written J/h or t/h is a frequency in hertz; the same symbol without h is an energy. Lattice depths are s=V0/ER. Every lattice number in this document — tunneling, band gap, on-site interaction — comes from the exact Mathieu spectrum through qsim_lattice, never from a Gaussian approximation, and reproduces the chapter’s own tables to the digits printed there. QS·WS 18

Entropy is per site, and per particle only when the two differ. At half filling in a Mott insulator they are the same number. Away from half filling they are not, and the conversion is stated wherever it is used. Temperatures are quoted as kBT/J for a spin model and as T/t for a Hubbard model, with Jex=4t2/U the superexchange that connects them. QS·WS 31

Entropy, not temperature, is the controlled variable. This is the single methodological choice that separates these solutions from a textbook treatment. A lattice is loaded adiabatically, so the experiment fixes s and the Hamiltonian decides T. Several of the chapter’s statements — including the recommendation to run at U/t=8 — are true at fixed temperature and change when the constraint is put where the apparatus actually puts it. Problem 17.4 makes this quantitative. QS·WS 17

Estimators are part of the physics. A light-cone velocity, a scatter exponent and a level-spacing ratio are all estimates extracted from data by a procedure, and in this chapter the procedure often matters more than the Hamiltonian. Each solution therefore states the estimator explicitly — which contour, which spectral window, which fit constraint — and, where it matters, shows how much of the reported effect is the estimator rather than the physics. QS·WS 24

Exact results are benchmarks, not decorations. The Bethe value J(1/4-ln2), Wannier’s residual entropy, the Majumdar–Ghosh energy and the Johnson–Krinsky–McCoy gap are used here as calibration standards: each solution states the exact number first and then asks how close a finite calculation, or a finite experiment, can get to it and at what cost. QS·WS 16

Exact where exact is possible. Four results in this chapter that the chapter obtains by fitting are available in closed form, and the closed forms are derived here: the Liouvillian gap of the preparation scheme, the uniqueness of its dark state, the parent-Hamiltonian gap that governs the adiabatic alternative, and the exceptional point of the damped two-level system. A fit that agrees with a closed form is a check; a fit quoted in place of one is a missed opportunity. QS·WS 25

Finite size is stated, not hidden. The largest exact calculation here is a 20-site chain; the largest experiment in the chapter is a 51-atom array. Every extrapolation to the thermodynamic limit is labeled as an extrapolation, and several are labeled as unjustified. QS·WS 24

Frequencies. Ω and Δ are angular quantities; νΩ=Ω/2π and νV=V/h are ordinary frequencies in hertz. A ratio written V/Ω always means νV/νΩ, which is the same number. Interaction energies are quoted as V/h in megahertz, following the chapter. QS·WS 13

Gate counting. A Pauli exponential exp(-iθP̂) of weight w is compiled as a CNOT ladder in, a single R̂z(2θ), and a ladder out, costing 2(w-1) two-qubit gates. This convention is used consistently in Problem 2.2 and again in Chapter 20. QS·WS 2

Gate counting. A Pauli exponential exp(-iθP̂) of weight w costs 2(w-1) CNOTs when compiled as a parity ladder, one R̂z, and the inverse ladder. A fermionic hopping term ĉi†ĉj+h.c. produces two Pauli strings of the same weight that share a ladder, so the pair costs 2w-1 rather than 4(w-1); this is the convention of Figure 21.3 and of the closed-form count in Problem 21.7, while Problems 21.7(c), 21.8 and 21.10 each state the per-gate cost they use instead. A fermionic swap costs three CNOTs, a Toffoli six. QS·WS 21

Gate errors are quoted as process infidelities. The coefficients Cd=0.383 and Cb=0.763 of Appendix B are 1-Fproc for the two-qubit map, and 1-Fproc=5/4(1-Favg) in dimension four. Where an average gate fidelity is wanted the conversion is stated explicitly. All Rabi frequencies written ν are ordinary frequencies in hertz and all written Ω are angular, with Ω=2πν; the blockade V always means V/h in hertz, so that νRabi/V is the dimensionless ratio the budget uses. QS·WS 28

Hubbard units throughout. The tunneling t is the unit of energy, temperatures are T/t with kB=1 inside any formula, entropies are s per site in units of kB, and the doping is δ=1-n. Absolute frequencies and nanokelvin appear only when a laboratory is meant, and they are always converted from the exact band-structure values of §6.9, never from the Gaussian approximation. QS·WS 17

Integers are computed twice. Every Chern number quoted below is obtained once from the Diophantine equation (18.14) and once from an independent Fukui–Hatsugai plaquette sum over the magnetic Brillouin zone. Where the two disagree the disagreement is reported; they never do, but the check is the point. QS·WS 18

Layer, round, shot. A layer is one parallel application of N/2 two-qubit gates; a round is one surface-code syndrome extraction, which is four layers plus a measurement; a shot is one execution of a circuit from array load to final image. These three words are used in exactly these senses throughout, and most of the chapter’s surprises come from the ratios between them. QS·WS 20

Lengths, and which oscillator length. Two different transverse lengths appear in this chapter and they differ by √2. The single-atom oscillator length is aho=√(ℏ/Mω); the relative-motion oscillator length, built from the reduced mass μ=M/2, is ar=√(ℏ/μω)=√2 aho. Two-body scattering problems are solved in the relative coordinate and therefore know only ar. Problem 10.7 shows where the distinction bites: Olshanii’s matching condition is naturally written in ar, and converting it to the a⊥=aho that §10.7 uses is exactly what puts the √2 into C=-ζ(1/2)/√2. Equation (10.9), by contrast, lives in aho from the start. Every solution below states which length it is using. QS·WS 10

Motional quantities are quantum, not classical. The position spread of a trapped atom is δr=x0√(2n+1) and its velocity spread is Δv=ω δr, not √(kBT/M). Near the ground state the two differ by thirty percent, in the direction that matters, and Problem 7.9 works out what that does to an atom released for a microsecond. QS·WS 7

Notation. N is the atom number, n0 the peak density, ω=(ωxωyωz)1/3, η=Utrap/kBT the truncation parameter, γel=n0σv/√2 with σ=8πa2 and v=√(8kBT/πM). Lattice depth is s=V0/ER; where the entropy symbol would collide with it, the depth is written V0/ER in full and the entropy keeps s. For 87Rb at 1064 nm, ER/h=2.028 kHz and ℏ/ER=78.49 μs, and these two numbers are used throughout. QS·WS 5

Notation. Magnetic moments are in Bohr magnetons, μ=gJJμB; electric dipoles in debye. The reduced field of a rigid rotor is x≡dE/Brot, and ⟨d⟩0, ⟨d⟩1 and d01 denote ⟨0|dcosθ|0⟩, ⟨1|dcosθ|1⟩ and ⟨0|dcosθ|1⟩ for the two lowest M=0 eigenstates. Cavity quantities are the single-atom coupling g, the cavity energy decay rate κ, the atomic linewidth Γ, and C1=4g2/κΓ. Equations carry three-part numbers: (14.4.2) is the second numbered equation of the solution to Problem 14.4. QS·WS 14

Notation. Czzc(d,t) is the connected equal-time correlator, S(t) the half-chain von Neumann entropy in natural units, ⟨r⟩ the mean level-spacing ratio (24.3), W the disorder strength, λ the integrability-breaking field, α the interaction exponent, ξ the Lieb–Robinson decay length in lattice sites, and Ω the PXP Rabi frequency. Equations carry three-part numbers: (24.4.2) is the second numbered equation of the solution to Problem 24.4. QS·WS 24

Notation and units. The notation is the chapter’s. Fine-structure states are n2S+1LJ, hyperfine states |F,mF⟩. Magnetic fields are quoted in gauss because the literature is, with 1 G=10-4 T and μB/h=1.399 624 MHz G-1. Polarizabilities are in atomic units, 1 a.u.=4πε0a03=1.648 777×10-41 C m2 V-1; scattering lengths in Bohr radii, a0=0.052 917 721 nm; intensities in W m-2 with mW cm-2 alongside, 1 mW cm-2=10 W m-2. Two recoil energies occur in this chapter and must never be confused: the photon recoil ERph=ℏ2kD2/2M built on the D2 wave vector, which is h×3.771 kHz for 87Rb and governs cooling; and the lattice recoil ER=h2/2Mλ2 built on the trap wavelength, which is h×2.028 kHz at 1064 nm and governs the Hubbard model. Equations carry three-part numbers: (3.4.2) is the second numbered equation of the solution to Problem 3.4. QS·WS 3

Notation and units. The notation is the chapter’s. Spin operators are Ŝα=1/2σ̂α, so a bare σ̂ carries a factor of two and every factor of two below is tracked. The itinerant tunneling is t, the spin exchange is J, the superexchange derived from an itinerant model is Jex=4t2/U, and the lattice coordination number is z — six for the cubic lattice, four for the square, two for the chain. Temperatures are quoted as ratios T/t or T/J and, where a species is named, in nanokelvin with kB/h=20.837 GHz K-1, so that h×1 kHz is 48.0 nK. Lattice depths are in recoil units V0/ER with ER=h2/2Mλ2. Entropies are per particle in units of kB. Equations carry three-part numbers: (4.4.2) is the second numbered equation of the solution to Problem 4.4. QS·WS 4

Operators and units. Hats denote operators throughout. Where ℏ would clutter an algebraic step it is set to unity and restored in the final result; every numerical evaluation carries its units explicitly. Couplings quoted as J/h or V/h are in hertz; the same symbol without h is an energy. QS·WS 2

Phases are directed. A Peierls phase θij belongs to a directed bond: it is the phase multiplying âi†âj, and the reverse bond carries -θij because the Hamiltonian is Hermitian. Every plaquette sum in this document is a directed sum around a closed loop taken counterclockwise. Half the confusion in this subject comes from writing a phase that is symmetric under i↔j, which no Peierls phase can be; Problem 18.1 is exactly that trap, and the chapter walks into it. QS·WS 18

Rates and their normalization. The engineered dissipation has rate Γ and the parasitic dissipation rate γ; both are true rates in s-1 and neither is ever divided by 2π, because a Lindblad rate is not a frequency. Γ is the coefficient absorbed into the jump operator, so L̂i†L̂i=4|A⟩⟨A| for the two-site operator (25.3) and the antisymmetric orbital is emptied at 4Γ rather than at Γ. Atomic linewidths and Rabi frequencies keep the opposite convention and are written Γe/2π and Ω/2π wherever they appear, so that the two never collide. The two-site operators carry no explicit prefactor in the chapter’s numerics; Γ=1 there, and every dimensionless time in this document is Γt. QS·WS 25

Shots and observables. M always denotes a shot count and Mobs the number of observables to be estimated; the chapter uses M for both, and the collision is the source of more than one misreading. Absolute precision is ϵ; relative precision on a signal of size C is η, so that ϵ=ηC. A duty cycle of 200 ms per shot is assumed throughout, which is the reload-dominated figure of. QS·WS 22

Spins, couplings and signs. Spin operators are Ŝα=1/2σ̂α throughout, so a single bond of the Heisenberg chain has Ŝi⋅Ŝj with eigenvalues +1/4 (triplet) and -3/4 (singlet). In every Heisenberg expression J>0 is antiferromagnetic; in every Ising expression the sign is stated explicitly at the point of use, because the chapter uses both conventions. Energies are quoted in units of J and lengths in lattice sites. QS·WS 16

The depth convention. Throughout, V(x)=V0cos2(kx), s≡V0/ER, d=λ/2, k=π/d, and ER=ℏ2k2/2M. Quasimomentum q is in units of ℏk, so the zone boundary is q=1. Every energy quoted in recoil units is a pure function of s; every energy quoted in hertz carries a species and a wavelength. QS·WS 6

The distinction between measuring the machine and verifying the answer is enforced throughout. Problems 29.4 and 29.5 concern protocols whose cost does not grow with the number of qubits; Problems 29.6, 29.7, 29.10 and 29.11 concern protocols whose cost does. The first group calibrates hardware, the second bounds or measures a physical claim, and a solution that blurs them has answered a different question from the one asked. Digital and analog machines are kept separate for the same reason: randomized and cross-entropy benchmarking are gate-model tools with no analog counterpart. QS·WS 29

The fidelity metric. Equation (13.2) is used throughout, so that every number here is directly comparable with the chapter and with code/ch13.py. Problem 13.1 shows that (13.2) is not the maximum over single-qubit Z rotations, and that the difference is exactly a factor of three for a pure conditional-phase error — a fact worth knowing before comparing two papers’ fidelities. QS·WS 13

The figure of merit. Equation (10.6) defines Q=(gn/h)/(L3n2), the interaction time being read as h/gn in the convention fixed in §9.1 and used throughout this book. Reading it instead as ℏ/gn multiplies every Q by 2π; where that reading would change the look of a conclusion it is quoted alongside, with the convention named, but the unqualified symbol Q always means the h version. QS·WS 10

The gate is a propagator, and every coefficient in this document was measured from it. Chapter 13 quotes two coefficients that control everything downstream: the blockade leakage ϵb=Cb(Ω/V)2 and the Rydberg decay ϵd=c tg/τ. Neither is a matter of taste; both are properties of the optimized two-pulse propagator and both are computed here directly, by evaluating (13.2) on the exact two-atom evolution. The values are Cb=0.763 and c=0.383, and Problem 13.4 shows how completely those two numbers determine the physics floor. QS·WS 13

The Levine–Pichler gate is the unit of time. Wherever a “gate” appears it is the two-pulse global CZ of §13.3, with Ωτp=4.292682 per pulse, ΔLP/Ω=0.377371 and an inter-pulse phase jump ξ=3.902422 rad, so the total duration is T=8.5854/Ω and the pulse area seen by a single atom is 8.5854 rad. Equations carry three-part numbers: (27.4.2) is the second numbered equation of the solution to Problem 27.4. QS·WS 27

The perturbation. Throughout, δJ/J≡ϵ denotes the root-mean-square fractional width of a zero-mean Gaussian distribution applied independently to every coupling, unless a correlated positional origin is stated. Disorder averages are written ⋅ and quantum expectations ⟨⋅⟩; the quantity plotted in Figure 15.1 is |Δ⟨m̂z(t)⟩|, the disorder average of the absolute error, which is not the same as the absolute value of the disorder average and behaves differently at long times. QS·WS 15

The ramp convention. The running example is the transverse-field Ising chain of the chapter, an open chain of L spins with. QS·WS 22

The residual is part of the answer. Chapter 9 defines an engineered Hamiltonian as a leading term plus an explicit accounting of what else is there. Every solution below that produces an effective Hamiltonian also produces the next order, states what operator it is, and evaluates it numerically. A part that stops at the leading term is not finished. QS·WS 9

The truncation penalty is carried explicitly. The chapter multiplies the diffraction-limited waist by 1.4 to allow for aperture truncation, residual aberration and the vacuum viewport. That factor changes w0, and therefore changes zR, the aspect ratio, both trap frequencies and the power per trap. Formulas are quoted in their ideal form and evaluated with the penalty, and which is which is always said. QS·WS 7

The working parameters are Appendix A, and they are used without apology. t1q=1 μs, t2q=0.4 μs, tmove=200 μs, tmeas=1 ms, treload=200 ms, p2q=5×10-3, p1q=10-4, ploss=10-3, T2=4 s. They are conservative by construction; where a conclusion depends on that conservatism, the solution says so and recomputes at the optimistic value. QS·WS 20

Two budgets, two currencies, and they must not be mixed. The digital budget is a dimensionless infidelity per two-qubit gate and it adds. The analog budget is a fractional coupling error δJ/J and it adds in quadrature, because its entries are independent random contributions to one number rather than independent ways of losing a gate. The first becomes a code distance through §20.9; the second becomes a validity horizon through (15.2). A sentence that converts one into the other without saying how is the commonest error in this subject. QS·WS 28

Two clocks, never mixed. Times internal to the model are quoted in units of ℏ/J; times internal to the apparatus are quoted in seconds. The conversion is ℏ/J=1/(2π J/h), so an interaction time at J/h=5 MHz is 31.8 ns and not 200 ns. The factor of 2π is the commonest arithmetic slip in this chapter’s subject matter and Problem 15.2 shows what it is worth. QS·WS 15

Two currencies for one error. Gate infidelity p2q is a probability; a miscalibrated analog block produces a coherent amplitude. These are not the same object, and several of the chapter’s headline factors depend on which is used. Problem 23.1 derives (23.2) in the chapter’s currency and then again in infidelity, and the two crossovers are compared; Problem 23.2 does the same for accumulation over many steps. Where a result is convention-dependent it is quoted in both conventions and the convention is named. QS·WS 23

Two currencies, kept apart. Digital claims are denominated in a Toffoli count, a code distance d, a physical-qubit count nphys and a wall-clock time; analog claims are denominated in an atom number N and a reachable dimensionless time Jt/ℏ=J/δJ. Nothing converts between them, and a comparison that appears to do so — as in Problem 30.6 — is a comparison of two independent estimates, not a conversion. QS·WS 30

Two currencies, kept separate. Product-formula costs are CNOT counts. Block-encoding costs are naturally Toffoli counts, and the chapter’s cBE≈4Γ is one. Where the two are compared in the same figure the Toffoli count must be converted, and Problem 21.5 does the conversion explicitly and shows what it does to the crossovers. QS·WS 21

Two currencies, never mixed. Phase-space density ϱ=nλdB3 scores the cooling chain up to the point where the gas becomes degenerate; entropy per particle s, in units of kB, scores everything after it. The reason for the change of currency is that lattice loading is adiabatic, so it conserves s and changes T, n and ϱ beyond recognition. Where a problem asks for one and the physics is naturally stated in the other, the conversion is done explicitly. QS·WS 5

Two formulations, used for two different purposes. All the physics computed here — pair production, the static potential, string breaking, the entanglement rate — uses the gauge-eliminated spin chain (19.4), in which Gauss’s law is an identity and gauge invariance cannot be violated even in principle. All the error computations — Gauss-law leakage, protection, compliance — use the U(1) quantum link model with explicit spin-1/2 links, in which Gauss’s law is a constraint that a bad term can break. Mixing the two would make either question meaningless. The modules are qsim_lgt.SchwingerSpin and qsim_lgt.QLM, and every number below was recomputed from them. QS·WS 19

Two frequency conventions, kept apart. Ω is always an angular Rabi frequency in rad s-1 and νRabi=Ω/2π is always the ordinary frequency quoted in megahertz; δν is a full width at half maximum in hertz, and Sϕ(f) is a one-sided phase-noise power spectral density in rad2 Hz-1, related to the single-sideband quantity by L(f)=Sϕ(f)/2. Polarizabilities are quoted in MHz (V cm-1)-2 in the convention ΔE/h=-1/2α0F2, so a positive α0 means a downward shift. QS·WS 27

Two radii, one window. rc=(C6/|ΔF|)1/6 bounds every design from below and Rb=(C6/ℏΩ)1/6 from above. Wherever an interaction is quoted inside 1.6 rc the exact two-channel value from (11.7) is given alongside the C6/r6 extrapolation, and the two are never conflated. QS·WS 11

Units. Couplings written V/h or J/h are frequencies in hertz; the same symbol without h is an energy. Rabi frequencies and detunings written Ω/2π and Δ/2π are in hertz, and Ω, Δ alone are angular frequencies. The figure of merit is Q=Vτ/h throughout, the number of interaction times available before decoherence ends the simulation. QS·WS 9

Units and the three parameters. Every energy is quoted in units of the correlated hopping w and every time in units of ℏ/w. The three parameters of the lattice Schwinger model are the hopping w=1/2a, the staggered fermion mass m and the electric coupling J=g2a/2; the single dimensionless combination that fixes the lattice is J/w=(ga)2=1/x of (19.5). Where a laboratory number appears it carries hertz explicitly, and w/h means a frequency. QS·WS 19

Units are mixed deliberately. Vacuum is quoted in torr and magnetic fields in gauss because that is how the components are sold; the SI equivalents are 1 Torr=133.322 Pa and 1 G=10-4 T. Lattice depths are in recoils, s=V0/ER, with ER=h2/2Mλ2 the recoil energy of the lattice light, and every absolute energy is quoted as a frequency by dividing by h. QS·WS 26

“Of your choosing” is answered by choosing. Problems 30.3, 30.5, 30.10, 30.11 and 30.12 ask the reader to supply a molecule, a gauge-theory row, an apparatus, a roadmap and a program. Each is answered here by working one named example all the way to a number — the oxygen-evolving complex of photosystem II, real-time string breaking in the lattice Schwinger model, a hundred-atom 87Rb tweezer array at 6 μm spacing, a published superconducting roadmap, and a three-stage upgrade ladder for that array. The example stands in for the reader’s own choice; the procedure is what transfers, and it is stated separately from the arithmetic in every case. QS·WS 30

Atom-Based Quantum Sensing and Metrology — เฉลยละเอียด · สัญลักษณ์และข้อตกลง 29

2D-MOT source. A vapor cell in which two-dimensional cooling and a push beam produce a slow, bright atomic beam through a differential-pumping aperture; see §11.3 of the guidebook. SM·WS 36

An assumed field degradation stays marked assumed. Six of the thirteen rows of the §38.5 register carry a field value that nobody has measured: the transportable gravimeter at ×3, the ring navigator at ×2.5, the free-space gyroscope at ×10, the SERF magnetometer at ×4 and the Rydberg electrometer at ×5. Every table below that uses one of these repeats the mark, because an assumed number is not evidence and a table that hides the distinction is worse than no table. SM·WS 38

Binomial weights. σP=√(P(1-P)/N) is the exact binomial standard deviation of one normalized population, and it is a function of the readout phase: with P=1/2[1-Ccosψ] one has P(1-P)=1/4(1-C2cos2ψ), so σP runs from 5.000×10-4 at mid-fringe down to 4.583×10-4 at the extrema for C=0.4 and N=106, a variation of 8.3 %. “Equal weights” means the mid-fringe value used at every point. The distinction is the whole of Problem 34.1(b). SM·WS 34

Conventions, stated once because two of them are contested. The collective spin is Ĵ=∑iŝ(i) with J=N/2, and |J,m⟩ are the eigenstates of Ĵz. The contrast is C=2|⟨Ĵ⟩|/N, the Kitagawa–Ueda parameter is ξK2=4Var(Ĵ⊥,min)/N of (8.10), the Wineland parameter is ξR2=ξK2/C2 of (8.14), and the metrological gain is Gm=-10log10ξR2. The twisting angle is μ=χt for Ĥ=ℏχĴz2, the convention fixed in §8.4; Kitagawa and Ueda’s own variable is μKU=2χt, and every “optimal twisting angle” in the literature must be read with its convention attached. In this convention the optimum is μopt=31/6N-2/3=1.2009 N-2/3, which is what qsm_metrology.oat_optimal returns; the value 241/6=1.6984 that is sometimes quoted is μKU,opt/√2 and belongs to neither convention. The cavity field decay rate is κ=πc/(2LF), half the intensity full width, and the chapter’s single-atom cooperativity C=2g2/κΓ is defined with that κ and that Γ; Problem 8.5 derives it from g2, κ and Γ explicitly and shows that the standard literature value is the same. SM·WS 8

Machines and conversions. The fountain M1 is N=106, T=100 ms, C=0.4, Tc=1.5 s, keff=1.610 575×107 m-1, so keffT2=1.610 575×105 rad/(m s-2), one milliradian is 0.6209 μGal and one microgal is 1.6106 mrad; its realized per-shot scatter is 9.900/√(1.5)=8.083 μGal from row 23 of Table 33.3. The ring navigator M2 is NARW=7.355×10-7 rad/√s, bias instability 10-8 rad/s at τ=104 s, rate random walk K=5.5×10-11 rad s-1/√s, Tc=1 s. The SERF head M3 is 10 fT/√(Hz) with a half width ΔB1/2=22.10 nT. For the one-meter gradiometer keffdT2=1.610 575×105 rad/s-2, so one milliradian of differential phase is 6.209 E with 1 E=10-9 s-2. Gravity at Chiang Mai is g=9.7847 m s-2 and ΩE=7.292 1159×10-5 rad/s at latitude 18.7883∘. SM·WS 34

Notation. The notation is that of the chapter. θ is a generic measurand and V̂(θ) its coupling term in Ĥs=Ĥ0+V̂(θ); ∂ϕ/∂θ is the scale factor. For the inertial sensors, keff=4π/λD2=1.610 575×107 m-1 for counter-propagating Raman beams on the 87Rb D2 line, vr=ℏkeff/M=11.7691 mm/s, T is the pulse spacing of the three-pulse Mach–Zehnder sequence, gCM=9.7847 m s-2 is the book’s baseline gravity at the RCQT laboratory (18.7883∘ N, 98.9853∘ E, 310 m), Γzz=3086 E is the standard free-air gradient with 1 E=10-9 s-2, 1 μGal=10-8 m s-2, and ΩE=7.292 115 9×10-5 rad/s. For the field sensors, γ=gFμB/ℏ, Q is a current dipole in A m, d a Rydberg transition dipole and Ω=dE/ℏ the Rabi frequency it produces. Thermodynamic temperature is written T to distinguish it from the interrogation time. The reference machines are those of the chapter: the fountain gravimeter (N=106, C=0.4, T=100 ms, Tc=1.5 s), the ring navigator (R=1 mm, N=105, C=0.5, one shot per second) and the SERF cell (n=1020 m-3, V=1.25×10-7 m3, T2=10 ms). SM·WS 4

Notation. The notation is that of the chapter. Ω1 and Ω2 are the single-beam Rabi frequencies, defined throughout by Ωj=Γ√(Ij/2Isat) with Γ=2π×6.0666 MHz and Isat=16.693 37 W/m2 the cycling saturation intensity of 87Rb; Δ is the one-photon detuning of beam 1 from the |F=1⟩→|F'=2⟩ transition of the D2 line, negative for red detuning and quoted as |Δ|; δ is the two-photon detuning; Ωeff=κ Ω1Ω2/2Δ is the two-photon Rabi frequency with κ the Clebsch–Gordan factor of §7.1; ΩAC is the differential light shift; ωhfs=2π×6.834 682 610 9 GHz. For the Bragg lattice V(z)=sErcos2(kz) with Er=ℏ2k2/2M, ωr=Er/ℏ=2π×3.7710 kHz at 780.241 nm, ΩL=s ωr/2 and β=s/16; Ωn is the n-th-order Bragg Rabi frequency, τB the Bloch period, ηp the single-pulse efficiency, and Sφ(f) the one-sided phase-noise spectral density of the Raman difference frequency. Vectors are bold, operators carry hats, and q denotes both a spherical polarization index and a quasimomentum in units of ℏk; context separates them without ambiguity. SM·WS 7

Notation. That of the chapter, and one convention is repeated because it matters on every page: η is the Eötvös parameter of (4.51) and nothing else; a noise-equivalent measurand per root hertz is written out in full. Γzz=3.086×10-6 s-2 is the free-air vertical gravity gradient; g=9.7847 m s-2 and ΩN=6.904×10-5 rad/s are the Chiang Mai baseline values; keff=1.610 575×107 m-1 for 87Rb two-photon Raman or Bragg diffraction and k=8.995 954×106 m-1 on the 87Sr clock line. n is the large-momentum-transfer order, N the atom number detected per shot, C the contrast, T the interrogation time and Tc the cycle time. de, dme and dg are the dilaton couplings; fφ=mφc2/h and κφ0=1.741×10-16/(fφ/Hz); τcoh=106/fφ, M is the number of coherence patches and Rstoch the Rayleigh penalty. h(f) is a strain amplitude spectral density and L a gradiometer baseline. SM·WS 26

Notation. That of the chapter. Sa(f) is the one-sided acceleration power spectral density in m2s-4/Hz and √(Sa) its amplitude density in m s-2/√(Hz); |Ha(ω)|2=keff2|H(ω)|2/ω4 is the acceleration transfer function of (2.45) and |H|2=16sin4(ωT/2) its short-pulse limit; T(f) is transmissibility and ζ=1/2Q; W(f) is a random-vibration profile in g2/Hz and grms its root-mean-square in multiples of gn=9.80665 m s-2; θtilt is tilt and θ0=45.2 μrad the null; αT a thermal expansion coefficient and Rth a thermal resistance; λf a failure rate, MTBF, MTTR and Av the reliability triple; τvac, σbg, KHe and ΦHe the vacuum quantities; mI, VI, PI the instrument’s mass, volume and power. SM·WS 36

Numbering. Equations carry three-part numbers: (4.3.2) is the second numbered equation of the solution to Problem 4.3. Equations of the chapter are cited by their chapter numbers, (4.15), (4.59), and so on; sections as §4.10; the sibling volumes as CT and QS. SM·WS 4

One convention is binding on every shield number in this document, and it is the chapter’s own. Chapter 29 owns the package, and §29.5’s expression BJ=Gμ0√(σckBTt)/a contains three geometric quantities, of which a enters as 1/a and t as √t. A shield-noise figure quoted without (a,t,σc) is therefore not a number at all; it is a number and an unstated geometry. Every value in these solutions carries its geometry with it, and Problem 29.5’s Verification block reconciles the four different stacks that this book quotes — 19.88, 28.72, 25.63 and 17.08 fT/√(Hz) — as four geometries and not four estimates of one. SM·WS 29

Signs and units. Δ is the one-photon Raman detuning, taken as a magnitude wherever an intensity, a power or a scattering loss is computed and as a signed quantity (negative to the red of the |F=1⟩→|F'=2⟩ transition of D2) wherever the hyperfine sum of Chapter 7 is called. A gauss is 10-4 T and a microgal is 10-8 m s-2; one eotvos is 10-9 s-2. Frequencies written ν are cyclic and measured in hertz, those written ω or Ω are angular and measured in rad/s, and a light shift quoted as “2π×518.2 Hz” is an angular frequency whose cyclic value is 518.2 Hz. The baseline gravity is the Chiang Mai value g=9.7847 m s-2 and the horizontal component of Earth’s rotation there is Ωh=6.903 56×10-5 rad/s. SM·WS 27

The decibel convention. s=10log10(e2r)=8.686 r is a variance ratio, so 3 dB is r=0.3454, 10 dB is r=1.1513 and 20 dB is r=2.3026. “Generated” always means the squeezing at the output face of the parametric oscillator, before any loss; “observed” always means what a homodyne detector at the end of a path of end-to-end efficiency η and rms local-oscillator phase jitter θ reports. The two differ by (25.14) and (25.16), and the whole engineering content of Part I of the chapter is the difference. SM·WS 25

The dipole convention. Throughout these solutions d denotes the sublevel matrix element. SM·WS 21

The equal-terms convention. δB√V takes V to be the sensing volume and never the package; for a point sensor the convention is stated with the number, and this chapter uses 1 nm3 for a single defect, as §24.10(ii) requires. Qf, the quantum-limited fraction of Chapter 38, is the demonstrated noise divided by the projection limit of the same device; for a superconducting loop the corresponding quantity is ε/ℏ in energy, or √(ε/ℏ) in amplitude, and the three are placed on one axis in Problem 24.11 and the Closing Note. SM·WS 24

The flux convention. Φ0=h/2e exactly, βL=2LI0/Φ0, VΦ in volts per flux quantum, SΦ a one-sided power spectral density in Wb2/Hz, and ε=SΦ/2L in units of ℏ. The reference device throughout is the chapter’s: I0=10 μA, L=100 pH, R=5 Ω, T=4.2 K, hence βL=0.9672. SM·WS 24

The fringe parameterization. The model (34.1) is written with the contrast C, but the Jacobian column quoted below it, -cos(φj+ϕ), is the derivative with respect to the half-contrast B=C/2. These solutions therefore take the parameter vector to be θ=(P0,B,ϕ,s) with B=C/2, which makes the four columns of J exactly the four the chapter lists. Nothing depends on the choice — a diagonal rescaling of one column rescales one row and one column of (JTWJ)-1 and leaves the ϕϕ element alone — but the two conventions give different off-diagonal elements, and the problem asks about those. SM·WS 34

The GUM discipline, binding on every uncertainty quoted below. This is the chapter in which the Guide to the Expression of Uncertainty in Measurement is applied rather than cited, so every uncertainty in these solutions carries three things explicitly and none of them is left to the reader to guess. First, its coverage factor: an uncertainty written u is a standard uncertainty, k=1; one written U is expanded, and the k that produced it is printed beside it. The book’s convention is k=2, and where a strict Student factor differs it is given with its degrees of freedom. Second, its degrees of freedom: a Type A uncertainty from n observations has ν=n-1; a Type B uncertainty carries the νi that Table 35.7 assigns it, and a quantity fixed by definition or convention carries ν=∞. Third, its correlation assumptions: every quadrature sum below states whether the terms are taken independent (r=0), fully correlated (r=1), or correlated with a stated rij, and where the assumption is debatable both values are computed. The arithmetic of every uncertainty in these solutions is reproducible from the numbers printed on the page. SM·WS 35

The gyromagnetic ratio. Throughout these solutions γe/2π=gμB/h=28.031 68 GHz/T with g=2.0028 (Felton et al. 2009), which is the chapter’s choice of (24.2); γe=2π×28.031 68 GHz/T=1.761 28×1011 rad s-1T-1 is the same quantity in angular units, and every sensitivity formula in which γe appears without a 2π takes the angular value. The free-electron value 28.024 95 GHz/T is used only where the 2.401×10-4 ambiguity between them is itself the point. SM·WS 24

The one-sided convention. Every spectral density in these solutions is one-sided, so that the vacuum quadrature variance is unity, a single-shot uncertainty and a density are related by S=2τ δ2, and the imprecision–back-action product is SximpSFba≥ℏ2 rather than ℏ2/4. A reader comparing with a source that symmetrizes and uses two-sided densities must multiply each density by two and the product by four, and the factor of four is the commonest reason two correct quantum-noise budgets disagree. SM·WS 25

The polarizability convention. α is defined by Δν=-1/2αE2, so the shift at 1 V/cm is α/2 and not α. Every “term of the sum” quoted below is a contribution to α, not to α/2. This matters immediately: §21.1 states that the largest single term of the sum at 50S1/2 is 149.0 MHz/(V/cm)2 and, in the same sentence, that the 50P term is +74.5 and the 49P term -60.4. The first is a contribution to α; the second pair are contributions to α/2. Problem 21.1 gives both columns. SM·WS 21

The quantum-limited fraction is always formed from an assembled budget. Qf=σtot/σp is meaningless unless the numerator and the denominator belong to the same machine under the same conditions, and §38.5 states the three conditions that make it well defined. Editorial ruling 10 fixes Qf=5.3, 1.4 and 156 for the fountain, the navigator and the SERF cell; those compare each machine’s target with its limit. Chapter 33 computes 5.21, 13.8, 105.9 and 270.5 from assembled budgets. Section 38.5 rules that the budget-based numbers govern, and this volume carries them everywhere. Problem 38.4 makes the ruling explicit, shows the one design decision it settles, and shows that for the fountain the two agree and for the navigator they differ by a factor of ten. SM·WS 38

The Rydberg row of the §38.5 register is corrected here. The register lists the Rydberg electrometer’s laboratory noise as 5×103 nV cm-1/√(Hz), hence Qf=2.41×104. That is the 2017 homodyne demonstration (Kumar, Fan, Kübler, Sheng and Shaffer, Oklahoma). The field’s record is the 2020 superheterodyne result of §21.5, 55 nV cm-1/√(Hz) (Jing, Hu, Ma, Zhang, Zhang and Jia), which against the same projection limit of 0.208 nV cm-1/√(Hz) gives Qf=265. The ranking column of §38.5 is understated by a factor of 91, that is by two orders of magnitude, and the corrected value is carried below. The conclusion of §21.14(vii) — that squeezing a Rydberg ensemble would buy nothing today — survives either value, and Problem 38.4 shows why in one line. SM·WS 38

The sensitivity convention. θn is a noise-equivalent measurand per root hertz, σ≥1 the readout noise factor, C the fluorescence contrast, N the number of contributing centers and T the free-precession time; tinit is the dead time per shot and every θn quoted at unity duty cycle sets it to zero. A figure quoted without its σ is not a specification, and none is quoted here without one. SM·WS 24

The status of a number is part of the number. Chapter 36 labels every figure it did not compute, and these solutions inherit those labels without exception. A row of Table 36.15 marked announced, vendor or unverified — the commercial optically pumped magnetometer figures in products since about 2016, the 2024–2025 quantum-navigation flight trials for which no peer-reviewed performance figure was found, AOSense’s products, CARIOQA’s programmatic status, and ACES/PHARAO, launched in 2025 with no published commissioning result — supports reasoning about what such a row would have to contain, and supports no quantitative conclusion. No solution here promotes one to a plain figure. The same applies to §36.8 and Appendix B: every qualification level in this document is a typical tailored value and must be verified against the current edition of the standard before it is written into a specification. The world unit counts of §37.11 are the author’s unverified estimate and are used nowhere in these solutions. SM·WS 36

The two signature compilations. This chapter owns two tables that the rest of the book is written against, and several of these problems are exercises in reading them rather than in deriving anything new. The control-knob table of §27.2, the seed of Compendium Table C.35, is read down its columns: group B and group C knobs multiply the phase and therefore enter δθ linearly or quadratically; group D knobs enter under a square root; group E knobs enter through the contrast and the estimator bias; group F knobs enter only the error budget. Problem 27.2 is that hierarchy made quantitative, and Problem 27.12 builds a new instance of the table from scratch. The failure table of §27.9, the seed of Compendium Table C.30, is read across its rows: every mechanism is an expansion, every expansion has a validity parameter εm, and every εm has a number at the reference machine. Problems 27.3, 27.4, 27.6, 27.7 and 27.10 each populate one row of it, and Problem 27.10 measures one rather than asserting it. SM·WS 27

Two effective wavevectors, and which is used here. Editorial ruling 1 fixes two definitions. For phase estimates and figures, keff=4π/λD2=1.610 575×107 m-1, with vr=ℏkeff/M=11.7691 mm/s and ℏkeff2/2M=2π×15.0839 kHz=4ωr. For accuracy work the scale factor is instead keff=2π(ν1+ν2)/c=1.610 561×107 m-1, smaller by 8.894×10-6, which on a fountain is 8.70 mGal. Chapter 7 is a chapter of phase estimates and device specifications, and §7.2 states explicitly that the accuracy-grade value belongs to Chapter 10; every solution below therefore uses 1.610 575×107 m-1, and each Verification block says so and states what the other choice would change. It changes nothing at the level of these problems: the largest gravity bias computed anywhere below is 6.20 μGal, and 8.894×10-6 of that is 5.5×10-5 μGal, four orders of magnitude inside the quoted precision. SM·WS 7

Which budget is the machine’s. Chapter 12 assembles twenty-five rows and reports ∑iui2=3.181 μGal2, two declared correlations and uc=1.97 μGal. Chapter 33 re-presents the same physics with one row Chapter 12 does not carry — the estimator bias of the four-parameter fringe fit, 0.050 μGal, which belongs to Chapter 34 — and with four declared correlations rather than two, giving ∑iui2=3.1832 μGal2, uc=1.9959 μGal and U=3.992 μGal at k=2. Section 35.7 rules that Chapter 33 owns the budget. These solutions therefore reproduce Chapter 12’s own arithmetic wherever a problem asks for it by name, and quote Chapter 33’s numbers whenever the question is what the machine’s total is; every Verification block that touches the total says which is which. SM·WS 12

β-separation line. The boundary Sν=8ln2 f/π2 separating the slow-modulation part of a frequency-noise spectrum, which broadens the line, from the fast-modulation part, which only feeds its wings; see §28.2 of the guidebook. SM·WS 28