---
title: 'Band-Schematic Pulses: Theory & Applications'
url: https://www.emergentmind.com/topics/band-schematic-pulses
type: topic
---

# Band-Schematic Pulses: Theory & Applications

to=arxiv_search.search ՞նչjson
{"query":"all:(band-selective pulse OR band-schematic pulse)","max_results":10,"sort_by":"relevance","sort_order":"descending"}
to=arxiv_search.search 【อ่านข้อความเต็มjson
{"query":"1710.09889 OR 1404.2519 OR 2105.13316 OR 2602.10861","max_results":10,"sort_by":"relevance","sort_order":"descending"}
Band-schematic pulses are deliberately structured pulses whose defining feature is their relation to a target band. In condensed-matter control, they are tailored electromagnetic transients whose spectral amplitude, phase, and time waveform are designed from a material band structure and interband transition conditions so that carriers are deposited in chosen regions of reciprocal space. In spin control, they are band-selective or band-uniform pulses whose excitation, inversion, or effective propagator is constrained over a prescribed offset interval. In ultrafast optics and spectroscopy, they are narrowband or band-shaped pulses with controlled temporal envelopes. In radio astronomy, they denote bursts whose energy is confined to a localized fractional bandwidth rather than spread smoothly across the observing band [1710.09889] [1404.2519] [1212.4957] [2105.13316]. The literature therefore uses the term in several non-identical but technically related senses, all centered on explicit shaping of spectral occupancy, passband/stopband structure, or band-resolved dynamics.

## 1. Conceptual scope and terminology

A useful way to organize the subject is by the object to which the “band” refers. In some works the band is a material band structure, in others a frequency passband, an offset band in NMR, a resonance band in an RF structure, or an occupied sub-band in a radio burst. A plausible implication is that “band-schematic” is best understood as a design philosophy rather than a single formal class.

| Domain | Operational meaning | Representative papers |
|---|---|---|
| Graphene and solids | Pulse spectrum mapped to interband transitions in $k$-space | [1710.09889] |
| Spin control and NMR | Uniform control over a selected offset band | [1404.2519], [1701.03286], [2602.10861] |
| Ultrafast optics | Narrowband pulses with engineered envelopes | [1212.4957], [2012.00957] |
| RF and microwave systems | Band-specific AM/PM waveforms or band-edge resonant pulses | [2505.22872], [2206.13564], [1709.07820] |
| Radio transients | Bursts confined to a narrow spectral band | [2105.13316], [2501.05875] |
| Sampling and diagnostics | Essentially multiband pulse models or impulse-based bandpass probing | [1010.3132], [1502.05862] |

Two distinctions recur across these settings. First, some band-schematic pulses are **selective**, meaning they suppress response outside a target band; others are **band-uniform**, meaning they enforce nearly constant action throughout the band. Second, some problems are formulated in the frequency domain, while others are formulated in terms of an effective propagator. In the multi-spin control framework of "Multi-spin control from one-spin pulses" [2602.10861], for example, a band-schematic pulse is defined by
$$
V_\mathrm{S}(\Omega)=Z\!\big(b\,\Omega\,T\big)\;U\;Z\!\big(a\,\Omega\,T\big),
$$
with schematic parameters $(a,b,u_\psi,u_\theta,u_\phi)$ approximately independent of offset across a prescribed band.

A common misconception is that band-schematic always means merely “narrow-band.” The record is broader. The NV-center work emphasizes a flat passband with steep edges rather than a narrow spectral line [1404.2519]; the graphene work uses broadband few-cycle THz pulses as a band-schematic dial in reciprocal space [1710.09889]; and the FRB literature uses the term for bursts with localized spectral peaks and fractional bandwidths of order $0.1$ [2105.13316].

## 2. Band-structure-guided carrier control in solids

The clearest band-structure usage appears in graphene, where low-energy carriers near a Dirac point satisfy
$$
E_\pm(k)=\pm \hbar v_F |k|.
$$
With minimal coupling to a spatially uniform THz field,
$$
H_k(t)=v_F \sigma\cdot(\hbar k + eA(t)),
$$
and the final conduction-band population is
$$
P_c(k)=|\langle u_c(k)\mid \Psi_k(t_f)\rangle|^2.
$$
Because interband resonances satisfy
$$
n\hbar\omega \approx 2\hbar v_F |k|,
$$
spectral components at $\omega$ select rings in $k$-space, so the pulse spectrum becomes a direct control knob for where carriers appear in the Dirac cone [1710.09889].

The reported implementation parametrizes the complex spectrum on a compact $30\ \mathrm{THz}$ band by a $B$-spline expansion,
$$
\hat E(\omega)=\sum_i c_i N_{i,p}(\omega),
$$
with order $p=3$, knot range $\omega_{\min}=2\pi\times 1\ \mathrm{THz}$ to $\omega_{\max}=2\pi\times 30\ \mathrm{THz}$, and endpoint constraints $c_1=0$ and $c_{N_s}=c_{N_s-1}=0$ for $N_s=10$. Time-domain synthesis uses
$$
E(t)=\sum_j \hat E(\Omega_j)\cos(\Omega_j t),\qquad
A(t)=-\sum_j \hat E(\Omega_j)\sin(\Omega_j t)/\Omega_j,
$$
and the fluence is normalized through
$$
U=\frac{c\epsilon_0}{2}\int E^2(t)\,dt.
$$
Optimization employs differential evolution over the spline coefficients, and optionally over $U/U_0\in[1/3,1]$ with $U_0=2.057\times 10^{-5}\ \mathrm{J/cm^2}$.

The principal result is energy-constrained control of the graphene carrier density by a factor of $\approx 4$ at equal pulse energy. With bandwidth fixed at $30\ \mathrm{THz}$ and fluence $U\approx 1.5\times 10^{-4}\ \mathrm{J/cm^2}$, broadband few-cycle spectra maximize the total conduction-band population $N_c$, whereas narrower multi-cycle spectra minimize it. In the representative fixed-fluence broad-window optimization at $U=4.11\times 10^{-5}\ \mathrm{J/cm^2}$, the maximizer is broadband in $[1,30]\ \mathrm{THz}$ and produces large $P_c(k)$ over the accessible window, while the minimizer contracts around $\sim 22\ \mathrm{THz}$ and exhibits pronounced oscillatory fringes attributed to Landau–Zener–Stückelberg interferometry. In a narrow window $k_x\in[0.0507,0.0558]\ \mathrm{\AA}^{-1}$, $k_y\in[0,0.076]\ \mathrm{\AA}^{-1}$, an optimized two-peak spectrum near $\sim 20\ \mathrm{THz}$ with $U=2.01\times 10^{-5}\ \mathrm{J/cm^2}$ and peak fields $\sim 5$–$7\times 10^7\ \mathrm{V/m}$ enhances a multiphoton feature with peak-to-valley ratio $\approx 2$, whereas a narrower spectrum near $\sim 25\ \mathrm{THz}$ with $U=6.94\times 10^{-6}\ \mathrm{J/cm^2}$ suppresses the same feature through deeper minima, consistent with coherent destruction of tunneling and stronger LZS interference [1710.09889].

This solid-state formulation also makes the design logic explicit. Broadband spectra brighten multiple $n\hbar\omega$ rings, narrow spectra isolate fewer rings, spectral nulls suppress undesired channels, and phase engineering can place sub-cycle transients when $\hbar k + eA(t)$ aligns with the polarization axis. The same paper states that the approach generalizes to parabolic bands and gapped semiconductors, where resonance conditions change from the Dirac form to direct-gap expressions such as $n\hbar\omega \approx E_g + 2|E_+(k)|$ [1710.09889].

## 3. Spin control, NMR, and composite passbands

In quantum control, band-schematic pulses are usually band-selective or passband-selective controls. The NV-center study "Band-selective shaped pulse for high fidelity quantum control in diamond" introduces REBURP 180, a BURP-family pure-phase inversion pulse with a flat passband, narrow transition region, and suppressed stopband. It was implemented with IQ modulation from a Tektronix AWG430 and a single microwave source/path, and in room-temperature measurements on NV centers it inverted all three $^{14}\mathrm N$ hyperfine-split lines simultaneously while suppressing nearby $^{13}\mathrm C$ transitions. An $800\ \mathrm{ns}$ REBURP 180 pulse covered the three $^{14}\mathrm N$ dips, and in the six-dip multi-flip experiment it maintained clean inversion of the targeted triplet for at least $16$ flips, whereas rectangular $\pi$ pulses of $\sim 90$–$110\ \mathrm{ns}$ degraded within a few flips; the reported improvement in control fidelity exceeded a factor of $2$ [1404.2519].

Several NMR formalisms turn this passband idea into an explicit dynamical construction. "Feedback Pulses" [2402.00058] defines a simulation-time feedback algorithm on a discrete offset grid $\{\Delta\omega_i\}$ governed by
$$
\frac{d\mathbf M_i}{dt}=\mathbf M_i\times \mathbf\Omega_i(t),\qquad
\mathbf\Omega_i(t)=\big(\omega_1(t)\cos\phi(t),\,\omega_1(t)\sin\phi(t),\,\Delta\omega_i\big),
$$
selects the “worst” offset by $d_i(t)=1+M_{z,i}(t)$ for inversion, sets the RF phase to $\phi(t)=\varphi_k(t)+\pi/2$, irradiates on resonance for that offset, and iterates until all offsets reach the south pole. The resulting broadband inversion pulse is converted into a broadband or band-selective excitation pulse by time reversal with a $\pi$ phase increment. The paper reports experimental broadband inversion over $40\ \mathrm{kHz}$ with $A=10\ \mathrm{kHz}$ and $T=3\ \mathrm{ms}$, broadband excitation over $40\ \mathrm{kHz}$ with $A=10\ \mathrm{kHz}$ and $T=2\ \mathrm{ms}$, and selective excitation over a $\sim 4\ \mathrm{kHz}$ passband with $A=5\ \mathrm{kHz}$ and $T=6.7\ \mathrm{ms}$ [2402.00058].

"Double swept band selective excitation" [1701.03286] solves the same phase-dispersion problem through a different mechanism. A band-selective excitation first drives spins in the target band to the Bloch equator with linearly dispersed phase as a function of resonance offset. That phase is then refocused by nesting free evolution between two adiabatic inversions. The key identity is that an adiabatic inversion $\Theta(\Delta\omega)$ flips the sign of a $z$-rotation, so a double-sweep block converts linear phase dispersion into a uniform final phase. In the reported $^{1}\mathrm H$ experiment at $750\ \mathrm{MHz}$ on residual HDO in $99.5\%\ \mathrm D_2\mathrm O$, one-double-sweep pulses of duration $3.89\ \mathrm{ms}$ and peak RF $10\ \mathrm{kHz}$ produced excitation bands $[-2,2]$, $[-4,4]$, and $[-8,8]\ \mathrm{kHz}$, while a two-double-sweep construction yielded a band-selective $x$-rotation in $6.77\ \mathrm{ms}$ [1701.03286].

Other NMR papers formalize band-schematic action through either robustness or schematic factorization. "Optimal control design of band-selective excitation pulses that accommodate relaxation and RF inhomogeneity" [1111.6650] extends selective-pulse design to demanding regimes with $T_1$ and $T_2$ equal to the pulse length. RC-SEBOP keeps spins near the $z$-axis during most of the pulse, tolerates residual $M_y$ during optimization, and removes that residual by crusher elements after excitation. For $T_p=T_1=T_2=1\ \mathrm{ms}$ it preserves more than $90\%$ of $M_x$ across a $4\ \mathrm{kHz}$ passband and yields a $60\%$ on-resonance signal gain over SLURP-1, while in an eSNOB-like $3\ \mathrm{kHz}$ case it gives about a factor-of-$2$ signal gain and improved stopband suppression.

The multi-spin framework of "Multi-spin control from one-spin pulses" [2602.10861] makes the band-schematic definition fully algebraic. A pulse is band-schematic if its action across a finite offset interval can be written as
$$
V_\mathrm{S}(\Omega)=Z\!\big(b\,\Omega\,T\big)\;U\;Z\!\big(a\,\Omega\,T\big),
$$
with offset-independent schematic parameters over the band. This allows single-spin GRAPE pulses with fixed active evolution times to control weakly coupled multi-spin systems without multispin optimization. The paper demonstrates a continuously irradiated JINEPT element for band-selective transfer
$$
I_z \rightarrow 2I_zS_z,
$$
with total scalar-coupling evolution time
$$
\Delta_{\mathrm{JINEPT}}=4aT.
$$
Experimentally, varying $\Delta$ across total durations $946$, $1285$, $1799$, and $2998\ \mu\mathrm s$ yielded intensities fit by
$$
I(\Delta)=A\sin(\pi J\Delta),
$$
with $J=138.6\ \mathrm{Hz}$ and $\Delta_{\max}=1/(2J)=3.6\ \mathrm{ms}$ [2602.10861].

Two additional constructions emphasize spectral shaping in probability or hardware space. "Narrowband and passband composite pulses for variable rotations" [2004.08456] derives symmetric narrowband and passband composite sequences whose transition profiles versus pulse-area error have vanishing wings. For the NB family,
$$
P(\epsilon)=P_0 \cos^{2(N-1)}(\pi\epsilon/2),
$$
whereas PB sequences are built by twinning two half-$\pi$ NB pulses to create a flat-top around the target probability. "Optimization of Pulses with Low Bandwidth for Improved Excitation of Multiple-Quantum Coherences in NMR of Quadrupolar Nuclei" [1912.05862] addresses the experimental mismatch caused by high-bandwidth GRAPE controls. GROUP constrains the RF waveform to a truncated sine Fourier basis, thereby suppressing high-frequency spectral content and phase transients from the probe and amplifier chain. In $^{87}\mathrm Rb$ MQMAS on $\mathrm{RbClO}_4$, the best GROUP pulse outperformed the best GRAPE pulse by more than $50\%$ experimentally, and the simulation–experiment correlation improved to $R^2\approx 0.6$ for low-frequency GROUP pulses versus $R^2\approx 0.18$ for GRAPE [1912.05862].

## 4. Ultrafast optical and THz pulse engineering

In femtosecond stimulated Raman spectroscopy, band-schematic pulses are narrowband ps-scale Raman pumps whose spectral bandwidth and temporal envelope are jointly optimized. "Optimally shaped narrowband pulses for Femtosecond Stimulated Raman Spectroscopy in the range 330-750 nm" [1212.4957] states that FSRS benefits from spectrally narrow pulses of order $10$–$15\ \mathrm{cm^{-1}}$ and a superior temporal profile, meaning a longer, symmetric, slowly varying envelope. The underlying source is second-harmonic spectral compression in a $25\ \mathrm{mm}$ BBO crystal, for which the paper gives
$$
\mathrm{FWHM}\approx 0.886/\mathrm{GDM},
$$
with $\mathrm{GDM}=L\times \mathrm{GVM}$ and BBO $\mathrm{GVM}$ values of $\sim 320$–$90\ \mathrm{fs/mm}$ as the fundamental tunes from $660$ to $1500\ \mathrm{nm}$. Theoretical SH bandwidths are therefore $\approx 4$–$13\ \mathrm{cm^{-1}}$, and the basic SH-SC method yields multi-$\mu\mathrm J$ pulses with $\sim 10\ \mathrm{cm^{-1}}$ bandwidth and $\sim 0.5\%$ efficiency over $330$–$520\ \mathrm{nm}$. A post-SHG double-pass spectral filter with a $1800\ \mathrm{lines/inch}$ grating and slit widths down to $0.15\ \mathrm{mm}$ suppresses spectral wings, converts the asymmetric “nose” pulse into a broader and more symmetric envelope, increases Raman gain for a given pump energy, suppresses ringing in the $802\ \mathrm{cm^{-1}}$ cyclohexane mode with $\tau\approx 2.0\ \mathrm{ps}$, and extends useful tunability to at least $330$–$750\ \mathrm{nm}$ [1212.4957].

A different route to narrow-band optical pulses is nonlinear generation from oppositely chirped inputs. "Temporal shaping of narrow-band picosecond pulses via non-colinear sum-frequency mixing of dispersion-controlled pulses" [2012.00957] uses non-colinear SFM with
$$
E_3(t)\propto \chi^{(2)}E_1(t)E_2(t),\qquad \omega_3=\omega_1+\omega_2,
$$
and spectral phases
$$
\phi_i(\omega)=\phi_{i,0}+\phi_{i,1}(\omega-\omega_{0i})+\frac{1}{2}\phi_{i,2}(\omega-\omega_{0i})^2+\frac{1}{6}\phi_{i,3}(\omega-\omega_{0i})^3+\cdots.
$$
Under the opposite-chirp condition $\phi_{1,2}=-\phi_{2,2}$, the instantaneous sum frequency becomes nearly stationary, while the output envelope follows
$$
I_3(t)\propto |E_1(t)E_2(t)|^2.
$$
The paper introduces a shape parameter
$$
\alpha=(\phi_3/\mathrm{ps}^3)/(\phi_2/\mathrm{ps}^2),
$$
and reports a numerical case with $\phi_{1,2}\approx +3.5\ \mathrm{ps}^2$, $\phi_{2,2}\approx -3.5\ \mathrm{ps}^2$, $\alpha\approx -0.128\ \mathrm{ps}$, a $2\ \mathrm{mm}$ type-I BBO crystal, and crossing angle $\approx 1.5^\circ$. Simulated SFM efficiency reaches $\approx 65\%$ at $515\ \mathrm{nm}$, practical efficiency is stated as $\lesssim 40\%$, and after subsequent SHG to $257\ \mathrm{nm}$ the shaped envelope improves LCLS-II photoinjector transverse emittance by at least $25\%$, with up to $\approx 30\%$ improvement at $\sigma_z\approx 0.5\ \mathrm{mm}$ [2012.00957].

These optical cases emphasize a recurring pattern. Narrow spectral support alone is insufficient; temporal symmetry, chirp control, group-velocity mismatch, and acceptance bandwidth determine whether a nominally narrowband pulse is practically useful. This is the same structural issue that appears in other domains as phase flatness, group delay, or effective propagator uniformity.

## 5. RF, microwave, and accelerator implementations

In accelerator RF systems, band-schematic pulses are digitally defined AM/PM waveforms adapted to the station transfer function. "High Precision RF Pulse Shaping with Direct RF Sampling for Future Linear Accelerators" [2505.22872] writes the drive as
$$
s(t)=A(t)\cos\big(2\pi f_0 t+\phi(t)\big),
$$
or equivalently
$$
I(t)=A(t)\cos\phi(t),\qquad Q(t)=A(t)\sin\phi(t).
$$
Predistortion is formulated through
$$
Y(\omega)=H(\omega)X(\omega),\qquad X(\omega)=T(\omega)/H(\omega),
$$
and cavity filling by
$$
\frac{dE(t)}{dt}+\frac{\omega_0}{2Q_L}E(t)=\kappa \sqrt{P_{\mathrm{in}}(t)}e^{j\phi(t)}.
$$
Using an RFSoC-based NG-LLRF platform with an NCO near $5.712\ \mathrm{GHz}$, the paper reports C-band tests at $10\ \mathrm{Hz}$ with $1\ \mu\mathrm s$ pulses at $5.17\ \mathrm{MW}$ and $0.45\ \mu\mathrm s$ pulses at $16.45\ \mathrm{MW}$. Demonstrated modulation schemes include square pulses, on/off every $0.2\ \mu\mathrm s$, phase reversal every $0.2\ \mu\mathrm s$, and a $360^\circ$ linear phase ramp over $1\ \mu\mathrm s$. In the SLED-like high-power test, reversal at $0.2\ \mu\mathrm s$ increased the reflection magnitude to more than twice the initial peak, consistent with rapid stored-energy extraction, and the same digital logic is stated to adapt to X-band by adding mixers [2505.22872].

A conceptually different microwave realization uses band-edge physics rather than direct waveform synthesis. "Pulse Generation Based on a Microstrip Circuit with Fourth Order Degenerate Band Edge" [2206.13564] exploits a fourth-order DBE at the Brillouin-zone edge, where four Bloch eigenmodes coalesce and the dispersion obeys
$$
\omega(k)\approx \omega_{\mathrm{DBE}} + A(k-k_{\mathrm{DBE}})^4,\qquad
v_g\approx 4A(k-k_{\mathrm{DBE}})^3.
$$
For a finite cavity, the ideal scalings are
$$
\Delta\omega \propto 1/N^4,\qquad Q\propto N^5.
$$
The reported six-cell RO3003 microstrip cavity exhibits a finite-length DBE resonance at $f_{d,r}\approx 2.86\ \mathrm{GHz}$. A central negative conductance of about $-2\ \mathrm{mS}$ sustains oscillation during the charging phase, and a MOSFET switch to a $50\ \Omega$ load collapses the loaded $Q$ and extracts a short ring-down pulse. The measured output is approximately $0.1\ \mathrm V$ peak and $1.1$–$1.15\ \mathrm{ns}$ wide, with the energy concentrated at the cavity center rather than distributed as in conventional Fabry–Perot or second-order band-edge cavities [2206.13564].

At sub-THz frequencies, the “band” is the pass band of a photonic-band-gap interaction circuit. "Photonic-Band-Gap Gyrotron Amplifier with Picosecond Pulses" [1709.07820] reports a $250\ \mathrm{GHz}$ traveling-wave amplifier with a $30\ \mathrm{cm}$ PBG circuit that confines the desired $\mathrm{TE}_{03}$-like mode and suppresses lower-order modes. The measured device gain is $38\ \mathrm{dB}$ with $8\ \mathrm{GHz}$ instantaneous bandwidth, the operational window tunes over $16\ \mathrm{GHz}$ with beam voltage and magnetic field, and the circuit gain exceeds $55\ \mathrm{dB}$ at $23\ \mathrm{kV}$ and $700\ \mathrm{mA}$. Picosecond input pulses from $260$ to $800\ \mathrm{ps}$ were amplified without observable broadening; at $249.7\ \mathrm{GHz}$ a $320\ \mathrm{ps}$ pulse compressed to about $260\ \mathrm{ps}$, while a $440\ \mathrm{ps}$ pulse emerged as about $425\ \mathrm{ps}$ [1709.07820].

Short transient RF bursts also define the operating paradigm in X-band photoemission guns. "Demonstration of sub-GV/m Accelerating Field in a Photoemission Electron Gun Powered by Nanosecond X-Band Radiofrequency Pulses" [2203.11598] uses $\sim 9\ \mathrm{ns}$, $11.7\ \mathrm{GHz}$ bursts produced by a PETS from an L-band drive-bunch train. The gun is strongly over-coupled with $Q_L\approx 180$, so its loaded bandwidth is $\Delta f\approx 65\ \mathrm{MHz}$ and the filling time is
$$
\tau_{\mathrm{fill}}=Q_L/\omega \approx 2.45\ \mathrm{ns}.
$$
Forward power is typically $250\ \mathrm{MW}$ and occasionally $\approx 300\ \mathrm{MW}$, with a $3\ \mathrm{ns}$ rising edge, less than $3\ \mathrm{ns}$ flat-top, and $3\ \mathrm{ns}$ falling edge. Beam-based metrology inferred a cathode field
$$
E_0=387.76^{+44.63}_{-35.88}\ \mathrm{MV/m},
$$
corresponding to an iris surface field of
$$
601.03^{+69.18}_{-51.61}\ \mathrm{MV/m},
$$
with no major RF breakdown or significant dark current over a three-week run. The physical rationale is the strong pulse-width dependence of breakdown,
$$
\mathrm{BDR}\propto E_0^{30}\tau_p^5,
$$
which favors nanosecond transients over conventional longer pulses [2203.11598].

## 6. Astrophysical, sampling, and diagnostic perspectives

In radio astronomy, band-schematic pulses are pulses whose spectra are localized rather than broadband. "Narrow-band giant pulses from the Crab pulsar" [2105.13316] fits individual pulse spectra between $400$ and $800\ \mathrm{MHz}$ with
$$
I(f)=\kappa (f/f_0)^{\alpha+\beta \log(f/f_0)},
$$
using $f_0=566\ \mathrm{MHz}$ as the reference frequency, and identifies narrow-band pulses by large negative curvature, $\beta\le -30$, together with small fitted FWHM. The fractional bandwidth is
$$
\Delta\nu/\nu \equiv \Delta\nu_{\mathrm{FWHM}}/\nu_0,
$$
or equivalently $(\nu_{\mathrm{high}}-\nu_{\mathrm{low}})/\nu_0$. Out of $1578$ giant pulses found in $77$ minutes of ARO observations, six are narrow-band, an occurrence rate of about $0.4\%$. Their $\Delta\nu/\nu$ values range from about $0.10$ to $0.25$, and they occur in both main-pulse and interpulse phases. Interstellar scintillation is rejected as an explanation because the expected decorrelation bandwidth at these frequencies is about $30\ \mathrm{kHz}$, far smaller than the observed $70$–$130\ \mathrm{MHz}$ widths; the paper therefore argues for an intrinsic origin and notes phenomenological similarity to narrow-band FRB 20190711A with $\Delta\nu/\nu=0.048\pm 0.005$ [2105.13316].

Search methodology has been adapted to this regime. "BASSET: Bandpass-Adaptive Single-pulse SEarch Toolkit -- Optimized Sub-Band Pulse Search Strategies for Faint Narrow-Band FRBs" [2501.05875] augments standard single-pulse searches with time-frequency correlation analysis that identifies and removes noise from zero-detection frequency bands. Applied to FAST observations of FRB 20190520B, BASSET discovered an additional $79$ pulses, increasing the total from the previously known $75$ to $154$ [2501.05875].

Sampling theory supplies a more abstract formulation of the same sparsity structure. "Sub-Nyquist Sampling of Short Pulses" [1010.3132] models multipulse signals with unknown pulse shapes and unknown positions, and extends the framework to essentially multiband signals in the class $\mathrm{MP}(N,W,T,\Omega;B,W_B)$. Using a compactly supported Gabor window $g$ and coefficients
$$
z_{k,l}=\langle f,M_{bl}T_{ak}g\rangle,
$$
the multichannel architecture obtains measurements of the form
$$
Y=DX^T,\qquad X=CZ,
$$
and recovers the jointly sparse coefficient matrix by MMV compressed sensing with $\ell_{2,1}$ minimization. For the essentially multiband case, the total number of samples scales as
$$
JM\approx 8\,N\,B\,(W_B+B)\,W\,\mu^{-1},
$$
which is substantially below Nyquist when both time occupancy and spectral occupancy are sparse [1010.3132].

Band-schematic pulses also appear in instrumentation and safety studies. "Bandpass calibration of a wideband spectrometer using coherent pulse injection" [1502.05862] treats a $350\ \mathrm{ps}$ injected calibration pulse as a broadband impulse with nearly constant spectral amplitude up to $1\ \mathrm{GHz}$. With the Bedlam board sampling at $1.024\ \mathrm{GHz}$, using $8192$ samples per buffer and a pulse repetition rate of $125\ \mathrm{kHz}$, coherent accumulation over $1\ \mathrm s$ gives $N\approx 1.25\times 10^5$ pulses and an observed SNR improvement of about $350{:}1$. Parkes telescope data showed delayed echoes spaced by about $180\ \mathrm{ns}$, corresponding to the known $\approx 5.7\ \mathrm{MHz}$ ripple in the frequency response [1502.05862]. In biomedical electromagnetics, "Peripheral Nerve Stimulation limits with fast narrow and broad-band pulses" [2012.06232] compares narrow-band sinusoidal excitation with broad-band triangular and trapezoidal waveforms using a non-resonant gradient-coil apparatus. Across $51$ volunteers, sinusoidal and triangular thresholds nearly coincide, with fitted means $\Delta B_r=30.3\pm 0.3\ \mathrm{mT}$ and $\tau_c=770\pm 110\ \mu\mathrm s$ for sinusoidal trains, and $\Delta B_r=30.4\pm 0.5\ \mathrm{mT}$ and $\tau_c=519\pm 53\ \mu\mathrm s$ for triangular trains; monophasic trapezoids are slightly more stimulatory, and the threshold in an MRI-like linearly inhomogeneous field is about twice that in a more homogeneous MPI-like field at $5\ \mathrm{kHz}$ for the volunteer explicitly tested in both configurations [2012.06232].

## 7. Cross-cutting design principles and limitations

Across these domains, band-schematic pulse design repeatedly trades selectivity against duration, phase complexity, or hardware burden. In FSRS, a narrower spectral bandwidth implies a longer pulse and the post-SHG filter improves temporal symmetry at the cost of spectral rejection [1212.4957]. In NV control, longer REBURP timescales sharpen the passband edges but increase gate duration [1404.2519]. In composite-pulse design, longer NB and PB sequences suppress the wings and flatten the central response but necessarily increase total pulse area and exposure to decoherence [2004.08456]. In graphene, broadband few-cycle pulses maximize total carrier transfer, while narrow multi-cycle pulses isolate and suppress specific resonances through LZS interference and coherent destruction of tunneling [1710.09889].

A second recurring constraint is the hardware transfer function. Low-bandwidth GROUP pulses outperform GRAPE experimentally because abrupt phase and amplitude changes in piecewise-constant controls excite high-frequency components that are truncated by the downstream RF chain, generating phase transients [1912.05862]. The NG-LLRF work makes the same point in accelerator form by measuring $H(\omega)$ and applying digital predistortion [2505.22872]. The coherent-pulse bandpass calibration method is, in effect, a direct measurement of that transfer function in time and frequency [1502.05862].

A third issue is completeness of observation. The Crab narrow-band giant-pulse paper notes that broad-band searches are intrinsically biased against events whose power occupies only a small fraction of the observing band [2105.13316]. BASSET responds by making the search itself bandpass-adaptive, thereby recovering pulses omitted by full-band summation [2501.05875]. The sub-Nyquist sampling framework provides a theoretical analogue: if time and frequency support are both sparse, the architecture should exploit both [1010.3132].

These patterns suggest a unifying interpretation. Band-schematic pulses are pulses whose design or identification is organized around a band-resolved constraint: a passband/stopband objective, a uniform effective action over an offset interval, a mapping from spectral components to a material band structure, or a narrow occupied sub-band in observational data. The common technical problem is not simply bandwidth reduction. It is the controlled relation between spectral support, phase evolution, and the system-specific transfer function that converts a nominal waveform into a physical effect.

Source: https://www.emergentmind.com/topics/band-schematic-pulses