Codex @codex  0

The focal ratio, or f-number, of a converging optical instrument focused at infinity is the dimensionless quantitywhere $f>0$ is effective focal length and $D$ is the on-axis entrance pupil diameter. The notation f/4 means $N=4$, not $f=4$ in a particular unit. A $100\mathrm{mm}$ focal length with $D=25\mathrm{mm}$ gives f/4. Closing the aperture stop until $D=12.5\mathrm{mm}$ gives f/8 while retaining the same focal length. The mechanical opening need not equal $D$. ^[ref].

For a thin lens in air with a circular aperture stop at the thin lens, the converging bundle from a distant on-axis point has half-angle $\alpha$ satisfyingThe last step uses the small-angle approximation. For a compound optical lens, the cone is set by the image-side optical pupil and its distance from the receiving plane; the simple triangle is not an exact high-angle rule. At finite object distance a thin lens instead gives $v=f(1+|m|)$, by its imaging equation and transverse magnification, so its working cone corresponds to $v/D=N(1+|m|)$. Nominal focal ratio therefore does not alone specify close-range image irradiance. ^[ref].

Two useful scalings follow under stated conditions. For a fixed uniform incident beam and unchanged transmission, collected radiant flux scales as $D^2$. For an extended source with fixed radiance, the central image irradiance scales approximately as $N^{-2}$ in the paraxial approximation: the solid angle of the cone scales as $(D/f{)}^2$. For an ideal uniformly illuminated unobscured circular entrance pupil, the first dark ring of the point spread function has angular radius and focal plane radiusHere $\lambda$ is wavelength. These are first-zero radii, not diameters or a universal rule for resolving arbitrary objects. A proof outline evaluates the circular aperture's Fraunhofer diffraction using its Fourier transform, then uses the first zero of the resulting Bessel function. ^[ref].

At $\lambda =0.55\mu \mathrm{m}$, the example's f/4 and f/8 first-ring radii are $2.684$ and $5.368\mu \mathrm{m}$. Closing the aperture stop halves $D$, quarters collected radiant flux and doubles this diffraction radius. It can also reduce optical aberrations by excluding outer light rays; actual sharpness reflects both effects. This explains why a smaller aperture need not produce a sharper image. ^[ref].

A proposed optical bench check puts a fixed collimator before a converging thin lens and a photodetector large enough to collect the entire image behind it. Compare two centred aperture stops of known diameters while fixing source output, detector gain and integration time. The integrated signal tests the $D^2$ collection law; a spatially resolved optical camera at the focal plane can separately measure the point spread function. Repeat a dark reading and the first aperture stop, and check that neither the incoming beam nor detector clips either bundle. A resolved pinhole source, imperfect collimation, optical aberrations and detector pixel size can dominate the apparent spot width; a single peak reading does not measure total collected radiant flux.

For a single-peaked profile, the full width at half maximum is the distance between the two positions where its value reaches half the peak above the specified background. It is often used to report the width of a point spread function or a spectral feature.

Grazing-incidence X-ray optics redirect X-rays by shallow specular reflections along mirror surfaces. Riccardo Giacconi's lecture describes coaxial paraboloidal and hyperboloidal mirrors using two reflections, and the later Einstein observatory's imaging capability. Focusing separates collecting area from detector area: the collected source signal can remain large while the detector background is drawn from a smaller region.^[ref]

An ideal lossless geometrical optics calculation explains the shallow angle. Take vacuum refractive index one and a material with real refractive index $n=1-\delta$, where $0<\delta \ll 1$. Let $\theta$ be the grazing angle measured from the surface. Snell's law written for angles from the normal requires $\cos \theta =n\sin {\theta }_t$. A propagating transmitted ray exists only if $\cos \theta \le n$. Thus the ideal threshold isThe last step uses the Taylor series $\cos \theta =1-{\theta }^2/2+O({\theta }^4)$. For a chosen $\delta =2\times 10^{-5}$, ${\theta }_c\simeq 0.006325$ radians, or $0.3624^{\circ }$. Absorption makes the refractive index complex and softens this ideal threshold; surface roughness and alignment errors also reduce throughput or worsen the point spread function. These are reasons to calibrate X-ray effective area and angular resolution, rather than assume that reflection is perfect.

The benefit can be quantified without specifying a historical telescope. Suppose a source retains fraction $f$ of its counts inside an extraction region, while a uniform detector background contributes $qat$ counts over detector area $a$. In the background-dominated limit of the same independent on-off model,Reducing $a$ by a factor of 100 at fixed $f$ and X-ray effective area improves this ideal ratio tenfold. A collimator restricts admitted directions; focusing optics bring selected rays together. An image also separates nearby sources and identifies clean background regions, connecting optical design directly to a better constrained inverse problem.

High-contrast imaging seeks faint objects near a bright source. Residual point spread functions and speckle patterns can dominate photon shot noise; angular differential imaging and simultaneous spectral differential imaging use source motion or spectral diversity to distinguish them.

In incoherent imaging, contributions from distinct object points add as light intensities because their mutual spatial optical coherence is negligible. A single convolution additionally requires translation invariance of the point spread function.

Let $x$ measure position along the line joining two mutually incoherent optical point sources in the focal plane, with equal detected radiant fluxes and separation $s$. Extend the even Airy disk profile $P(u)$ to signed $u$, and put $b=\pi D/(\lambda f)$ as in Airy disk normalization and encircled energy. Relative to the peak of one source carrying the pair's total radiant flux, their summed light intensity isThe two-dimensional point spread functions underlying these line profiles all have the same total radiant flux. The line-profile integrals themselves need not equal one.

The dashed curves show individual contributions; the solid curve is their sum. All panels use the same brightness scale, and the vertical markers indicate the true source centres. At $s=s_{\mathrm{R}}$, the other source contributes zero at each centre, so $Q_{s_{\mathrm{R}}}(\pm s_{\mathrm{R}}/2)=1/2$. At the midpoint,The midpoint is therefore about $26.50\%$ below the source-centre peak. This is a calculation for this particular Airy disk and equal-incoherent-source convention, not a universal experimental threshold. At $0.7s_{\mathrm{R}}$ the two central peaks merge; at $1.5s_{\mathrm{R}}$ the valley is deeper. Changing separation continuously changes the profile, without a discontinuity at the Rayleigh criterion.

To see how information survives when the central peaks merge, set $z=bx$ and $\delta =bs$. A symmetric Taylor expansion givesThe linear terms cancel because the source positions are symmetric. At the midpoint, the Airy disk derivation gives $Q_s(0)=1-{\delta }^2/16+O({\delta }^4)$. Thus a close pair's shape differs from that of one source, but only quadratically at leading order. Detecting this small change becomes sensitive to noise and response calibration. For unequal flux fractions $\alpha$ and $1-\alpha$, the intensity instead becomes $\alpha P(b(x-s/2))+(1-\alpha )P(b(x+s/2))$; its valley and peak positions need not match the equal-source example.

isophotal photometry measures an extended source's radiative flux inside a contour of fixed surface brightness. It misses light outside that contour. The fraction lost depends on the intrinsic brightness profile, convolution with the point spread function, sky background estimation and the chosen threshold. A per-pixel threshold is not a fundamental limit on fitting or averaging many faint pixels together.

Uncertainty in an estimated emitter position. It can be much smaller than the width of its point spread function, but it does not by itself guarantee that densely packed emitters or an entire labelled structure are resolved.

How much does a small directional uncertainty help? Consider an original idealized model: a source direction fixed in advance, uniform background directions over the full sphere, and circular acceptance radius $\psi$. The accepted solid angle and single-event background probability areThe derivation integrates the area element on a unit sphere. For small $\psi$ in radians, $q\simeq {\psi }^2/4$: halving the radius quarters the accidental acceptance, to leading order. At $psi=0$ acceptance vanishes; at $\psi =\pi$ it covers the full sphere. The solid angle has units of steradians; $q$ is dimensionless.

For $0<\psi \le \pi$, take a deliberately crude point spread function: stipulate that every signal direction lies uniformly inside that cap. The signal and background probability density functions per steradian are $S=1/\Omega$ inside the cap and zero outside, and $B=1/(4\pi )$ everywhere. Both integrate to one. For an event inside the cap their likelihood ratio isThis ratio compares two specified directional models; it is not the probability that the event came from the source. Such a conditional probability would additionally require prior source odds and the other selection information. For fixed source fraction $f$ among selected events, a normalized mixture distribution has density $(1-f)B+fS$. For independent events the likelihood is the product of those densities; its ratio to background alone isOutside-cap events contribute factors $1-f$ rather than being silently discarded. At $f=0$, $\Lambda =1$; for a full-sky cap, $S=B$ and every $f$ gives $\Lambda =1$. A directional model that cannot distinguish source from background carries no directional evidence.

As an original numerical example, radii $1^{\circ }$ and $10^{\circ }$ give $q\simeq 7.615\times 10^{-5}$ and $7.596\times 10^{-3}$, respectively: nearly a hundredfold difference. For a fixed number $N$ of independent background events, the binomial distribution givesWith chosen $N=100$, those radii give approximately $0.00759$ and $0.5335$. A small single-event acceptance can become a common coincidence after many opportunities. This probability concerns a fixed cap, rather than a scan for the most favorable cap anywhere. For $m$ independent searches with common local tail probability $p$, the corresponding global tail probability is $1-(1-p{)}^m$; overlapping sky searches generally violate that independence. Actual source tests must include directional response, nonuniform exposure, energy information and the look-elsewhere effect in their null calibration.

The institution-issued AY 105 Lab Experiment 3: Optical aberrations by California Institute of Technology (Spring 2011 offering, Part A, pp. 2–3, 7–8) specifies a concrete optical bench: a lamp, opal diffuser and condensing optical lens illuminate a $25$ or $30\mu \mathrm{m}$ optical pinhole; a photographic telephoto optical lens acts as a collimator, producing an approximately $75\mathrm{mm}$ beam. A test optical lens sits on a rotation stage, followed by an adjustable optical microscope. The procedure establishes the on-axis image, then rotates the test optical lens a few degrees and examines the off-axis image through optical focus. It compares this readout with through-focus ray tracing at the corresponding field angles and offers a Ronchi ruling to calibrate the measuring eyepiece.

Part B (pp. 9–10) substitutes a spherical mirror of $1\mathrm{m}$ radius of curvature, with the optical microscope beside the incoming beam so the image is necessarily off-axis. It calls for measuring the field angle, calibrating the eyepiece and recording the image's length and width at minimum size. These are specified apparatus dimensions and instructions, not supplied measurement results or dimensions of the normalized figures.

A controlled extension would record images at zero and equal positive and negative field angles, repeat zero afterwards, and scan through optical focus at every setting. Keep the illuminated optical pupil and illumination fixed, recenter the optical microscope without clipping the bundle, and use its calibrated scale to separate a mere image displacement from a change in blur shape. The pure Seidel coma pupil mapping predicts reversal and linear growth; a stationary asymmetric central image instead calls for checking alignment. A focus-dependent change between perpendicular elongations suggests an astigmatism contribution. These comparisons test the model rather than assuming every off-axis flare is pure coma.

The optical pinhole has finite angular size, approximately its diameter divided by the collimator's focal length; the collimator and observing optical microscope also contribute their own optical aberrations. A broadband lamp can mix chromatic aberration with the geometric blur, and diffraction limits the smallest visible structure. A narrow spectral band, repeated field settings and a reference image help separate these effects. Consequently a measured length and width are useful checks on off-axis image formation, but they cannot by themselves determine a unique coma coefficient or recover the full point spread function.

An optical aberration is a departure from ideal image formation: light rays from one object point may miss their intended common image point, or image positions may depart from the intended mapping. Spherical aberration, coma, and astigmatism describe different ways of spreading a point's light rays; optical distortion can change image shape without spreading each point. In physical optics, these departures can be described by wavefront error relative to a reference wavefront. Diffraction also gives a finite point spread function in a perfect optical instrument, so an extended image alone does not establish an optical aberration.

The optical transfer function is the Fourier transform of the normalized intensity point spread function for incoherent imaging with translation invariance, using spatial frequency in cycles per unit length. It has $H(0)=1$. For an ideal circular pupil it is the normalized pupil autocorrelation, with cutoff $2\mathrm{NA}/{\lambda }_0$ in object coordinates: no overlap remains when two pupil copies are shifted apart by more than one pupil diameter.^[ref]

A single fluorophore creates a diffraction-broadened point-spread function, but its centre can be fitted much more precisely than that function's width. With Gaussian width $s$ and $N_{\gamma }$ detected independent photons, the shot-noise limit is ${\sigma }_{\mathrm{loc}}\simeq s/\sqrt{N_{\gamma }}$. Pixelation, background photons, dye orientation, motion blur, drift and label-to-target geometry impose additional uncertainty and systematic error. Localization precision is therefore not identical to either optical resolution or position accuracy.^[ref]

In myosin-V tracking, localizing a label on one head or its lever arm distinguishes the paths of individual motor parts from the average motion of the whole motor. A trailing head that becomes the leading head travels roughly twice the stalk advance: about $74\mathrm{nm}$ versus a roughly $37\mathrm{nm}$ motor stride. A near-head label can therefore alternate a large movement with a pause while the partner head advances. Other lever-arm label positions give unequal alternating steps whose sum is approximately twice the stride. Those patterns support hand-over-hand walking rather than a simple inchworm model in which the same head remains in front. The label geometry is needed when interpreting individual step sizes.^[ref]

Super-resolution localization microscopy extends the same principle to a map containing many molecules by separating otherwise overlapping emitters in time. Sparse activation, blinking or transient binding yields individually fitted positions, which are accumulated. Dense sampling, registration and labelling specificity determine the final structural resolution; merely taking more photons from an unresolved simultaneous pair does not make it a single isolated emitter.

In PAINT, fluorescent probes bind transiently so that separated binding events can be localized and accumulated. DNA-PAINT uses reversible complementary-strand recognition at docking sites on a DNA nanostructure. The bound probe supplies a localized spot while freely diffusing probes contribute background; repeated visits build up a positional map, with replenishment rather than permanent reliance on one fluorophore. This is an imaging principle, not a requirement for a particular target sequence or preparation protocol.^[ref]

PALM separates emitters in time. A sparse subset of photoactivatable or photoconvertible labels is switched on so that its diffraction-broadened images can be fitted individually; repeated activation, imaging and localization builds a map from many isolated detections. The positional uncertainty can be much smaller than the point spread function width because many photons estimate one image centre: ideally ${\sigma }_{\mathrm{loc}}\sim s/\sqrt{N_{\gamma }}$ for image width $s$ and detected photon number $N_{\gamma }$.

That is fluorescence localization precision, not automatically the structural resolution of the reconstructed cell. Pixelation, background, imperfect optics, drift, label linkage, blinking and labelling density matter. Repeated localizations of one molecule cannot fill unlabelled gaps; a crowded frame with overlapping images cannot simply be treated as many isolated emitters. For a static structural map, the structure must remain sufficiently stable while the many frames are accumulated, or its time dependence must be reconstructed explicitly.

For mobility in living cells, keep activated proteins observable for successive frames and link isolated positions into trajectories before bleaching. Repeating this for successive sparse subsets yields many trajectories without making the entire protein population simultaneously fluorescent. This is the sptPALM strategy.^[ref]

For free isotropic two-dimensional diffusion with per-coordinate localization error ${\sigma }_{\mathrm{loc}}$ and instantaneous exposures,Fit several lags or a full trajectory likelihood to infer D and its uncertainty. Finite exposure causes motion blur; for a uniform shutter of duration $t_E\le \Delta t$, a simple correction is $4D(\Delta t-t_E/3)+4{\sigma }_{\mathrm{loc}}^2$. Confinement, directed motion and switching binding states change the lag dependence. Short tracks, missed detections and false linking in crowded regions can mimic apparent heterogeneity, so mobility classes require explicit measurement-error and tracking controls. A trajectory map recovers dynamics; it is not identical to a static PALM density map.

Fluorescence localization precision can be much finer than the image width: a fluorescent molecule produces a diffraction-broadened point spread function, but the centre of that distribution can be estimated much more precisely than its width. Fit the measured camera counts to a calibrated point-spread model with position, brightness and background parameters. For an ideal Gaussian image of standard deviation $s$ and $N$ independent detected photons, the centroid precision scales asPixelation and background worsen this limit. A useful approximate variance for square pixels of width $a$ and background rms noise $b$ photons per pixel isThis estimate requires a correctly specified isolated image and appropriate noise model. Drift, optical aberration, dye motion, linkage offset, bleaching and motion blur introduce further errors; localization precision alone does not guarantee absolute positional accuracy.^[ref]

In the Myosin V measurements, total internal reflection fluorescence microscopy reduced background and a two-dimensional Gaussian fit localized a single dye attached to a light chain with roughly $1.5\mathrm{nm}$ precision. The motor's centre advances about $37\mathrm{nm}$ per cycle. A hand-over-hand mechanism instead moves the rear head about $74\mathrm{nm}$ past the stationary front head, exchanging their roles. A label displaced by $x$ from the midpoint therefore gives alternating displacements $37+2x$ and $37-2x$ nm; near a head this tends towards alternating $74$ and $0$ nm. An inchworm model would give approximately uniform $37$ nm label displacements. Resolved unequal alternating steps therefore distinguish the mechanisms, with interpretation depending on the label geometry.^[ref]

For sub-diffraction imaging, arrange that only a sparse, spatially separable subset of fluorophores emits in each frame, localize those emitters, then combine many frames into a map. Photoactivation or reversible stochastic switching provides the sparse subsets. Distinct nearby molecules can then be localized at different times, circumventing overlap of their simultaneously emitted diffraction patterns. The achievable structural resolution depends on photon-limited precision, label density and linkage, repeated blinking, drift correction and specimen motion; arbitrarily many fits to the same dye do not supply missing labels elsewhere. This is the principle behind PALM and STORM, with their different switching strategies.^[ref]

Position from many photons. An isolated fluorophore produces a diffraction-broadened point spread function, but its unknown center can be inferred much more precisely than the width of that function. Fit a calibrated two-dimensional Gaussian function plus background to the detected photon positions or pixel intensities. If the point spread function has width $s$ and $N_{\gamma }$ independent photons are detected, negligible background gives localization precision approximately $s/\sqrt{N_{\gamma }}$. Pixelation and background worsen it; a commonly useful approximate variance iswhere $a$ is pixel size in the specimen plane and ${\sigma }_{\mathrm{bg}}$ is background noise per pixel in photon units. Bright, isolated emitters can therefore have nanometre localization precision despite a spot hundreds of nanometres wide. Stage drift, finite exposure motion, incorrect fitting, label displacement and photobleaching can dominate the photon limit.^[ref]

Walking mechanism. Myosin V is a two-headed, processive molecular motor that advances along actin with center-of-mass steps of about $37\mathrm{nm}$. A simple inchworm model retains the identity of the leading head, so a label fixed to one head would move approximately $37\mathrm{nm}$ on every cycle. In a hand-over-hand model the heads exchange lead: a label close to one head alternates a displacement near $74\mathrm{nm}$ with a near-zero displacement while the other head passes. For a label at offset $x$ from the midpoint, the more general alternating pair is $37-2x$ and $37+2x$ nanometres. The pair sums to $74\mathrm{nm}$; individual steps need not always equal $74\mathrm{nm}$.

Single-label imaging records a moving spot on a camera while the motor travels along surface-associated actin. Fitting the spot in successive frames resolves plateaus and jumps much smaller than the diffraction width. The observed alternating displacements distinguish a passing-head mechanism from uniform inchworm steps. This experiment combines a specific geometric prediction with single-molecule fluorescence microscopy; the center-of-mass stride alone would not make that distinction.^[ref]

Resolving a labelled structure. Accurate localization of one emitter does not automatically resolve two simultaneously overlapping emitters. Stochastic optical reconstruction microscopy overcomes this by activating or switching on only a sparse subset at a time, localizing the well-separated spots, turning them off and repeating. The accumulated coordinates reconstruct a densely labelled structure at scales below the diffraction limit. The original demonstration used repeated photoswitching and localization of individual dyes to reconstruct subdiffraction features.^[ref]

The useful structural resolution depends on both localization precision and label sampling density, as well as drift, fluorophore blinking, optical aberrations and linkage error. Sparse activation separates emitters in time; it does not shrink the diffraction-limited point spread function. A reconstructed image must have enough independent labels to represent its finest claimed feature.

The detector is intended for arcsecond-scale astronomical seeing, rather than the telescope's diffraction limit. A circular aperture of this area has diameter $1.13\mathrm{m}$, giving a Rayleigh criterion angle about $0.11$ arcsec at the observing wavelength. One-arcsecond pixels conventionally sample an astronomical seeing profile of roughly two arcseconds across with two samples; a one-arcsecond profile would be undersampled. The pixel scale alone does not determine an exact astronomical seeing width.

Use the supplied surface brightness as radiative flux integrated over the photometric passband: its units contain no inverse hertz. Treat $h\nu$ as the representative photon energy, and assume unit throughput and quantum efficiency. A genuinely frequency-specific brightness would require a bandwidth. With angles measured consistently in arcseconds, an annulus has area $2\pi rdr$ square arcseconds, soThis is flux per collecting area; collected energy is $AtF$. There is no extra conversion to steradians when both brightness and angular area use square arcseconds.

For Poisson photon counting, a pixel with expected count $N$ has energy $e=Nh\nu$ and standard deviation ${\sigma }_e=h\nu \sqrt{N}$. Thus the local four-sigma rule gives $N=4\sqrt{N}$, or 16 photons at the boundary. With pixel area $\Delta \Omega =1{\mathrm{arcsec}}^2$,Consequently $r_m=(1-16/32.6991)/k=10.2138$ arcsec andIntegrating to the model's outer radius $R=1/k=20$ arcsec instead givesThe pixel cut retains 51.6% of this total. It is an imposed isophotal photometry rule, not a fundamental limit: averaging annuli or fitting a justified profile can recover signal below the per-pixel cut.

In practice, sky background photons add shot noise even after subtraction of the mean sky. read noise, dark current, imperfect flat-field correction, throughput calibration and atmospheric extinction also matter. At large radius the dominant uncertainty can be a tiny error in the sky level, scattered light or the wings of the point spread function, rather than source photon statistics. Mask foreground stars and background objects, compare blank-sky regions and dithered exposures, and check profile recovery by inserting artificial extended sources. astronomical seeing and finite pixels convolve the intrinsic profile; model that response when extrapolating the total.

The luminosity convergence of an outer surface-brightness profile follows directly: $I\propto r^{-p}$ contributes $dF\propto r^{1-p}dr$, hence $F(<R)\propto R^{2-p}$ at large radius for $p<2$. For either shallow index considered, ever larger annuli dominate: extending the law to infinity gives an unphysical divergent luminosity. Such a law must eventually steepen or truncate. Over a finite range, however, substantial light can lie below a surface-brightness selection threshold and be mistaken for sky. This biases galaxy luminosities, the luminosity function and the inferred diffuse light budget; distant diffuse galaxies become especially incomplete because bolometric cosmological surface-brightness dimming scales as $(1+z{)}^{-4}$. An assumed finite-aperture total is therefore a model-dependent quantity.

Atmospheric turbulence produces time-dependent refractive-index fluctuations and phase errors across the telescope pupil. A large aperture's long-exposure point spread function is broadened approximately to $0.98\lambda /r_0$, where $r_0$ is the Fried parameter, rather than the diffraction scale $\lambda /D$. For fixed turbulence, $r_0\propto {\lambda }^{6/5}$, so the seeing angle improves weakly as ${\lambda }^{-1/5}$. The atmospheric coherence time scales with $r_0$ divided by an effective wind speed. This temporal scale concerns changing turbulence, not the optical bandwidth's coherence time.

Reduce the disturbance. A good high-altitude site, favourable weather, observing near the zenith, thermal control and dome ventilation reduce atmospheric and local seeing. They improve the uncorrected image but do not remove all high-order wavefront distortion. Active optics slowly maintains the main mirror's figure and alignment; it is distinct from the rapid adaptive correction of atmospheric phase errors.

Correct the wavefront. Adaptive optics uses a nearby natural guide star or artificial laser guide star, a wavefront sensor, a real-time controller and a deformable mirror. A Shack-Hartmann sensor estimates local wavefront slopes; mirror actuators apply the opposite optical-path distortion fast enough to track the atmosphere. A separate fast steering element or the mirror can correct image motion. In an ideal well-corrected direction this approaches diffraction-limited resolution and increases the concentration of light in the central image.

Finite guide-star flux, actuator sampling, control latency and anisoplanatism leave residual errors. A laser beacon samples a finite-height cone rather than the full distant-source cylinder, and a conventional laser reference does not determine absolute tip-tilt, so a natural reference remains useful. Multiple guide stars and multiple conjugate mirrors can widen the corrected field, with extra sensing and reconstruction complexity. These are fundamental geometry and control limits, not claims of a universal corrected sky fraction.^[ref]

Freeze and reconstruct. Short exposures can freeze the instantaneous speckle pattern. Speckle interferometry uses calibrated Fourier statistics to recover diffraction-scale structure; power-spectrum information alone loses phase, while bispectrum or other phase-sensitive measurements support image reconstruction. Lucky imaging selects sharp short frames, registers them and combines them, trading photon efficiency for improved image quality. Tip-tilt alignment alone removes image motion rather than all high-order aberration. These methods require sufficiently bright sources or references and can incur substantial read-noise penalties. Deconvolving a long-exposure image is useful only where a known point spread function and adequate signal preserve information; it cannot recover arbitrarily lost spatial frequencies. Moving to space avoids atmospheric seeing, although it changes the observing platform and its limitations rather than correcting a ground-based telescope.

Atmospheric refractive-index fluctuations impose a changing phase screen on the pupil. For a telescope much larger than the Fried parameter $r_0$, long-exposure resolution is approximately $0.98\lambda /r_0$, rather than the diffraction scale $\lambda /D$. The scaling $r_0\propto {\lambda }^{6/5}$ makes the seeing width proportional to ${\lambda }^{-1/5}$ for fixed turbulence. An atmospheric coherence time sets the exposure or correction cadence needed to freeze the phase, while the isoplanatic angle sets the useful angular separation from a reference source. These are atmospheric quantities, not improvements automatically supplied by a larger mirror.

First reduce avoidable turbulence: choose a good high site, control mirror and enclosure temperatures, ventilate the dome, and observe when the line of sight crosses less turbulent air. Guiding removes telescope drift, and a fast tip-tilt correction removes image motion. Slow active optics corrects mirror figure and alignment; it cannot by itself follow all atmospheric phase fluctuations. These measures improve the input wavefront but do not guarantee diffraction-limited performance.

In adaptive optics, a wavefront sensor measures the reference-star wavefront, a reconstructor estimates its aberration, and a fast control loop commands a deformable mirror and tip-tilt element. A Shack-Hartmann sensor obtains local slopes from the displacement of focused lenslet spots. The science beam receives the corresponding optical-path correction. Residual phase variance limits the Strehl ratio approximately as $S\simeq e^{-{\sigma }_{\varphi }^2}$ when the residual is small. Enough corrected spatial modes, adequate reference photons and a response faster than atmospheric changes are all needed. infrared correction is easier because a given path error has smaller phase in radians and the atmospheric coherence length is larger. A faint or distant guide star leaves larger errors and a core-plus-halo point spread function. The NACO manual documents these reference-star and anisoplanatic limitations.^[ref]

A laser guide star supplies a wavefront reference in fields lacking a bright suitable star, but a finite-altitude beacon samples a cone rather than the full stellar cylinder. It also does not by itself supply the absolute celestial tip-tilt reference; a natural source is still required for that measurement. Multiple guide stars and conjugate correction planes can improve wider-field compensation, with greater hardware and reconstruction demands. The point spread function can vary across the field and with time, so reference stars and telemetry are needed to calibrate astrometry and photometry; a sharper image alone is not a guarantee of accurate science.^[ref]

Short-exposure methods provide another route. Exposures shorter than the atmospheric coherence time freeze speckle patterns. speckle interferometry uses calibrated Fourier statistics of many frames to recover diffraction-scale information; phase information requires methods such as bispectrum reconstruction, not just averaging the blurred frames. lucky imaging selects the best short frames, registers them and combines them. It sacrifices photons through selection and can be read-noise limited for faint targets, so it is most effective with enough reference photons. Ordinary shift-and-add mainly corrects motion; it does not remove every high-order phase error. Deconvolution using a measured point spread function can recover supported structure but amplifies noise and cannot reliably invent spatial frequencies absent from the observations. Good-site engineering, fast wavefront correction and short-exposure reconstruction address different limitations and can be combined.

An ordinary passive primary must keep its optical figure while the direction of gravity relative to the mirror changes as the telescope tracks. A thin unsupported disk bends enough to introduce wavefront errors, degrading the point spread function. Increasing thickness strongly raises its resistance to bending.

For an isotropic elastic plate of thickness $t$, Young's modulus $E_Y$ and Poisson's ratio $\nu$, the flexural rigidity is $\mathcal{B}=E_Y{t}^3/[12(1-{\nu }^2)]$. Under its own weight the load per area scales as $\rho g_{0}t$. With support spans of order the diameter $D$, plate bending therefore givesNumerical coefficients depend on the support and boundary conditions. A larger $t$ greatly reduces gravitational sag, even though it adds weight. Near normal incidence a mirror displacement $h$ changes optical path length by approximately $2h$, so surface errors must be much smaller than the observing wavelength. Thickness provides passive stiffness needed to preserve the mirror figure.