For Darcy-Bénard convection, the dimensionless Darcy law is , with incompressibility and a heat equation about . Here is a Darcy thermal-time Rayleigh number, with permeability replacing the squared-depth dependence in the fluid-layer Rayleigh number. The stated stream function convention gives , so . The vertical-plane curl of Darcy law gives ; omitting the quadratic perturbation advection from the heat equation gives .
On the impermeable connected square boundary, zero normal velocity makes constant along the boundary; choose that constant as zero. The prescribed conductive boundary temperature fixes on all four sides. Thus the perfect-conductor wording is interpreted through the supplied homogeneous perturbation data: the side temperatures retain the background vertical profile, rather than imposing a different isothermal side state.
At marginal stability put and . The two coupled equations imply
The gauge transformation for conducting-square Darcy onset removes the first derivative:
The square's Dirichlet Laplacian eigenfunctions are , with eigenvalues of equal to , . The first marginal value is therefore
It is genuinely the first instability threshold, not just an isolated neutral value. Eliminating gives the temperature linear operator , where is the Dirichlet realization of an elliptic operator for the Laplacian. Integration by parts shows the second term is self-adjoint and nonnegative, since its quadratic form is . Thus the spectrum is real, starts negative at , and its leading eigenvalue cannot cross into growth without a zero eigenvalue. The smallest zero value is the one found above.
Let , and . Multiplying the lowest real by independent real and imaginary constants gives two real physical eigenfunctions:
The subscripts indicate even and odd parity about . Because is even in , the first has even and odd , while the second reverses those parities. They are linearly independent and satisfy all four Dirichlet boundary conditions. The lowest complex eigenfunction is simple over the complex numbers, but its arbitrary complex multiplier supplies two independent real pairs; this explains the physical degeneracy.
Figure 1.
Neutral and oscillatory modes in porous convection
. Details.
For a real steady amplitude, integrate the conserved-field equation twice. Periodicity forces the integration slope to vanish, and the zero mean fixes the remaining constant. Hence
Substituting this into the steady amplitude equation gives the nonlocal ordinary differential equation
For , use the hyperbolic secant identity . Vanishing of the two independent coefficients gives
Choose and ; an overall negative sign is another solution. Thus . The exact average of the proposed profile over the finite interval is
For , the hyperbolic tangent differs from one by , so . Eliminating gives the requested leading relation
Equivalently , with . Positive real roots exist precisely when ; they are and are reciprocal. Each candidate must additionally satisfy . For , the leading necessary range is , with equality at the coalescence . Away from this leading-order fold, the inequality is strict.
The strict finite-window condition can also be seen by retaining the average exactly. Its consistency relation is . Since , it implies , and hence
The strict endpoint distinction is exponentially small in the large- approximation; the leading polynomial alone permits its double root. When , this algebra imposes no finite upper bound on . The printed interval therefore needs : for example , gives a large positive root of the leading equation and , despite the printed upper endpoint being negative.
Finally, the single localized pulse of a conserved-field convection model is an approximate periodic solution. Its values agree at the interval ends, but its derivatives have opposite signs there; their magnitude is exponentially small, . Periodic-tail corrections are needed for an exact smooth periodic boundary condition. The range derived from the truncated profile is a consistency condition, not a proof that every permitted parameter gives an exact stable periodic pulse.
Assume . The symbol in this part is the convection roll amplitude, not the Rayleigh number of the previous part. Set for the algebra. The periodic boundary condition requires and modulation wavenumbers . The conserved-field equation preserves the imposed zero mean of , since the integral of an derivative over a period vanishes.
Equating the coefficients after linearization of the convection amplitude coupled to a conserved field gives
The conjugate sideband is needed because the variation of couples a perturbation to its complex conjugate. Put , . The linear operator on is
Its characteristic polynomial, expressed conveniently without expanding every coefficient, is
At it has roots . The phase root is neutral; the uniform root is excluded by the prescribed zero-flux-perturbation mean. Nonzero arbitrarily long-wave modes remain allowed on a sufficiently large domain. For fixed , the two roots tending to zero have at simple limiting roots. Dividing the polynomial by yields
Repeated limiting roots may require a further expansion, but do not affect the strict instability criterion. If , the constant term is negative and the two real roots have opposite signs. The positive root gives a growing long-wave sideband instability. Using gives
This is sufficient for instability when a sufficiently small nonzero allowed is available. It is not an unconditional instability theorem for every fixed finite period. For example , , , satisfy the inequality, but the nonzero allowed obey . Their roots are and , all negative. The prescribed zero mean removes the uniform flux root and the remaining uniform phase is merely neutral. The corrected conclusion is the long-wave instability with a conserved mean field, with domain size and mode availability stated.
Eliminate the instantaneous Stokes flow velocity in favour of temperature. On a horizontal Fourier mode , the Stokes temperature-slaving operator maps to , where and . The temperature evolution has linear operator and bilinear map . Under the homogeneous thermal Dirichlet boundary conditions, is self-adjoint. Normalize its critical eigenfunction as and set ; the critical vertical velocity is .
At order , the critical eigenfunction equation gives . At order , the weakly nonlinear expansion contains the imposed second harmonic and the quadratic products of the critical mode: a horizontally uniform temperature correction proportional to and, in a general vertical-mode calculation, a second harmonic proportional to . These corrections are found by solving the noncritical boundary value problems, with homogeneous thermal data except for the imposed forcing.
At order , the method of multiple scales produces the slow derivative , the detuning term , and the two cross-advection terms involving first- and second-order fields. Project the component onto the adjoint eigenfunction using the vertical inner product. This is the solvability condition in the method of multiple scales: divide each resonant projection by . The detuning supplies with ; interactions of horizontal wavenumbers and permit with ; self-interaction through the slaved mean and second harmonic supplies . Other products have the wrong horizontal wavenumber. Reflection permits real coefficients with this cosine forcing. Thus the symmetry-allowed spatially forced convection amplitude equation is
There is a useful specialization that should not be silently missed. For the literal one-vertical-mode Stokes flow problem, the vanishing two-to-one forcing coefficient for Stokes convection makes at this order. To see this, write a positive second-harmonic forcing component as , incorporating the cosine's factor . Its coupling to the negative critical harmonic has projected integrand, apart from sign and its factor ,
The integral vanishes because at both plates, even though is nonzero. This proves the cancellation without solving the forced profiles. The permitted coefficient is therefore zero times ; symmetry alone does not establish nonzero phase pinning for the equations actually supplied.
The same normalization makes the remaining coefficients explicit. Since , . The quadratic second harmonic cancels for , while the uniform correction is . Projecting gives . Thus for the literal model and this temperature normalization,
A generic nonzero would require a nonvanishing projection in an amended physical model or a different forcing structure. It is still meaningful to classify the real-coefficient amplitude equation requested independently.
Write . Then and . These are a gradient flow for , so local minima give stable equilibrium points. At the origin the two eigenvalues are and . The origin has exponential asymptotic stability if , retains asymptotic stability with algebraic decay at , and is unstable if . At equality, obeys , since both linear coefficients are nonpositive. Integrating this inequality proves attraction even in the zero-eigenvalue direction.
For the stable nonzero equilibrium points are real; for they are imaginary:
The real branch has Jacobian matrix eigenvalues ; the imaginary branch has . The oppositely aligned branch, when it exists, is a saddle equilibrium. No mixed real-imaginary nonzero equilibrium is possible when .
For , the origin is stable for , with algebraic decay at zero. If , the circle is radially attracting. Each point has Lyapunov stability but has a neutral phase direction, so it does not have individual asymptotic stability; the circle has orbital stability. This is the literal model's unpinned family. The general nonzero- branches instead exhibit phase locking to one of two phases separated by .
Take and . Taking the divergence of the Stokes flow momentum equation gives ; applying the Laplacian to its vertical component then eliminates pressure and gives . The forced velocity and temperature are , so their advection product is and drops out at leading order. The steady heat equation reduces to .
Consequently the boundary value problem for the forced profiles is
Its six boundary conditions are
The horizontal velocity field is , which directly verifies the incompressible flow and stress-free boundary condition. In the detuning used subsequently, may be replaced by in this leading problem, since its correction enters at order . The vertical profiles depend linearly on .
For reference, the unforced free-slip convection neutral curve is , with and . At the neutral curve has , so the forced second horizontal harmonic is noncritical at , and the stated linear problem has no critical-mode resonance.
Take the layer depth as the length unit, its thermal diffusion time as the time unit, and its imposed temperature difference as the temperature unit. Then is the velocity field, the pressure after the conductive hydrostatic pressure has been subtracted, the departure from the conductive temperature , the upward unit vector, and the Rayleigh number. Here is the dimensional depth, the kinematic viscosity, the thermal diffusivity and the thermal expansion coefficient. The gradient, divergence and Laplacian act on the dimensionless coordinates. The term in the heat equation is advection of the background gradient, since . Mass density variations enter only the buoyancy term under the Boussinesq approximation; the incompressible flow constraint is .
For fixed, impermeable, stress-free plates, the no-penetration boundary condition, stress-free boundary condition and perfectly conducting thermal boundary condition are
For a nonzero horizontal Fourier mode, incompressibility gives , so the velocity conditions are equivalently on both plates. A uniform horizontal velocity is not fixed by these boundary conditions; take its mean to be zero when discussing the stationary forced pattern.
The printed small-Prandtl-number label does not generally justify these equations. In thermal-time units the full momentum equation contains , where the Prandtl number is . Dropping that term gives infinite-Prandtl-number convection or another explicitly justified Stokes flow approximation. It is not the general limit. For example the horizontal shear , , satisfies the full linear momentum equation and the plate conditions, and persists at fixed as . The printed inertia-free equation would instead require this shear to vanish. The following parts solve the equations actually displayed; applying them at small Prandtl number needs an additional slaving or creeping-flow ordering.
First make explicit the usual method of steepest descent hypotheses: the saddle point is simple, and ; a contour deformation leads to the two descent arms with one clockwise half-indentation around the pole; and endpoints or other contour portions contribute smaller terms. The sign of the exponential matters: , so decay is toward valleys of .
Use a local analytic function as coordinate. Nondegeneracy allows an analytic square root after factoring the double zero of , so the coordinate can be chosen with
The sign of is selected by the orientation of the contour integral, with running from negative to positive real values. Matching cubic coefficients gives . The transformed amplitude therefore has the Laurent series
The Cauchy principal value contribution of the pole on the two real arms is zero by oddness. Its clockwise semicircular indentation contributes , since there. The regular term contributes the Gaussian integral . Consequently the simple saddle coinciding with a simple pole has the additive expansion
The remainder written here assumes the other contour contributions are negligible to that order. Otherwise the local result has to be supplemented by them. The next local correction is generically times the displayed exponential. If , it can vanish; odd Taylor terms integrate to zero between the symmetric arms. When , the leading term gives the printed asymptotic equivalence. When , asymptotic equivalence to zero is not meaningful: the additive formula is still valid and a regular saddle-point approximation may lead instead.
Nondegeneracy is necessary and is not explicitly printed. For a counterexample to the unrestricted reading, take , and . There is one critical point, but it is degenerate. Choose the contour from to , with a clockwise quarter-circle avoiding zero. Both rays are decaying valleys for . Their straight-ray contributions cancel, and the indentation gives , not . Thus the requested coefficient requires the simple saddle point and contour deformation assumptions above; the mere existence of one critical point is insufficient.
Apply the method of multiple scales with independent fast time and slow time . A preliminary slow time produces no resonant forcing at first order, so its leading amplitudes are constant; the first nontrivial modulation is at . Write , with
At order , the inhomogeneous linear differential equation is . Absorbing its homogeneous part into the amplitudes gives
At order , the forcing is . Its resonant Fourier series coefficients are
The solvability condition in the method of multiple scales sets both coefficients to zero, giving the amplitude equations. Define
For bounded initial conditions , , the requested leading approximation is
This holds on bounded intervals of , hence , with the constant in the error allowed to depend on the bounded interval. If desired, retain the displayed and adjust by to match initial conditions at that order. No particular initial conditions were supplied.
Since , the matrix exponential is when . When , put and replace this with . At it is . Thus the leading second instability tongue of a weak Mathieu oscillator has exponentially growing solutions for , and the leading modulation is bounded for every initial condition precisely when
At either displayed endpoint, is a nonzero nilpotent operator; some initial conditions give a secular term proportional to . This leading calculation alone must not be mistaken for an exact finite- stability decision at the endpoints.
For an exact small- interpretation, put . The standard Mathieu equation parameters are and . Its two relevant Mathieu characteristic values are obtained by projecting onto even and odd Fourier series modes. Near the unperturbed value , the even modes have diagonal entries and adjacent couplings in an orthonormal basis; the odd modes have diagonal entries and couplings . Substituting into their characteristic polynomials gives
Higher modes of the Fourier series first contribute at higher order. Therefore the fourth-order edges of the second Mathieu instability tongue are
Floquet theory gives boundedness of every solution outside the closed gap, or , in this fixed- neighborhood. At the actual gap edges the second independent solution is generically unbounded. In particular, if is fixed exactly at or , it lies just outside the true gap and is stable for each sufficiently small positive . The strict boxed criterion is the robust leading-order, uniformly separated criterion; the endpoints require this higher-order refinement, and their amplification bounds need not remain uniform as .
Use the characteristic polynomial in the convention :
For fixed with , all three limiting eigenvalues are simple. Expanding this polynomial at each root, or applying first-order perturbation of a simple eigenvalue, gives
These are fixed- asymptotic expansions, not uniform through . At the displayed correction to vanishes because is an exact root for every ; there is no second nonzero term for that branch.
The two roots near interact when the correction is comparable to their unperturbed gap . Thus the distinguished limit is , with bounded real . Set and . The polynomial becomes
At orders and ,
With , the two interacting eigenvalues are
The correction to the pair is included to resolve the next perturbative term as well as the leading splitting. In particular the previously excluded fixed case is :
The half-integer powers reflect square-root splitting of a defective double eigenvalue: the limiting matrix has a nontrivial Jordan block at .
For matching, write and take the overlap . For , , , while , . Consequently the distinguished branches reduce to
The regular coefficient satisfies , proving agreement. For the labels and exchange roles and the same matching holds. The isolated negative branch also matches because . As a spectral check, positive diagonal symmetrization of a matrix transforms this tridiagonal matrix to a real symmetric matrix with off-diagonal entries and ; its eigenvalues are real for positive .
For freezing, the liquid concentration decreases away from the interface, so the liquidus increases there. Constitutional supercooling means somewhere ahead of the interface. Since these temperatures agree at the interface, its local onset is
For the similarity solution and salt rejection condition, this is exactly
It is also the global criterion for this freezing solution when and : the ratio is a positive constant times . For this ratio decreases with . If the ratio is at most one at the interface, cannot become negative; if it is greater than one there, a supercooled interval appears immediately ahead. Equality marks onset, not a finite supercooled interval.
For fixed positive and , the left side of the inequality is and the right side is positive of order one, so freezing is constitutionally supercooled for sufficiently small . To locate the boundary between the two regimes, resolve the smaller scale , . The salt equation and liquidus then give
Here , so the thermal equation improves to ; this justifies the second expansion including latent heat. Also and the left side of the exact criterion is . Therefore the small-diffusivity constitutional-supercooling threshold is
This printed condition is an asymptotic onset condition, rather than an exact finite- inequality. In its transition window the complete algebraic system and exact gradient criterion must be used.
Let . The salt equation in the saline Stefan problem gives . Since the complementary error function is positive on the real axis and , the signs of and agree. Thus freezing is equivalent to , or . Using the leading thermal balance,
at leading order away from a vanishing temperature difference. In fact the exact onset is also : at the salt balance forces , and the Stefan condition then forces . The solution with has a stationary interface and balancing conductive heat fluxes, despite nontrivial diffusion of the initial temperature discontinuity. For completeness, the exact freezing criterion follows without relying on that local crossing. By the salt-rejection function for a saline Stefan front, decreases continuously from to as positive increases. If , let be where . On , the solid-side thermal term decreases, the liquid-side term increases, and the latent-heat term increases: the factors and decrease and increase respectively. These monotonicities follow from the positive integral representation of . The thermal residual starts at and is negative at , so a positive root exists uniquely precisely when . Beyond both thermal contributions to freezing are negative, so no further root is possible. If , the same sign argument excludes every positive root.
The temporal integration by parts terms are
The terminal states are free and the objective has no explicit terminal-velocity dependence. Therefore
The factor two follows from the displayed objective without a prefactor. The initial scalar field perturbation is fixed, so and there is no independently prescribed adjoint-scalar condition at . Its value there is obtained by backward integration. Independent variation of the initial velocity state gives .
Variation of the candidate in the energy-augmented functional gives the fixed-energy initial-condition optimality condition. For the usual divergence-free no-slip control space , write for its orthogonal projection, the Leray-Helmholtz projection in the usual incompressible velocity space. This is the standard formal control condition when pressure is determined by incompressible flow and normal momentum. If the independently imposed direct pressure flux is retained as an additional constraint, the admissible control variations must also satisfy its compatibility conditions along the trajectory; they need not fill the usual space . The standard projection formula alone does not prove stationarity for that more restricted problem. Then
For the natural solenoidal adjoint space, . Equivalently, its component tangent to the sphere in a normed vector space of fixed kinetic energy vanishes:
Here is needed for the sphere in a normed vector space to be regular and for division by . The multiplier is real and can have either sign; normalizing the initial adjoint with a prescribed positive sign is not a general necessary condition for minimization. If , the only feasible initial velocity is zero and this tangent formula for the sphere in a normed vector space is inapplicable. First-order stationarity alone also allows maxima or saddles; a local minimum requires the appropriate nonnegative constrained second variation.
The spatial boundary terms vanish with periodic adjoints in , homogeneous no-slip boundary conditions at , and there. To see this, velocity variations vanish at the wall while their normal derivatives need not, so the viscous boundary term forces the adjoint velocity to vanish. Scalar variations have zero normal derivative but free values, forcing the adjoint scalar's normal derivative to vanish. Pressure variation gives adjoint incompressible flow; no independent terminal or initial datum is assigned to . Its additive time-dependent constant may be fixed by a zero spatial mean.
In particular, a homogeneous Neumann condition for the direct pressure does not by variational transposition require . For a smooth adjoint solution, the wall-normal adjoint momentum equation instead supplies
at the flat walls, because the direct scalar normal derivative is zero there. A zero adjoint-pressure derivative would be an additional compatibility choice, valid only when this right-hand side vanishes.
There are also literal direct-data compatibility qualifications in the printed setup. The initial total scalar field is , whose wall derivative is . It is extraordinarily small but mathematically not zero. Thus exact initial scalar data and exact zero wall flux are incompatible for a classical solution smooth at . One may use a parabolic mild solution of an abstract Cauchy problem with the boundary condition enforced for , or replace the initial profile by an exactly compatible smooth one; these are conventions, not an equality . The reference scalar field is also not a stationary profile of the diffusion equation at finite : the full term must be retained.
The pressure compatibility at a no-slip wall is similarly important. Direct normal momentum gives . The separately imposed zero pressure derivative requires this right side to vanish. It is not automatic for arbitrary fixed-energy initial velocities. For example, the divergence-free no-slip field generated by the periodic stream function with has , and , while . With adjusted to any positive energy, no classical solution smooth up to the initial wall can satisfy that extra pressure condition. The displayed adjoints are the formal necessary equations along admissible smooth trajectories; a well-posed physical formulation normally determines pressure from incompressible flow and normal momentum, rather than imposing independent homogeneous pressure flux for every control.
Use , , , and . Work along a sufficiently regular direct trajectory satisfying the given constraints. The zero spatial mean of the scalar field is conserved by incompressible flow, impermeable walls and zero scalar flux, so minimizing is equivalent to minimizing scalar variance, up to the fixed domain volume.
The printed functional fixes the initial state to a candidate ; it contains no term that enforces its kinetic energy. For the optimization over that candidate, add the real Lagrange multiplier constraint
Equivalently, one can restrict all control variations to the sphere in a normed vector space of fixed kinetic energy. This term changes the initial-control optimality condition, not the interior adjoint equations.
Let , and . Linearization of the momentum and scalar transport residuals gives
Both appearances of the perturbation velocity in the nonlinear momentum term have been differentiated. In particular, the coefficient is the gradient of the total velocity, not just the base shear.
For the negative-constraint convention of the functional, integration by parts gives the interior coefficients of as
Thus the adjoint equations for Boussinesq scalar mixing are
The transpose is essential: the th component of is . The coupling transposes advection of the scalar by a velocity perturbation; transposes buoyancy feedback. Dropping the latter would give a passive scalar adjoint, not the active scalar problem.
These equations are integrated backward, not forward. If , they read
with direct coefficients evaluated at . Both terms from the diffusion equation now have the usual forward sign in . A direct-adjoint looping method stores or reconstructs the forward trajectory, solves these equations backward, and uses the initial adjoint as the control gradient. The endpoint and fixed-energy conditions below give necessary conditions for a local optimizer, not a global optimality theorem.

Pinned article: Introduction to the OurBigBook Project

Welcome to the OurBigBook Project! Our goal is to create the perfect publishing platform for STEM subjects, and get university-level students to write the best free STEM tutorials ever.
Everyone is welcome to create an account and play with the site: ourbigbook.com/go/register. We belive that students themselves can write amazing tutorials, but teachers are welcome too. You can write about anything you want, it doesn't have to be STEM or even educational. Silly test content is very welcome and you won't be penalized in any way. Just keep it legal!
We have two killer features:
  1. topics: topics group articles by different users with the same title, e.g. here is the topic for the "Fundamental Theorem of Calculus" ourbigbook.com/go/topic/fundamental-theorem-of-calculus
    Articles of different users are sorted by upvote within each article page. This feature is a bit like:
    • a Wikipedia where each user can have their own version of each article
    • a Q&A website like Stack Overflow, where multiple people can give their views on a given topic, and the best ones are sorted by upvote. Except you don't need to wait for someone to ask first, and any topic goes, no matter how narrow or broad
    This feature makes it possible for readers to find better explanations of any topic created by other writers. And it allows writers to create an explanation in a place that readers might actually find it.
    Figure 1.
    Screenshot of the "Derivative" topic page
    . View it live at: ourbigbook.com/go/topic/derivative
  2. local editing: you can store all your personal knowledge base content locally in a plaintext markup format that can be edited locally and published either:
    This way you can be sure that even if OurBigBook.com were to go down one day (which we have no plans to do as it is quite cheap to host!), your content will still be perfectly readable as a static site.
    Figure 5. . You can also edit articles on the Web editor without installing anything locally.
    Video 3.
    Edit locally and publish demo
    . Source. This shows editing OurBigBook Markup and publishing it using the Visual Studio Code extension.
  3. https://raw.githubusercontent.com/ourbigbook/ourbigbook-media/master/feature/x/hilbert-space-arrow.png
  4. Infinitely deep tables of contents:
    Figure 6.
    Dynamic article tree with infinitely deep table of contents
    .
    Descendant pages can also show up as toplevel e.g.: ourbigbook.com/cirosantilli/chordate-subclade
All our software is open source and hosted at: github.com/ourbigbook/ourbigbook
Further documentation can be found at: docs.ourbigbook.com
Feel free to reach our to us for any help or suggestions: docs.ourbigbook.com/#contact