Infinite Layered Frequency-Time Schedules with Divergent Gradients and Bounded Amplitudes

Hello. In the previous lesson, a single layer was amplified and then rotated so that its large temperature gradient survived while its leading vorticity contribution was reset. The remaining question is one of assembly: how can one repeat this indefinitely, with later layers becoming finer and stronger, without requiring an infinite amount of time or making the underlying fields themselves large?

This lesson builds a deliberately simplified ledger for that task. It is not yet a solution of the Boussinesq equations with smooth forcing. Rather, it isolates the numerical logic a genuine construction must satisfy: summable stage durations, summable field amplitudes, and nonsummable derivative sizes.

---

The finite-time scheduling requirement

Index the layers by $q=1,2,\dots$. Give layer $q$ a time interval of duration

$${\delta }_q=2^{-q}.$$

Let its activation time be

$$t_q=\sum _{j=1}^{q-1}{2}^{-j}=1-2^{-(q-1)}.$$

Thus the $q$-th layer occupies

$$I_q=[t_q,t_q+{\delta }_q].$$

The total available time is finite:

$$T=\sum _{q=1}^{\infty }{2}^{-q}=1.$$

The intervals become shorter and shorter, but their endpoints approach $T=1$. Therefore, an arbitrary number of layers can be completed before the terminal time.

There is also an essential regularity feature hidden in this elementary fact. If $t<1$, only finitely many stages have started by time $t$. Thus, provided each individual layer is smooth in space and time, the partial construction is smooth at every strictly preterminal time. Singular behaviour can emerge only in the limiting approach to $T$.

This is the temporal architecture behind a finite-time blowup construction:

1. Each layer is allotted less time than its predecessor.
2. The total of all allotted times remains finite.
3. Each successive layer carries a more extreme spatial scale.
4. At any time before the terminal instant, the system has undergone only finitely many extreme-scale operations.

The schedule by itself is easy. The substantive issue is choosing frequency and amplitude scales compatible with it.

---

A toy layer ledger

Work on a periodic coordinate $y$, and suppose that after its growth-and-reset stage, layer $q$ is held in the vertical form

$${\theta }_q(y)=a_q\sin (K_{q}y).$$

Vertical orientation is useful because it makes ${\partial }_x{\theta }_q=0$ in the idealized two-dimensional Boussinesq geometry. The large gradient is retained, but the completed layer is dormant as a leading vorticity source.

Choose

$$a_q=2^{-q},K_q=2^{2q}=4^q.$$

The layer’s temperature-gradient amplitude is then

$$G_q=a_q{K}_q=2^{-q}{2}^{2q}=2^q.$$

So the temperature amplitude decreases geometrically, but the gradient amplitude increases geometrically.

Quantity for layer $q$	Choice or resulting scale	Role
stage duration	${\delta }_q=2^{-q}$	makes the total time finite
temperature amplitude	$a_q=2^{-q}$	makes temperature contributions summable
frequency	$K_q=2^{2q}$	creates increasingly fine scales
gradient amplitude	$a_q{K}_q=2^q$	diverges with $q$
maximal vorticity excursion	$G_q{\delta }_q=1$	remains order one during a stage
induced velocity amplitude	$1/K_q=2^{-2q}$	becomes very small at high frequency

The middle row is the key consistency calculation. During the active part of the $q$-th stage, the wave model from the previous lesson gives schematically

$${\dot{b}}_q\sim a_q{k}_{q,1},$$

and therefore

$$|{\dot{b}}_q|\lesssim a_q{K}_q=G_q.$$

If the stage lasts ${\delta }_q$, then the vorticity amplitude accumulated before the designed reset has size at most

$$|b_q|\lesssim G_q{\delta }_q=2^q{2}^{-q}=1.$$

Thus the increasingly large source is active for an increasingly short time. The two effects balance.

By Biot–Savart scaling, the velocity generated by this oscillatory vorticity has amplitude

$$|u_q|\asymp \frac{|b_q|}{K_q}\lesssim 2^{-2q}.$$

At the end of the stage, the controlled rotation resets $b_q$ to zero at leading order, while retaining the large vertical temperature gradient. In this toy model, there is at most one currently active vorticity layer, and every completed layer is dormant.

That is already enough to see the intended asymmetry:

$$\boxed{\text{field amplitudes decrease or remain controlled,}\text{derivative amplitudes increase without bound.}}$$

---

Bounded temperature, unbounded gradient

After $N$ layers have been completed, write the stored temperature field as

$${\theta }^{(N)}(y)=\sum _{q=1}^N{2}^{-q}\sin (4^{q}y).$$

The field itself is uniformly bounded:

$${\Vert {\theta }^{(N)}\Vert }_{L^{\infty }}\le \sum _{q=1}^N{2}^{-q}<1.$$

This is simply the triangle inequality. Each new layer contributes a smaller temperature amplitude than the last, so the total remains bounded even if infinitely many layers are ultimately stored.

Now differentiate:

$${\partial }_y{\theta }^{(N)}(y)=\sum _{q=1}^N{2}^{-q}{4}^q\cos (4^{q}y)=\sum _{q=1}^N{2}^q\cos (4^{q}y).$$

At the common phase point $y=0$, every cosine equals $1$. Hence

$${\partial }_y{\theta }^{(N)}(0)=\sum _{q=1}^N{2}^q=2^{N+1}-2.$$

Consequently,

$${\Vert \nabla {\theta }^{(N)}\Vert }_{L^{\infty }}\ge 2^{N+1}-2.$$

The same collection of modes therefore has two radically different aggregate behaviours:

Aggregate quantity	Behaviour as more layers complete
$\Vert {\theta }^{(N)}{\Vert }_{L^{\infty }}$	stays below $1$
$\Vert \nabla {\theta }^{(N)}{\Vert }_{L^{\infty }}$	grows at least like $2^N$
active vorticity amplitude	remains order one in the toy schedule
velocity generated by active high-frequency vorticity	decreases rapidly with frequency

This is the core blowup pattern: a continuous, bounded field accumulates finer and finer oscillations at a rate that destroys any uniform derivative bound.

The use of $y=0$ was convenient rather than profound. It prevents cancellation among the completed gradients. A real construction cannot merely assume all waves are globally aligned in this way; it instead manages geometry, supports, phases, and error terms carefully. But the bookkeeping principle is identical.

---

Why the time scale and frequency scale must be chosen together

It would not be enough to say “take frequencies very large.” If we had kept stage durations fixed, say ${\delta }_q=1$, then the active vorticity excursion would be

$$G_q{\delta }_q=2^q,$$

which becomes uncontrollably large. Conversely, if we chose a duration much shorter than $2^{-q}$, the vorticity would be easy to control but the schedule might be unnecessarily restrictive.

The toy choice satisfies the balancing relation

$${\delta }_q\asymp G_q^{-1}.$$

It keeps the time-integrated forcing from each active layer at order one:

$$\text{forcing size}\times \text{active duration}\asymp G_q{\delta }_q\asymp 1.$$

This is analogous to controlling total exposure in a risk process: a very large instantaneous quantity need not create a large cumulative effect if the period over which it is exposed is sufficiently short. Here, however, the construction also preserves a permanent record of that episode in the form of a high-frequency temperature gradient.

There is an important limitation. A stage that raises a wave’s frequency from a moderate scale to $K_q=4^q$ over a time interval of length $2^{-q}$ requires increasingly intense deformation. In an affine model, the relevant strain rate is roughly comparable to

$$\frac{\log K_q}{{\delta }_q},$$

which grows rapidly with $q$. A toy schedule can declare that this happens; a theorem must show that the required background deformation is genuinely generated by the fluid and that the external forcing remains smooth. That is where the detailed multi-layer design enters.

---

---

The real schedule: faster than dyadic separation

The paper’s construction has the same logic as our dyadic toy model but uses substantially more aggressive scales. Its frequencies satisfy a recursion of the form

$${\lambda }_q={\lambda }_{q-1}^{Q_q},$$

where $Q_q$ is already large and increases with $q$. Thus later frequencies are separated from earlier ones by far more than a fixed factor such as $4$.

The theorem’s estimates encode the following schematic facts:

$$P_q\lesssim {\lambda }_q^{-7/8},A_q\gtrsim {\lambda }_q^{1/8}.$$

Here $P_q$ is the temperature amplitude of layer $q$, while $A_q$ is its temperature-gradient amplitude. The distinction is exactly the one illustrated by $a_q$ and $a_q{K}_q$ in the toy model.

Because the phase-gradient length remains comparable to one, the physical frequency factor ${\lambda }_q$ gives the relation

$$A_q\sim {\lambda }_q{P}_q.$$

So a very small temperature mode can still carry a large gradient:

$${\lambda }_q\cdot {\lambda }_q^{-7/8}={\lambda }_q^{1/8}.$$

The scale ${\sigma }_q$ measures the growth rate available to subsequent stages. The theorem gives, schematically,

$${\sigma }_q^2\asymp {\lambda }_q^{1/8},$$

and also enforces extremely rapid separation, including a bound of the form

$${\sigma }_q\ge {\sigma }_{q-1}^3.$$

Meanwhile, the $q$-th stage has a lifetime bounded by a reciprocal earlier growth scale:

$$|I_q|\lesssim \frac{1}{{\sigma }_{q-2}}.$$

Because the ${\sigma }_q$ grow at least cubically from one stage to the next, these reciprocal lifetimes form a convergent series. This proves that the limiting terminal time $T^{\ast }$ is finite.

The paper’s choices are more elaborate than the toy relation ${\delta }_q=2^{-q}$, but the logic is the same:

Toy construction	Paper construction
$a_q=2^{-q}$	$P_q$ decays as a negative power of ${\lambda }_q$
$K_q=4^q$	${\lambda }_q$ grows extraordinarily rapidly
$a_q{K}_q=2^q$	$A_q$ grows as a positive power of ${\lambda }_q$
${\delta }_q=2^{-q}$	stage lifetimes are controlled by reciprocal $\sigma$-scales
reset sets old $b_q$ to zero	steering selects a pulse that achieves ${\Omega }_q(t_b^q)=0$
finite sum of durations equals $1$	the endpoints have a finite limit $T^{\ast }$

The actual construction uses smooth activation functions rather than sharp switching. This matters: a literal instruction to “turn on layer $q$ at $t_q$” would create a discontinuity in time and therefore nonsmooth forcing. Smooth cutoffs are made flat at their endpoints, so each stage joins the next smoothly. At any preterminal time, only finitely many stages have made nontrivial contributions, while the infinitely many future layers are either absent or present only at extremely small seed size.

---

What this toy schedule has established

The assembly mechanism can now be stated compactly.

For every layer:

1. Insert a very small temperature amplitude.
2. Amplify its frequency, creating a much larger gradient.
3. Keep the active period short enough that the vorticity accumulated during growth is controlled.
4. Use steering and rotation to reset its leading vorticity effect.
5. Retain the fine-scale temperature structure as a dormant record.
6. Start the next layer on a shorter time interval and at a much higher frequency.

The two convergences that make the design possible are different:

$$\sum _q{\delta }_q<\infty$$

puts infinitely many stages before a finite terminal time, while

$$\sum _q{a}_q<\infty$$

keeps the temperature field bounded.

At the same time, the derivative contributions are deliberately not summable:

$$a_q{K}_q/⟶0.$$

Indeed, in the toy choice they grow exponentially. The terminal obstruction is therefore not that the fluid fields must become pointwise infinite; it is that no uniform bound on the relevant derivative norm can survive all the accumulated fine scales.

Next, we will address the major omission in this global-wave picture: real layers must be localized to compact, nested affine cores. Spatial cutoffs make that possible, but they also create new residual errors that the forcing and correction scheme must absorb.