Time asymmetry, memory and observation: what a hidden variable can explain

1. What changes when an observer retains only part of a state?

My published paper Dual-Time Topological Geometry and the Emergence of Temporal Asymmetry in Non-Equilibrium Dynamics develops a two-parameter temporal formulation and studies projection, path dependence and memory representations. The website reading guide uses an exact hidden-mode calculation to make the reduction mechanism concrete.

Here I start with an even simpler example: a rotation. Its advantage is that the full dynamics preserves a quadratic quantity exactly. We can therefore see the effect of incomplete observation before adding any damping or relaxation.

Let a real state consist of two numbers, x and y, and choose a positive frequency ω. Their equations are

dx/dt = ωy,
dy/dt = −ωx.

Every initial pair has a unique future and past. The quantity E = (x² + y²)/2 is constant because its two derivative terms cancel. Reversing the parameter and changing the sign of y maps a solution to another solution. These are elementary, exact statements about the specified model.

Suppose the readout reveals only x. The preparations (0, 1) and (0, −1) produce the same initial readout, but their subsequent visible values are sin(ωt) and −sin(ωt). The visible present does not contain all the information needed to predict its own future.

2. Eliminating one variable produces an exact memory equation

Integrate the second equation from the chosen initial time zero:

y(t) = y(0) − ω ∫₀^t x(s) ds.

Substituting into the first equation gives a closed description of the visible coordinate once the hidden initial datum is supplied:

dx/dt = ωy(0) − ω² ∫₀^t x(s) ds.

The equation now contains a weighted history, with constant kernel −ω². It describes exactly the same rotation as the two local equations. No energy loss has been introduced. The dependence on earlier times records the predictive information missing from x alone.

Differentiating again yields d²x/dt² = −ω²x. A second-order equation requires both x(0) and its initial derivative. Since that derivative equals ωy(0), the additional initial datum restores precisely the information omitted by the instantaneous readout.

This gives three equivalent descriptions of the example: a first-order evolution of the full pair, a history-dependent equation for the visible coordinate, and a second-order equation with two initial data. The memory representation is informative about observation and state choice, but it does not uniquely establish an underlying temporal ontology.

3. A decaying memory requires an additional mechanism

Now modify the hidden equation by adding a relaxation rate λ > 0:

dx/dt = ωy,
dy/dt = −ωx − λy.

Multiplying the hidden equation by exp(λt) and integrating gives an exponentially weighted history. The resulting visible equation is

dx/dt = ωe^−λty(0)
− ω² ∫₀^t e^−λ(t−s)x(s) ds.

Older contributions now receive less weight. For this particular system, the same quadratic quantity satisfies dE/dt = −λy² ≤ 0. The added relaxation term supplies the decrease; changing the description to include a memory integral did not create it.

The decrease is a mathematical stability statement. Calling E a physical energy, identifying its destination, or deriving a thermodynamic entropy law requires an account of the reservoir, preparation and physical observables. A finite deterministic equation can also remain invertible as a map over finite time while having a decreasing quantity; invertibility alone is different from invariance under a specified time-reversal operation.

The published temporal framework motivates investigating how richer dynamics and projection can organize memory. These two examples are explanatory calculations with stated coefficients. They do not establish that every memory kernel dissipates, that every curvature produces a visible arrow, or that an arbitrary hidden sector has the required physical interpretation.

4. What a stronger dual-time model would have to explain

Within my proposed Single Monad Model and Duality of Time Theory, the research aim is to derive both the generative process and the completed records that an observer can access. In the later nested formulation, six inner levels organize generation and the seventh is outer completed observation. The published paper's two-coordinate mathematical domain needs an explicit correspondence with those roles; they cannot be identified simply by sharing the language of dual time.

A constructive next step would specify a source, its retained memory and a completion map, then calculate the visible response. The choice of readout would determine which preparations become indistinguishable, and the dynamics would determine whether that lost distinction reappears in a later measurement.

For example, prepare two source states with the same completed readout and compare their later responses to the same controlled perturbation. A predicted difference would show that the initial readout was not a sufficient state description. To support a stronger claim, the model must also explain which alternative descriptions can reproduce the result and which independently derived relation rules them out.

Another useful target is to derive several response features from the same source parameters. A relaxation time determined in one protocol could then fix a phase lag or transient in another, without adjusting it again. Such a relation would give experts a definite calculation to assess.

Question for collaboration: Which completion maps preserve enough predictive information to recover a local state description, and when do their restrictions enforce a measurable asymmetry? I welcome collaboration with experts in non-equilibrium dynamics, mathematical physics and the philosophy of time on that connection.

Sources and further reading

1. Mohamed Haj Yousef (2026). Dual-Time Topological Geometry and the Emergence of Temporal Asymmetry in Non-Equilibrium Dynamics. Mathematics, 14(5), 853. The published temporal geometry and memory-representation framework.
2. SMONAD: Temporal asymmetry—geometry, memory, and reduction. Explains hidden-mode elimination, stability and the distinction between geometric path dependence and visible response.
3. SMONAD: Eliminating hidden variables creates memory. Further detail on exact reduction and the initial-state term.
4. SMONAD: When memory can be represented by one time. Helps separate the usefulness of a memory representation from a claim of unique temporal ontology.