Mathematics becomes most interesting to me when a formula stops being just a formula and starts describing a process.

Recently, I have been returning to differential equations from a very different perspective. Not as a collection of methods to remember, but as a language for describing systems that evolve with time.

And the more I read, the more interesting the questions become.

What if a process began long before the moment we started observing it?

What if its present state depends not only on what is happening now, but also on what happened before?

And if we are able to influence such a system, how do we decide what the best influence should be?

These questions sit close to the mathematical direction I am beginning to explore.

A system that changes with time

A simple way to describe the state of an evolving system is to write

where x represents position and t represents time.

The function y tells us what the system looks like at a particular place and at a particular moment.

For many physical and mathematical processes, partial differential equations describe how this state changes.

At first, this sounds straightforward: specify how the system begins, write the equation, and study what happens next.

But this immediately raises a question.

Do we always know where the process began?

What if there is no meaningful starting point?

A classical evolution problem often begins with an initial condition such as

At time t = 0, we know the state of the system. From there, the equation describes its future evolution.

But some processes may have started long before the moment that interests us.

In that case, choosing one artificial moment and calling it “the beginning” may not describe the situation very naturally.

Instead, we can imagine a process evolving over

There is still a moment T at which we may want to observe the system, but there is no finite initial time from which everything begins.

I find this idea surprisingly beautiful.

The mathematics does not ask us to invent a beginning simply because a model is easier to write that way. Instead, it asks how the behaviour of the system over its entire past can determine the state we observe now.

Can an equation remember?

There is another complication.

The present state of a system may depend not only on what is happening at time t, but also on what happened before t.

A simple way to imagine this is through an interval of past time,

The length of this interval may even change with time.

This introduces the idea of delay — or, more intuitively, memory.

The system carries part of its history with it.

That idea changes the way I look at an equation. It is no longer only a description of what happens “now”. The past becomes mathematically present inside the model.

In a sense, an equation can remember.

And what if we can influence the system?

Once a system evolves, remembers its past, and reacts to different influences, another natural question appears:

Can we control it?

Suppose a choice v changes the behaviour of the system. Different choices produce different states:

But not every possible influence is equally good.

Perhaps we want the system to approach a desired state. Perhaps we also want to avoid using an unnecessarily large control.

Then we need a way to evaluate each choice:

The functional J gives a numerical measure of how good a particular control is.

And now the question becomes an optimisation problem:

Which admissible control gives the best result?

This is where differential equations begin to meet optimal control.

The direction I am beginning to explore

The area I am currently entering lies near the intersection of several ideas:

parabolic equations, processes without classical initial conditions, systems with memory, and optimal control.

At this stage, I am not writing about new results.

I am trying to understand how these pieces fit together.

What assumptions are necessary?

Which methods already work?

What changes when memory is introduced?

Where do existing arguments stop working?

And what would be needed to make them work again?

For me, this is one of the most interesting parts of research: before there is a theorem, there is a period of learning how to formulate the right question.

Beyond the formulas

This is also what I want Beyond the Formulas to become.

Not a collection of textbook definitions.

Not a place where every question already has a polished answer.

But a space where I can follow mathematical ideas as they become clearer — through reading, research, intuition, questions and, eventually, proofs.

Because sometimes research does not begin with a theorem.

It begins with learning how to ask the right question.