A central topic of study in analytic number theory is the behaviour of the Riemann zeta function. Many theorems and conjectures in this area are closely connected to concepts from probability theory. In this talk, we will discuss several results on the typical size of the zeta function on the critical line, over different scales. Along the way, we will see the role that is played by some probabilistic phenomena, such as the central limit theorem and multiplicative chaos.

It is known that there exists certain randomness in the values of the Moebius function. It is widely believed that this randomness predicts significant cancellations in the summation of the Moebius function times any 'reasonable' sequence. This rather vague principle is known as an instance of the 'Moebius randomness principle'. Sarnak made this principle precise by identifying the notion 'reasonable' as deterministic. More precisely, Sarnak's Moebius Disjointness Conjecture predicts the disjointness of the Moebius function from any arithmetic functions realized in any topological dynamical systems of zero topological entropy. In this talk, I will firstly introduce some background and progress on this conjecture. Secondly, I will talk about some of my work on this. Thirdly, I will talk some related problems to this conjecture.

We will first introduce the modular method for solving Diophantine Equations, famously used to
prove the Fermat Last Theorem. Then, we will see how to generalize it for a totally real number field $K$ and
a Fermat-type equation $Aa^p+Bb^q=Cc^r$ over $K$. We call the triple of exponents $(p,q,r)$ the 
signature of the equation. We will see various results concerning the solutions to the Fermat equation with
signatures $(r,r,p)$ (fixed $r$). This will involve image of inertia comparison and the study of certain
$S$-unit equations over $K$. If time permits, we will discuss briefly how to attack the very similar family
of signatures $(p,p,2)$ and $(p,p,3)$. 

Let $p$ be a prime, let $1 \le t < d < p$ be integers, and let $S$ be a non-empty subset of $\mathbb{F}_p$ (which may be thought of as being $\{0,1\}$). We will establish that if a polynomial $P:\mathbb{F}_p^n \to \mathbb{F}_p$ with degree $d$ is such that the image $P(S^n)$ does not contain the full image $A(\mathbb{F}_p)$ of any non-constant polynomial $A: \mathbb{F}_p \to \mathbb{F}_p$ with degree at most $t$, then $P$ coincides on $S^n$ with a polynomial $Q$ that in particular has bounded degree-$\lfloor d/(t+1) \rfloor$-rank in the sense of Green and Tao, and has degree at most $d$. Likewise, we will prove that if the assumption holds even for $t=d$ then $P$ coincides on $S^n$ with a polynomial determined by a bounded number of coordinates and with degree at most $d$.

In 1993, Erdős, Sárközy and Sós posed the question of whether there exists a set $S$ of positive integers that is both a Sidon set and an asymptotic basis of order $3$. This means that the sums of two elements of $S$ are all distinct, while the sums of three elements of $S$ cover all sufficiently large integers. In this talk, I will present a construction of such a set, building on ideas of Ruzsa and Cilleruelo. The proof uses a powerful number-theoretic result of Sawin, which is established using cutting-edge algebraic geometry techniques.

Serre's conjecture (now a theorem) predicts that an irreducible 2-dimensional odd
Galois representation of $\mathbb Q$ with coefficients in $\bar{\mathbb F}_p$ comes from the mod p reduction of
a modular form. A key feature is that two modular forms of different weights can have the same
mod p reduction. Fixing a modular form $f$, the weight part of Serre's conjecture seeks to find all
the possible weights where one can find a modular form congruent to $f$ mod $p$. The recipe for these
weights was conjectured by Serre, and it depends only on the local Galois representation at $p$. I
will explain the ideas involved in Edixhoven's proof of the weight part, and if time allows, I
will briefly say something about what the generalizations beyond $\operatorname{GL}_2/\mathbb Q$ might look like.