We present a formalization of quasi-compact and quasi-separated schemes (qcqs-schemes) in the Cubical Agda proof assistant. We follow Grothendieck's functor of points approach, which defines schemes, the quintessential notion of modern algebraic geometry, as certain well-behaved functors from commutative rings to sets. This approach is often regarded as conceptually simpler than the standard approach of defining schemes as locally ringed spaces, but to our knowledge it has not yet been adopted in formalizations of algebraic geometry. We build upon a previous formalization of the so-called Zariski lattice associated to a commutative ring in order to define the notion of compact open subfunctor. This allows for a concise definition of qcqs-schemes, streamlining the usual presentation as e.g. given in the standard textbook of Demazure and Gabriel. It also lets us obtain a fully constructive proof that compact open subfunctors of affine schemes are qcqs-schemes.

Abstract This is a foundation for algebraic geometry, developed internal to the Zariski topos, building on the work of Kock and Blechschmidt (Kock (2006) [I.12], Blechschmidt (2017)). The Zariski topos consists of sheaves on the site opposite to the category of finitely presented algebras over a fixed ring, with the Zariski topology, that is, generating covers are given by localization maps for finitely many elements $f_1,\dots, f_n$ that generate the ideal $(1)=A\subseteq A$ . We use homotopy-type theory together with three axioms as the internal language of a (higher) Zariski topos. One of our main contributions is the use of higher types – in the homotopical sense – to define and reason about cohomology. Actually computing cohomology groups seems to need a principle along the lines of our “Zariski local choice” axiom, which we justify as well as the other axioms using a cubical model of homotopy-type theory.