The left part of a pair of adjoint functors is one of two best approximations to a weak inverse of the other functor of the pair. (The other best approximation is the functor's right adjoint, if it exists. ) Note that a weak inverse itself, if it exists, must be a left adjoint, forming an adjoint equivalence.
A left adjoint to a forgetful functor is called a free functor; in general, left adjoints may be thought of as being defined freely, consisting of anything that an inverse might want, regardless of whether it works.
The concept generalises immediately to enriched categories and in 2-categories.
Given posets (or prosets) $C$ and $D$ and a monotone function $U: C \to D$, a left adjoint of $U$ is a monotone function $F: D \to C$ such that
for all $x$ in $C$ and $y$ in $D$.
Given locally small categories $C$ and $D$ and a functor $U: C \to D$, a left adjoint of $U$ is a functor $F: D \to C$ with a natural isomorphism between the hom-set functors
Given $V$-enriched categories $C$ and $D$ and a $V$-enriched functor $U: C \to D$, a left adjoint of $U$ is a $V$-enriched functor $F: D \to C$ with a $V$-enriched natural isomorphism between the hom-object functors
Given categories $C$ and $D$ and a functor $U: C \to D$, a left adjoint of $U$ is a functor $F: D \to C$ with natural transformations
(where $F;U$ etc gives the composite in the forwards, anti-Leibniz order) satisfying certain triangle identities.
Given a 2-category $\mathcal{B}$, objects $C$ and $D$ of $\mathcal{B}$, and a morphism $U: C \to D$ in $\mathcal{B}$, a left adjoint of $U$ is a morphism $F: D \to C$ with $2$-morphisms
satisfying the triangle identities.
Although it may not be immediately obvious, these definitions are all compatible.
Whenever $F$ is a left adjoint of $U$, we have that $U$ is a right adjoint of $F$.
Left adjoint functors preserve
colimits;