Homotopy Type Theory Sandbox (Rev #57, changes)

Showing changes from revision #56 to #57: Added | Removed | Changed

Euclidean semirings

Given a additively cancellative commutative semiring $R$, a term $e:R$ is left cancellative if for all $a:R$ and $b:R$, $e \cdot a = e \cdot b$ implies $a = b$.

A term $e:R$ is right cancellative if for all $a:R$ and $b:R$, $a \cdot e = b \cdot e$ implies $a = b$.

An term $e:R$ is cancellative if it is both left cancellative and right cancellative.

The multiplicative submonoid of cancellative elements in $R$ is the subset of all cancellative elements in $R$

A Euclidean semiring is a additively cancellative commutative semiring $R$ for which there exists a function $d \colon \mathrm{Can}(R) \to \mathbb{N}$ from the multiplicative submonoid of cancellative elements in $R$ to the natural numbers, often called a degree function, a function $(-)\div(-):R \times \mathrm{Can}(R) \to R$ called the division function, and a function $(-)\;\%\;(-):R \times \mathrm{Can}(R) \to R$ called the remainder function, such that for all $a \in R$ and $b \in \mathrm{Can}(R)$, $a = (a \div b) \cdot b + (a\;\%\; b)$ and either $a\;\%\; b = 0$ or $d(a\;\%\; b) \lt d(g)$.

Non-cancellative and non-invertible elements

Given a ring $R$, an element $x \in R$ is non-cancellative if: if there is an element $y \in \mathrm{Can}(R)$ with injection $i:\mathrm{Can}(R) \to R$ such that $i(y) = x$, then $0 = 1$. An element $x \in R$ is non-invertible if: if there is an element $y \in R^\times$ with injection $j:R^\times \to R$ such that $j(y) = x$, then $0 = 1$.

 On real numbers and square roots

There is a significant difference between square roots and $n$-th roots. Square roots are the inverse operation of the diagonal $f(x, x)$ for any binary operation $f$, while $n$-th roots are inverse operations of the $n$-dimensional diagonals $g(x, x, \ldots, x)$ for $n$-ary operations, which we typically do not formally talk about in typical practice for rings, etc…

Derivatives

Given an Archimedean ordered integral domain $A$ such with that element the dyadic rationals$\mathrm{<span><del\; class="diffmod">\; 𝔻</del><ins\; class="diffmod">\; \gamma </ins></span>}<span><del\; class="diffmod">\; \subseteq </del><ins\; class="diffmod">\; :</ins></span>A$ \mathbb{D} \gamma:A \subseteq A , a pointwise continuous function$\mathrm{<span><del\; class="diffmod">\; f</del><ins\; class="diffmod">\; p</ins></span>}:A0<span><del\; class="diffmod">\; \to </del><ins\; class="diffmod">\; <</ins></span>\mathrm{<span><del\; class="diffmod">\; A</del><ins\; class="diffmod">\; \gamma </ins></span>}$ f:A p:0 \to \lt A \gamma , is and$q:\gamma \lt 1$, a function $f:A \to A$ is pointwise differentiable if it comes with a function $\partial D(f):A\to A$ \partial(f):A D(f):A \to A called the derivative and a function $M:{\mathrm{<span><del\; class="diffmod">\; 𝔻</del><ins\; class="diffmod">\; A</ins></span>}}_{+}\to {\mathrm{<span><del\; class="diffmod">\; 𝔻</del><ins\; class="diffmod">\; A</ins></span>}}_{+}$ M:\mathbb{D}_+ M:A_+ \to \mathbb{D}_+ A_+ on the positive dyadic elements rationals of called the$A$ called the modulus of differentiability such that for every positive dyadic element rational$ϵ:{\mathrm{<span><del\; class="diffmod">\; 𝔻</del><ins\; class="diffmod">\; A</ins></span>}}_{+}$ \epsilon:\mathbb{D}_+ \epsilon:A_+ , for every term element$h:A$ such that $0 \lt \max(h, -h) \lt M(\epsilon)$ , and for every term element$x:A$,

Given an Archimedean ordered integral domain $A$ with element $\gamma:A$, $p:0 \lt \gamma$, and $q:\gamma \lt 1$, a function $f:A \to A$ is smooth if it comes with a sequence $D^{(-)}(f):\mathbb{N} \to (A \to A)$ called the iterated derivative sequence and a sequence of functions $M^{(-)}:\mathbb{N} \to (A_+ \to A_+)$ in the positive elements of $A$ called the moduli sequence of iterated differentiability, such that

• $D^{0}(f) = f$

• for every natural number $n:\mathbb{N}$, for every positive element $\epsilon:A_+$, for every element $h:A$ such that $0 \lt \max(h, -h) \lt M^{n+1}(\epsilon)$, and for every element $x:A$,

  $$\max(h D^{n + 1}(f)(x) + D^{n}(f)(x) - D^{n}(f)(x + h), D^{n}(f)(x + h) - D^{n}(f)(x) - h D^{n + 1}(f)(x)) \lt M^{n}(\epsilon)$$

Algebra

Given a commutative ring $R$, there is a commutative ring $A$ where $R$ is a subring of $A$, with a function $(-)\circ(-):A \times A \to A$ called composition, a term $x:A$ called the composition identity, a function $S_{(-)}:R \times A \to A$ called the shift, and a function $\partial:A \to A$ called the derivative such that

rules for composition:

• for all $a:R$, $a \circ f = a$
• for all $f:A$, $f \circ x = f$
• for all $f:A$, $x \circ f = f$
• for all $f:A$, $g:A$, and $h:A$, $f \circ (g \circ h) = (f \circ g) \circ h$
• for all $f:A$, $g:A$, and $h:A$, $(f + g) \circ h = (f \circ h) + (g \circ h)$
• for all $f:A$, $g:A$, and $h:A$, $(f g) \circ h = (f \circ h) (g \circ h)$

rules for shifts:

• for all $a:R$ and $f:A$ $S_a f = f \circ (x - a)$

rules for derivatives:

• $\partial(x) = 1$
• for all $a:R$, $\partial(a) = 0$
• for all $f:A$ and $g:A$, $\partial(f + g) = \partial(f) + \partial(g)$
• for all $f:A$ and $g:A$, $\partial(f g) = \partial(f) g + f \partial(g)$

Symbolic representations of formal smooth functions on the entire domain.