Here is just a collection of definitions of various mathematical structures that I came across in my readings. I like to have this page to quickly remind me of what mathematicians mean when they say something.
Sets
See: set-theory-axioms
Algebraic Structures
Group
A group is a set of objects (group items) along with a binary operation that satisfies the following axioms, where :
- Identity element:
- Inverse elements:
- Associativity:
A group is said to be Abelian if the group operation is commutative (for all ):
Info
Group theory is often thought of as the theory of symmetry. Symmetry transformations (e.g. length preserving rotations) follow the properties above. Any two symmetry transformations combine into a third symmetry transformation (think, two rotations is also a rotation). The identity element is “do nothing”. The inverse element is “reverse the transformation” (think, rotate the other way). The ordering of symmetry transformations might matter (for a Non-Abelian group) but many symmetry transformations in sequence can be calculated in arbitrary groupings (associativity).
Example: Rotations of 3-D Euclidean space
The rotation group of 3-D Euclidean space is SO(3)—the special orthogonal group of dimension 3. This is a continuous group of dimension 3 (a Lie group) which can be, e.g., parameterized by the 3 Euler angles. This group is non-abelian (think, a rotation around the x-axis followed by a rotation around the z-axis is not the same if you flip the order) and can be represented by 3x3 orthogonal matrices () with . The “special” part (corresponding to the condition ) means we don’t consider reflections as proper rotations, or, equivalently, we only consider orientation-preserving rotations.
Ring
A ring is a set along with two binary operations and satisfying the following axioms, where :
- is an abelian group
- Associativity of multiplication:
- Multiplicative identity:
- Left distributivity:
- Right distributivity:
A ring is commutative if .
Example: Integers ( )
Check it, integers under addition satisfy all of these axioms.
Field
A field is a commutative ring in which every nonzero element has a multiplicative inverse:
Example: Rational numbers , Real numbers , Complex numbers .
Why not integers? (Hint: What’s ?)
Quaternions are not fields as they are not commutative.
Vector Space
A vector space is a collection of objects (vectors) over a field along with two operations and (we usually omit the symbol) that satisfy the following 8 axioms, where and :
- Associativity of addition:
- Commutativity of addition:
- Additive identity:
- Additive inverse:
- Compatibility of scalar multiplication:
- Multiplicative identity:
- Distributivity over vector addition:
- Distributivity over scalar addition:
Musing
Notice that properties 1-4 make the elements of the vector space an Abelian group over addition. So sometimes a (lazy lol) author might just use the words “Abelian group over addition” instead of listing out properties 1-4.
Module
A module is basically a vector space but you replace the field with a ring. Formally, a left -module over the ring is an Abelian group equipped with a scalar multiplication obeying properties 5-8 of the above ( and ):
- Compatibility of scalar multiplication:
- Multiplicative identity:
- Distributivity over vector addition:
- Distributivity over scalar addition:
I gave the definition of a left -module here so I could literally copy paste. Use your imagination for what a right -module means. The main thing to keep in mind for modules is if is a division ring or not (i.e. if every non-zero element in R has an inverse element). In the case that is a division ring then the module (and therefore all vector spaces, since fields are division rings) will have a basis. If is not a division ring, it could be (though it doesn’t have to be) that the module has no basis.
Example: The Hairy Ball
I’m gonna introduce some notation and concepts that aren’t all contained in this page. It gives you incentive to explore my other pages. Maybe you’ll find relevant definitions or maybe you won’t. In any case, it’s not that important. The standard example for when a module has no basis is the set of smooth vector fields on a 2-sphere over the ring of smooth functions (check for yourself that it satisfies the conditions of a ring). Crucially, is not a division ring since a function which is 0 at some point has no inverse (taking it to the constant function which is everywhere 1). I tried writing out the proof for why over does not admit a basis but it was too long. Just know it has to do with the Hairy Ball theorem and the fact that there are no everywhere non-vanishing smooth vector fields on the 2-sphere.
Musing
It’s unfortunate that we are reusing the letter here. Most of the time we’ll use it for manifold. Meh. You can tell how lazy I am by now right?
Algebra
An algebra over a field is a vector space equipped with a bilinear product . Bilinearity means the product is linear in each argument:
for all and . The product need not be associative or commutative.
Example 1:
matrices over (with being matrix multiplication)
Example 2:
Lie algebras with the anticommutative, , Lie bracket following the Jacobi identity: . A simple example of this is 3-D Euclidean space with the cross product.
Analytic Structures
TODO: Metric space, Normed space, Banach space.
Inner Product Space
An inner product space is a vector space over or equipped with an inner product satisfying the following axioms for all and :
- Conjugate symmetry:
- Linearity in the second argument:
- Positive-definiteness: , with
The inner product induces a natural norm , making every inner product space a normed space.
Mathematician's convention
Mathematicians take the inner product to be linear in the first argument instead, i.e. . The two conventions are equivalent up to a swap of arguments.
Example:
The familiar dot product is an inner product on .
Hilbert Space
A Hilbert space is a complete inner product space. What do we mean by complete? Suppose we have a countably infinite set of orthonormal basis vectors: We can create a vector space out of this by allowing all finite linear combinations:
where is a finite number. This will obey all the axioms of a vector space. The trouble comes when we consider an infinite sum . The standard axioms of vector spaces do not guarantee that this infinite sum exists or, even if it exists, will be part of the vector space. What we can do is complete by requiring that all square summable vectors must exist in the Hilbert space (i.e. Hilbert space is the completion of with respect to the norm topology). A square summable vector is a vector where .
Info
In physics, we work on square summable / square integrable functions in which happens to be separable (contains a countable set of orthonormal basis vectors). In a separable space, the above definition I gave is equivalent to the following formal definition:
Define a Cauchy sequence of vectors as a sequence where for any there exist such that Then a vector space is complete (and therefore a Hilbert space) if and only if every Cauchy sequence of vectors converges to a vector in the space. By convergence here we mean the standard definition of convergence: the sequence converges to if and only if . In plain language, “any time I have a sequence which is Cauchy, it converges to some limit inside my space”.
If the space is not separable (i.e. it must use an uncountable set of orthonormal basis vectors) then this definition in terms of Cauchy sequences is still valid but my earlier one is not. Note that we almost never use non-separable spaces in physics. Even unbound states (e.g. wave packets) still live in a separable Hilbert space.
TODO: Rigged Hilbert Spaces.
Measure-Theoretic Structures
-Algebra
A -algebra on a set is a collection of subsets of closed under countable set operations:
- Closed under complements:
- Closed under countable unions:
The pair is called a measurable space, and elements of are called measurable sets. A function is a measure if and is countably additive: for disjoint . The triple is then called a measure space.
Probability Space
A probability space is a measure space where . Here is the sample space (all possible outcomes), is the -algebra of events, and is the probability measure. A random variable is a measurable function , and its expectation is the Lebesgue integral:
TODO: Lebesgue measure, Proper Class, Category