I am quite confused as to what is the difference between a frame and a co-ordinate system. The wikipedia page was not very helpful for me. I would be very happy if someone could give me a non-rigorous idea about what exactly the difference is. My background involves basic differential geometry. I have also done some very basic differential topology and am aware of manifolds and some topological properties associated with them.

I'll try and elaborate a bit about the context. The Frenet frame fields are useful in that they express the rate of change of the unit vectors constituting the frame in terms of the vectors themselves. Also the fact that it is a "moving frame" is supposedly useful. I am thinking, this means analysis of the various properties associated with the curve is easier as the frame moves along with the observer.Though I couldn't explicitly point out what these properties are. Am I right??

Why exactly is this frame so useful??Also doesn't the notion of a frame reject path independence?

I apologise if my question seems a bit vague. I do hope its enough to convey what I am looking for. I am not much exposed to physics. But I would welcome answers involving examples from physics if it helps matters.

P.S.: If this can be tied in with notions about the coordinate basis of vector space, then all the better.I am looking for something all encompassing. I hope that doesnt come across as foolish.


2 Answers 2


A coordinate system is a homeomorphism from a neighborhood of $M$ to $\mathbb{R}^n$. This gives you coordinates and induces coordinate vector fields $\frac{\partial}{\partial x_i}$. These form a basis in each tangent space.

If you specify a metric on $M$, you will see that unless your manifold has no curvature, the basis in each tangent space induced by coordinates is not orthonormal. By performing Gram-Schmidt in each tangent space to the coordinate fields, you can create a "frame field" - $n$ orthonormal vector fields which form an orthonormal basis of each tangent space in this neighborhood. These fields cannot be integrated to coordinates (again because of curvature).

The short answer to your question: coordinate systems are charts to $\mathbb{R}^n$ that give coordinate functions locally on a manifold, while frames are vector fields that give orthonormal bases for each tangent space in a neighborhood of the manifold.

The Frenet frame is an orthonormal basis of the pullback of the tangent bundle $T\mathbb{R}^3$ via a curve in $\mathbb{R}^3$. It encodes information about the embedding of the curve in $\mathbb{R}^3$. (A frame field on a vector bundle with a bundle metric is a little bit more general than a frame field on the tangent bundle on a manifold, but the basic idea is still the same.)

  • $\begingroup$ Thanks, that is as close to an answer I was hoping for. Can you add anything about Frenet frame or moving frame though, as to how exactly they fit in into the scheme of things? $\endgroup$
    – Vishesh
    Oct 30, 2012 at 15:18

It is worth clarifying the difference between a moving frame on an abstract manifold and a moving frame along a submanifold inside a higher dimensional space.

A moving frame on an abstract manifold or on an open subset of the manifold is simply set of vector fields $V_1, \dots, V_n$ such that $V_1(x), \dots, V_n(x)$ form a basis of the tangent space $T_xM$ for each $x$ where the vector fields are defined. If the manifold is assumed to have a geometric structure, it is common to assume that the frame is somehow adapted to the geometric structure. For example, if there is a Riemannian metric, then it is often convenient to assume that the vector fields are orthonormal.

A moving frame on a submanifold $M^n$ inside a higher dimensional ambient space $A^N$ is a set of sections $V_1, \dots, V_N$ of the tangent bundle of the ambient space restricted to the submanifold. Again, if the ambient space has a geometric structure, then the frame is usually assumed to be somehow adapted to the geometric structure. For example, if $A^N$ is Euclidean $N$-space or some other Riemannian $N$-manifold, then it is often convenient to assume that the frame is orthonormal and that the first $n$ vectors are tangent to the submanifold and the last $N-n$ are normal.

It is possible to work with the moving frame directly but it is usually easier to work with the dual basis, which is a frame of sections of the cotangent bundle. This is because the exterior derivative is easier to compute with than the Lie bracket.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.