# How to say a given space is a manifold?

How to say a given space is a manifold?

The only thing that props in my mind is to check if every open set has a euclidean coordinate chart on it. But, what if the space I am dealing with is not fully understood apriori?

As in, how were the spaces of thermodynamic equilibrium states, phase and configuration spaces in classical mechanics, space-time (of general relativity) etc given a manifold structure? How to justify that every open set of these spaces correspond to an euclidean space?

Last edited:

How to say a given space is a manifold?

The only thing that props in my mind is to check if every open set has a euclidean coordinate chart on it. But, what if the space I am dealing with is not fully understood yet?

As in, how were the spaces of thermodynamic equilibrium states, phase and configuration spaces in classical mechanics, space-time (of general relativity) etc given a manifold structure? How to justify that every open set of these spaces correspond to an euclidean space?

You just need to satisfy the definition of a manifold for that space.

First, you probably need to guess the dimension. How many degrees of freedom does the space have after you've applied all the constraints? Once you understand the dimension, it should usually be clear how to break the space up (really, you can probably break it up in number of ways, as long as you cover the space. Once you've done that, you have a manifold. I image the hard part would be to show the manifold is smooth or satisfies other properties.

Even after I break it up into its covering set (meaning, I have laid out the open sets), how to justify that each of these open sets correspond to an euclidean chart?

Do we have to worry about differentiable-ness, because once I have assigned one chart to each open set I can then assign charts that will be $$C^\infty$$-compatible with the original and call it as the atlas.

Even after I break it up into its covering set (meaning, I have laid out the open sets), how to justify that each of these open sets correspond to an euclidean chart?

When picking out the charts, there is no real "justification" to do. You simply need to provide the chart itself. Find the function, x:M->R^n, which gives coordinates to each point. Then find the inverse function, x^-1:R^n->M, which maps points to coordinates.

Do we have to worry about differentiable-ness, because once I have assigned one chart to each open set I can then assign charts that will be $$C^\infty$$-compatible with the original and call it as the atlas.

Yes. If you want a smooth manifold, you must all the transition maps are smooth.

What kind of spaces are you working with in particular?

I dont have a specific space in mind. Let us take the example of thermodynamic systems (the axiomatic formulation of Caratheodory)...
The assumption that each state (or point in manifold) of the system be described by some variables (euclidean coordinates) is fine. But, the requirement that there be a neighbourhood of each such point which can be described by the same euclidean chart isnt obvious to me.

Or even talking about the generic space-time, it is fine to talk of 4 coords to represent each point in space-time, but what is the basis for assuming the same for an open neighbourhood of each such point?