Newtonian force as a covariant or contravariant quantity

In summary, the conversation discusses the concept of force and its relationship to vectors and dual vectors in the context of Newtonian mechanics. This idea is presented by Burke in his book called Div, Grad, and Curl are Dead. It is mentioned that force is a 1-form and dual to vectors, which is supported by the statement that energy is a scalar and displacement is a contravariant vector. However, there is some disagreement about whether energy is truly a scalar or a scalar density. The conversation also touches on the issue of symmetry between vectors and their duals and the role of a metric in writing Newton's Second Law. Finally, it is mentioned that while the dual space of \mathbb{R}^{3} is isomorphic
  • #141
I don't agree. One has to distinguish between a representation of an object in particular coordinates, and the object itself. The typical physics index gymnastics notation refer to representations (components) of an object.

It is true that you could say that a tuple of of basis covariant vectors transforms as the components of a single contravariant vector. But this is not standard in physics, where we refer to transformation of the components. (See the Warning on p42 of http://books.google.com/books?id=DUnjs6nEn8wC&source=gbs_navlinks_s.)
 
Last edited:
Physics news on Phys.org
  • #142
bcrowell said:
Carroll (and by proxy physicists in general) is trying to simultaneously satisfy incompatible desires 1-5. A clear way of seeing this is that he wants to write [itex] \omega_{\mu}[/itex] as a synonym for [itex]\omega[/itex]. Now suppose that (in old-fashioned non-abstract index notation, indicated by the Greek indices), we have [itex]\omega_0=1[/itex] for [itex]\mu=1[/itex] and all the other components zero. Then we have [itex]\omega=dx^\mu[/itex].

No, you can't write that. But you can write [itex]\omega = dx^1[/itex], which is perfectly sensible.

But [itex] \omega_{\mu}[/itex] is a synonym for [itex]\omega[/itex], so can substitute in and make it [itex]\omega_{\mu}=dx^\mu[/itex].

That is also nonsense. At best you can write [itex]\omega_\mu = \delta_\mu{}^1[/itex].

There is nothing wrong or abusive about writing [itex]\omega = \omega_\mu \, dx^\mu[/itex]. The summation rule is being followed ("one upper and one lower index are summed"), and the indices are in places that indicate their purpose:

1. Lower indices on real-number-type quantities indicate which component of a covariant object.

2. Upper indices on 1-form-type quantities indicate which 1-form. Such indices are written upper in order to be consistent with the summation rule.

If you were to consistently stick with index notation (either abstract or not), you would never write something like [itex]\omega = \omega_\mu \, dx^\mu[/itex]. You would just talk about "the 1-form [itex]\omega_\mu[/itex]".

At any rate, the constant arguing about what exactly indices "mean" is one of the reasons I hate this notation. Such discussions don't lead anywhere useful. It's even worse when you try to use orthonormal frames and you get into "flat space indices" vs. "curved space indices".
 
  • #143
Just like we think of vectors at a point as a linear combination of derivations, we can write the dual vectors at a point as a linear combination of the dual basis to the derivations. We write [itex]V = V^{i}\partial _{i}[/itex], [itex]\omega = \omega _{i}dx^{i}[/itex] so that [itex]\omega(V) = \omega _{i}dx^{i}(V^{j}\partial _{j}) = \omega_{i}V^{j}dx^{i}(\partial _{j}) = \omega_{i}V^{j}\delta ^{i}_{j} = \omega _{i}V^{i}[/itex]. The initial two expressions are not meant to be inner products of some form. The indices on the derivations and their dual represent the elements of the basis and dual basis respectively, not components.
 
  • #144
bcrowell said:
Appeals to authority are useful only with respect to #5. Since we don't all have the same printed books on our bookshelves, it's useful to have an online reference that's accessible to all of us, so let's use the online version of Carroll: http://ned.ipac.caltech.edu/level5/March01/Carroll3/Carroll2.html

Well, your version of Carroll says the same thing as we have been saying:

We are now going to claim that the partial derivative operators [itex]\{\partial_{\mu}\}[/itex] at p form a basis for the tangent space Tp. (It follows immediately that Tp is n-dimensional, since that is the number of basis vectors.)

and

Therefore the gradients [itex]\{dx^\mu\}[/itex] are an appropriate set of basis one-forms; an arbitrary one-form is expanded into components as [itex]\omega = \omega_{\mu} dx^\mu[/itex].
 
  • #145
bcrowell said:
But I'm not clear on how to notate this idea that in both examples, the force 1-form is parallel to the surface. Do you represent the surface as, say, a 1-form created by taking the (infinite) gradient of a step function across the wall?

The constraint surface can be represented by an equation of the form f(Xi) = 0. The constraint surface in your second example is

f(φ, θ) = θ - φ = 0

A constraint force which is parallel to the constraint surface can be written as a multiple of df, since df is parallel to the constraint surface.

F = λdf = λ(dθ - dφ)
 
Last edited:
  • #146
bcrowell said:
No, I'm pretty sure you're wrong about this. The operator [itex]\partial_a=\partial/\partial x^a[/itex] is written with a lower index precisely because it's a covector and transforms like one. The notation isn't misleading. The notation shows exactly how the thing transforms. For instance, you can convince yourself of this by writing down a rescaling of the coordinate and applying the tensor transformation law.

Yeah, these conventions are confusing, but they make some sense if you think about it.

In differential geometry, a vector [itex]V[/itex] is defined via its effects on scalar fields. Specifically, if we choose a particular coordinate basis, we can write:

[itex]V(\Phi) = V^\mu \partial_\mu \Phi[/itex]

with the implicit summation convention. Now, if [itex]V[/itex] happens to be a basis vector [itex]e_\sigma[/itex], then that means that [itex]V_\sigma = 1[/itex], and all the other components are equal to 0. So we have, in this case:

[itex]e_\sigma(\Phi) = \partial_\sigma \Phi[/itex]

Since this is true for any [itex]\Phi[/itex], it's an operator equation:

[itex]e_\sigma = \partial_\sigma[/itex]
 
  • #147
micromass said:
The notation [itex]\frac{\partial}{\partial x^i}\vert_p[/itex] represents a tangent vector,and not a 1-form. A 1-form would look like [itex]dx^i\vert_p[/itex]. See "Introduction to Smooth Manifolds" by Lee.

Yeah, this stuff is pretty confusing. Even though [itex]\frac{\partial}{\partial x^\mu}[/itex] is a vector, the prototypical covector is [itex]\nabla \Phi[/itex] for some scalar field [itex]\Phi[/itex], which has components [itex]\frac{\partial}{\partial x^\mu} \Phi[/itex]

To complete the confusion, even though [itex]dx^\mu[/itex] is a 1-form, the prototypical vector is a tangent vector to a parametrized path [itex]P(\lambda)[/itex], which has components [itex]\dfrac{dx^\mu}{d\lambda}[/itex].
 
  • #148
stevendaryl said:
Yeah, this stuff is pretty confusing. Even though [itex]\frac{\partial}{\partial x^\mu}[/itex] is a vector, the prototypical covector is [itex]\nabla \Phi[/itex] for some scalar field [itex]\Phi[/itex], which has components [itex]\frac{\partial}{\partial x^\mu} \Phi[/itex]

To complete the confusion, even though [itex]dx^\mu[/itex] is a 1-form, the prototypical vector is a tangent vector to a parametrized path [itex]P(\lambda)[/itex], which has components [itex]\dfrac{dx^\mu}{d\lambda}[/itex].
It is confusing, but even more if one doesn't distinguish the geometrical object from its component representation. [itex]dx^\mu[/itex] is a 1-form, but transforms contravariantly, however if it had components other than unity they would transform covariantly.

So I'm still waiting for someone to explain why the rate of change of momentum is not naturally a tangent vector.
 
  • #149
TrickyDicky said:
It is confusing, but even more if one doesn't distinguish the geometrical object from its component representation. [itex]dx^\mu[/itex] is a 1-form, but transforms contravariantly, however if it had components other than unity they would transform covariantly.

So I'm still waiting for someone to explain why the rate of change of momentum is not naturally a tangent vector.

If momentum is defined to be [itex]\vec{P} = m \frac{\vec{dU}}{dt}[/itex], then it is naturally a vector. If it is defined via a Lagrangian [itex]L[/itex] by [itex]P_\mu = \frac{\partial}{\partial U^\mu}L[/itex], then it is naturally a covector.

Another way to see that momentum should be considered contravariant might be:

In curvilinear coordinates, the equations of motion for a free particle is

[itex]\dfrac{d}{dt} U_\mu = 0[/itex]

not

[itex]\dfrac{d}{dt} U^\mu = 0[/itex]
 
  • #150
I think the basic cause of the ugly inconsistency in Einstein notation is not that there's anything wrong with Einstein notation but that it's being mixed inappropriately with Sylvester's notation. Sylvester was the one who originally introduced terms like "contravariant" and "covariant." The pure Sylvester notation is described in this WP article: http://en.wikipedia.org/wiki/Contravariant_vector Unfortunately Sylvester's notation and terminology are much more cumbersome than Einstein notation, and it doesn't correspond to the way physicists customarily describe the objects themselves as having transformation properties, rather than the objects staying invariant while their representations transform.

In Einstein notation, a sum over up-down indices represents a tensor contraction that reduces the rank of an expression by 2. In Sylvester notation, a vector is expressed in terms of a basis as a sum over up-down indices, which is not a tensor product. Mixing Einstein and Sylvester notation produces these silly-looking things where it looks like we're expressing a covector as a sum of vectors or a vector as a sum of covectors.

The abstract index form of Einstein notation is carefully designed so that you *can't* inadvertently refer to a coordinate system; anything written in abstract index notation is automatically coordinate-independent. This means that in abstract index notation, you can't express a vector as a sum over basis vectors. You don't want to and you don't need to.

There is also an issue because in Einstein notation we refer to vectors as covariant and contravariant, whereas in Sylvester notation a given vector or 1-form is written in terms of components and basis vectors that are covariant and contravariant. Neither is right or wrong, but mixing them leads to stuff that looks like nonsense.
 
Last edited:
  • #151
Ok, I actually can't understand the Lewis notes I linked to in #127.

These guys http://www.google.com/url?sa=t&rct=...uLu-lMgYA63SlGua6NB5lOA&bvm=bv.41867550,d.b2I do it differently. In proposition 7.8.1 they do define a force via contraction with a vector field. However they contract with a symplectic 2 form, not a metric. I think the relevant definition of the 2 form is on p161.

Incidentally the Lewis notes say ad hoc "motivation" is that the Euler-Lagrange equations transform as the components of a one form. He say this is hand-waving because the EL equations are not a one-form.
 
  • #152
I guess the Lewis notes http://www.mast.queensu.ca/~andrew/teaching/math439/pdf/439notes.pdf are fine after all. Force is a one-form. What I was confused by is their notation F: R X TQ -> T*Q. All they mean is that a force is an assignment of a one form to every velocity at every position. Just as they say a vector field is V: Q -> TQ by which they mean a vector field is an assignment of a vector at every position.
 
Last edited by a moderator:
Back
Top