Home / Math and Physics Files / Math / Curvilinear Systems / Tensor Doc and Support / support notes and rejects
new dAn fiddle
DOCX · 23.3 KB
Open DOCX file
Phil's draft notes, in a folder of support notes and rejects for a curvilinear-coordinates tensor document. They derive dx(n), dAn and dV in terms of coordinate variation products, assign tensor weights (0, -1, -1), and transform the objects to x'-space. They also compute magnitudes using scale factors h'n and the cofactor identity for the metric, ending in a summary table and the volume result dV = sqrt(g') dV'. Contains inline reminders and a rejected idea.
AI-written summary; may contain errors. This description is approximate.
Extracted text (machine-read; may contain errors)
New dAn fiddling
1. Initial assumptions in x-space
I am pretty sure these two expressions for dAn are correct:
dAn = |det(Sab)| en Πi≠ndx'i = g'1/2 en Πi≠ndx'i
dAn = σ (-1)n-1 Πxi≠nei Πi≠ndx'i
= σ (-1)n-1 { εkabc...x (e1)a(e2)b ...... (eN)x } Πi≠ndx'i // n missing
Correspondingly we have
dx(n) = en dx'n
dV = | dAn dx(n)| = |σ (-1)n-1 εkabc...x (e1)a(e2)b ...... (eN)x Πidx'i|
= | σ det(S) Πidx'i | = |σ (g')1/2 Πidx'i| = (g')1/2 Πidx'i
which results are consistent with Appendix B. The 3 quantities listed here have absolutely clean and unambiguous meaning in terms of the differential N-piped in x-space.
It is useful to define the following "coordinate variation products" as follows, for later use:
dL'n ≡ dx'n
dA'n ≡ Πi≠ndx'i
dV' ≡ Πidx'i = dA'n dL'n
and then the above become
dx(n) = en dL'n
dAn = σ (-1)n-1 Πxi≠nei dA'n = g'1/2en dA'n
dV = (g')1/2 dV'
The quantities dL'n, dA'n, dV' may regarded as length, area and volume in the Cartesian-view of x'-space as discussed *****. As noted there, this is just a visualization tool.
2. Weights
The weights of our three assumed tensor objects are
dx(n) weight = 0 vector
dAn weight = -1 vector density
dV weight = -1 scalar density
Throughout, we regard things like dx'n and Πi≠ndx'i and Πidx'i as positive scalar constants, and have been calling them the "variation products" of the curvilinear coordinates.
The weight for dAn is obvious from the second formula above, given that the en have weight 0 and the true ε tensor has weight -1. This weight is NOT obvious from the formula d Πi≠ndx'i, although I have tried hard to extract the weight from this formula. It is an equation involving x-space objects en and dAn but includes the x'-space object g'1/2 and therefore does not fit the covariant mold. It might fit the "cross mold" as follows
An = g'1/2 en
A'n = g1/2 e'n // wrong!
but that is pure conjecture. If these were true, we could apply WCT#2 with w=+1 and en having weight 0, but this would tell us that An had weight W+w = 0+1 = +1, and this is wrong. So forget this idea completely!!
3. The objects in x'-space.
Assuming all three objects are tensors of weights shown, we may conclude that
dx'(n) = R dx(n)
dA'n = J R dAn = g'1/2 R dAn
dV' = J dV = g'1/2 dV
The first object may be written
dx'(n) = R [en dx'n] = [Ren] dL'n = e'n dL'n
so that
[dx'(n)]i = (e'n)i dL'n = δni dL'n
For the third object, using the expression dV = (g')1/2 dV above, one may write
dV' = g'1/2 dV = g' dV'
The remaining second object dA'n object may be calculated as follows:
(dA'n)i = g'1/2 Rik (dAn)k = // g'1/2 = |J| = |J|-(-1), weight = -1
= g'1/2 Rik σ (-1)n-1 { εkabc...x (e1)a(e2)b ...... (eN)x } dA'n // n missing
= g'1/2 Rik σ (-1)n-1 { εkabc...x Sa1 Sb2 ...... SxN } dA'n // n missing
= g'1/2 Ski σ (-1)n-1 { εkabc...x Sa1 Sb2 ...... SxN } dA'n // n missing
= g'1/2 σ { (-1)n-1εkabc...x Sa1 Sb2 .. .. Ski..... SxN } dA'n
= g'1/2 σ {εabc..k..x Sa1 Sb2 .. .. Ski..... SxN } dA'n
If i ≠k, then two S second indices must be the same and result is 0. If i =k, then {} = det(Sab) = σg'1/2. Therefore
(dA'n)i = δni g' dA'n
To summarize, the x'-space objects can be written in this manner
[dx'(n)]i = δni dL'n length
(dA'n)i = δni g' dA'n area
dV' = g' dV' volume
4. Evaluation of x-space length and area magnitude.
Consider from above
dx(n) = en dL'n
The magnitude squared is given by
| dx(n)|2 = dx(n) dx(n) = (dL'n)2 en en = g'nn (dL'n)2 = h'n2(dL'n)2
so that
dx(n) ≡ | dx(n)| = hn' dL'n
Turning now to the area, from above
dAn = g'1/2en dA'n
so the magnitude squared is given by
|dAn|2 = dAn dAn = (g'1/2 dA'n)2 en en = (dA'n)2 g' g'nn = (dA'n)2 g' (1/h'n)2 .
Therefore one way to write the magnitude is this
dAn ≡ |dAn| = (1/h'n) g'1/2 dA'n
Another way to write this same dAn arises from the following fact:
g' g'nn = cof(g'nn)
which we now prove:
(g'up)ab ≡ g'ab (g'dn)ab ≡ g'ab
g'up = (g'dn)-1 = cof(g'dnT)/det(g'dn) = cof(g'dn)/det(g'dn)
=> (g'up)nn = cof[(g'dn)nn]/det(g'dn)
or
g'nn = cof[g'nn] / g' QED
Therefore, one can write
|dAn|2 = (dA'n)2 g' g'nn = (dA'n)2 cof(g'nn)
and then one ends up with two forms for the area magnitude dAn ≡ |dAn| and one form for | dx(n)\
dAn = (1/h'n) g'1/2 dA'n
dAn = dA'n
dx(n) = h'n dL'n
5. Evaluation of x'-space length and area magnitude.
dx'(n) = dx(n) since weight = 0
dA'(n) = g'1/2dA(n) since weight = -1
dV' = g'1/2 dV since weight = -1
Therefore
dx'(n) = hn' dL'n
dA'(n) = g'1/2 dA'n = (1/h'n) g' dA'n
dV' = g' dV'
5. Summary of all results
dL'n ≡ dx'n
dA'n ≡ Πi≠ndx'i
dV' ≡ Πidx'i = dA'n dL'n
x-space x-space mag
length dx(n) = en dL'n dx(n)= hn' dL'n
area dAn = g'1/2en dA'n dAn = dA'n = (1/h'n) g'1/2 dA'n
volume dV = (g')1/2 dV' dV = (g')1/2 dV'
x'-space x'-space mag
length (dx'(n))i = δni dL'n dx'(n) = hn' dL'n
area (dA'n)i = δni g' dA'n dA'(n) = g'1/2 dA'n = (1/h'n) g' dA'n
volume dV' = g' dV' dV' = g' dV'
What is the famous volume result?
dV = (g')1/2 dV'
dV = J dV'
and this agrees with Section 8.