Home / Math and Physics Files / Math / Curvilinear Systems / Tensor Doc and Support / Files related to May 2015 update
scratch
DOCX · 24.5 KB
Open DOCX file
Informal scratch notes by Phil, apparently written while revising the Appendix E and Section 6-7 notation of his curvilinear coordinates/tensor document. He reviews sections E.1 to E.7 for misuse of the covariant dot product, tests whether the completeness relation for dual bases can be written as a matrix sum, and tries to build a bra-ket analogy with outer and inner products. Several attempts are abandoned or revised.
AI-written summary; may contain errors. This description is approximate.
Extracted text (machine-read; may contain errors)
scratch junk
Maybe better to write this as
Σn( n)a (bn)b = Σn( Bn )a (bn)b = Σn ac(Bn)c (bn)b = ac (Bn)c (bn)b
So then completeness says
ac (Bn)c (bn)b = δa,b
and I avoid having floating around here. But this just does not look nice to me. There is no way to write this in matrix form.
But suppose I apply gda to both sides. I get
gda ac (Bn)c (bn)b = gdaδa,b
or
δdc (Bn)c (bn)b = gdb
(Bn)d (bn)b = gdb
or
(Bn)a (bn)b = gab
Maybe THIS is what I should be calling "completeness" ? The matrix form would be
Σn Bn bnT = gdn
and in Standard notation
Σn bn bnT = gup
Can I reverse the n tilt here? Feeding time. Back. How might I re-analyze the analogy to the bra-ket notation?
I may have made an error again with the notation ion Appendix E . Lets review
E.1 All OK, used covariant notation.
E.2 OK down to (E.2.11), and I retain my red commend about trying to tilt the other way, put that off for the moment.
Orthonormal Basis section: seems OK as is. I think all of E.2 is OK.
E.3 Polyadic notation. Seems all OK, not dot products appear.
E.4 OK, not dot products, keeping up and down indices in right places.
E.5. Perhaps first signs of trouble. First problem is in (E.5.6) where I say
aTb = a b = (a1 a2) = a1b1 + a2b2 = a scalar (if a and b are vectors) . (E.5.5)
This is NOT the covariant dot product!!! Beware!! I can fix this section simply by replacing the above line with this
aTb = (a1 a2) = a1b1 + a2b2 . (E.5.5)
Yes, that rescues the entire Section E.5 so install that change right now.
E.6 Now more big trouble with dot notation! I have highlighted in red chunks that are problematical, but most of it survives for now.
E.7 Here is where I try to get bra-ket going. The big problem is this:
Σi i biT = 1 (7.18.6) → Σi |bi><bi| = 1 = Σi |bi><bi| // completeness
Let's try this rewrite of the above
Σi i biT = 1 (7.18.6) → Σi |i><b| = 1 // completeness
abT = (b1 b2)
The thing on the left in abT is the column vector which is the |a> thing. I don't like the above idea because I would like to see the regular vector |bi> appear.
Question 1: What happens if I transpose both sides of the left matrix equation?
[ Σi i biT]T = 1T = 1
Σi ( i biT)]T == 1
Σi bi ( i)T == 1
This still is a matrix on the left in this sense
abT = (b1 b2) = = a matrix // same as matrix ab = ab (E.5.3)
Now lets try this connection
Σi bi ( i)T == 1 → Σi |bi><i| = 1
The thing on the left in abT is the column vector which is the |a> thing. So at least in the above form I get the normal ket on the left which is |bi>.
Question 2: Can I move the bar to the other term? Go way back to this
Σn( n)a (bn)b = δa,b // completeness relation for the dual set {bn, Bn } (6.2.15)
Go back to notes above:
I am wondering now about my completeness thing
Σn( n)a (bn)b = δa,b Σn n bnT = 1
Maybe better to write this as
Σn( n)a (bn)b = Σn( Bn )a (bn)b = Σn ac(Bn)c (bn)b = ac (Bn)c (bn)b
So then completeness says
ac (Bn)c (bn)b = δa,b
Now play with the last equation
ac (Bn)c (bn)b = δa,b
(Bn)c ac (bn)b = δa,b
But this does NOT produce (Bn)c n = δa,b. So the answer is: you cannot just move the bar to the other term!
Plan A. Let's try to construct the whole analogy from scratch assuming this connection from above,
Σi bi ( i)T == 1 → Σi |bi><i| = 1
Here we go:
bi → |bi> // vector
biT → <bi| // transpose vector
abT (E.5.3) → |a><b| // a matrix (outer product)
aTb (E.5.5) → <a | b> // a number (inner product)
a b = aibi →
V = Σi [V(b)]i bi (7.18.7) → |V> = Σi [V(b)]i |bi> // vector expansion ...
The next item is suspicious, since I have a covariant dot on the left side:
[V(b)]i = bi V (7.18.7) → [V(b)]i = <bi|V> // and coefficients
How can I fix this up?
bi V = (i)a Va
STOP and go back to dev notation
Σn( n)a (bn)b = δa,b // completeness relation for the dual set {bn, Bn } (6.2.15)
How do you translate this equation to standard notation??
Rule 1: n → bn
Rule 2: contravariant indices go up
We then get
Σn( bn)a (bn)b = δab = δa,b = gab
This looks good! Then in matrix notation this becomes
Σn( bn)(bn)T = 1
and this is MUCH better. So what about (7.18.7) now?
(7.13.1) Σn(n)i(en)j = δi,j Σn (bn)i(bn)j = δi,j // completeness (7.18.5)
(7.13.1) Σn n enT = 1 Σn bn bnT = 1 // completeness (matrix form) (7.18.6)
Remember from (7.13.1) that (n)i →(en)i , so the label goes up, the index stays down!!!