17
$\begingroup$

I am currently using Fulton and Harris for a course on representation theory, and I have noticed that there are a few errors throughout the book. A search on google with the keywords "Errata for Fulton and Harris" doesn't come up with anything.

I believe the book is widely used in many representation theory courses, and it would be helpful if a list of errata could be compiled. Does anyone know of any list of errata for this?

Thanks.

Errors I have found: On page 150 (in the middle) the line "...so the representation $W = \Bbb{C}\cdot x^2 \oplus \Bbb{C} \cdot xy \oplus \Bbb{C} \cdot y^2 = W_{-2} \oplus W_0 \otimes W_2$ is the..." should have a direct sum in place of tensor product in between $W_0$ and $W_2$.

$\endgroup$
2

4 Answers 4

13
$\begingroup$

Being somewhat error-prone myself, I'm well aware of the need to collect errata in some systematic way. In the Internet age this has often been done in ad hoc ways on individual homepages, though some publishers (like AMS) are trying to establish durable book pages at their site with updates and errata posted by the authors from time to time.

All of us who consult the book of Fulton & Harris tend to view some of the passages as written down too informally, but there are also some outright errors. For example, many people seem to have trouble following the proof of Proposition 15.15, while Exercise 15.19 has an obvious error in the special case of Weyl's dimension formula: $a+b+1$ should be $a+b+2$.

I did try (unsuccessfully) at one point to get direct clarification from the authors, but it would be optimal for Springer to coordinate the collection of errata. Printing technology now favors print-on-demand and e-books, but these are cheapest when no changes are made in the original printing plates made from a TeX file. Even though it's easy to correct a TeX file, that by itself doesn't motivate publishers to issue corrected reprints. So they do have a responsibility to provide more help to readers in other ways. (By the way, my copy of Fulton & Harris is a first printing, so I'm unsure what if anything has been changed in later printings.)

$\endgroup$
0
6
$\begingroup$

My answer is similar to that of J.Humphreys in that I think there should be a central place for errata ; but I have a more audacious proposition : use the book's wikipedia page!

That way, we don't need the publisher to do something, and we're able to do something by ourselves (where "we" is the mathematical community).

And it can be done likewise for other reference works.

$\endgroup$
2
  • 6
    $\begingroup$ This is so going to be reverted as unverifiable. $\endgroup$
    – Ryan Reich
    Oct 5, 2012 at 13:39
  • 4
    $\begingroup$ Ryan: it can be placed to Wikibooks that abides by weaker policies. (An even bolder proposal: convince Wikimedia to open Project Errata.) $\endgroup$ Oct 6, 2012 at 1:37
6
$\begingroup$

I think there may be a "serious" mistake in the section of branching rules: equation (25.37) and (25.39) on Page 427 (GTM 129, 1991). The formulas actually only true for the "stable case", which is $\lambda_i = 0$ when $i>\lfloor m/2\rfloor$ in case $(O_{m}\mathbb{C},GL_{m}\mathbb{C})$; $\lambda_i = 0$ when $i>n$ in case $(Sp_{2n}\mathbb{C}, GL_{2n}\mathbb{C})$. One may read "Roger Howe, Eng-Chye Tan, Willenbring, Stable Branching Rules for Classical Symmetric Pairs" for a conceptually simpler proof of this formula. However, a "clean" (or rather "useful") branching formulas for non-stable case are still unknown currently (up to my understanding). I think finding such formula is still an active research area recently, while this mistake may cause confusions to non-expert.

$\endgroup$
2
  • 2
    $\begingroup$ If I have understood the question; the branching rules in the non-stable range for the inclusion of $Sp_{2n}$ in $GL_{2n}$ have been found by Sheila Sundaram (in her thesis, supervised by Richard Stanley). I will leave you to decide if this is "clean" or "useful". $\endgroup$ Oct 5, 2012 at 16:55
  • 1
    $\begingroup$ Thank you Bruce, Sheila Sundaram's thesis definitely extends my understanding on branching rules. $\endgroup$
    – Jia-jun Ma
    Oct 6, 2012 at 15:46
3
$\begingroup$

I may have found another error. In exercise 14.37, they ask to show that the Killing Form on $SO(n,\mathbb{R})$ is positive definite. But the formula for the Killing form on the underlying Lie algebra, which is $so(n,\mathbb{R})$ consisting of skew-symmetric matrices, is $(n-2)tr(XY)$. Firstly, for $n=2$, that isn't even definite. Even when $n>2$, it isn't positive. For any skew-symmetric matrix, its square is negative-semidefinite, which can be verified with any simple example. Thus, $(n-2)tr(X^2)\leq 0$ for skew-symmetric matrices. Am I going crazy?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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