I am trying to understand strong convergence for whole-plane spectral sequences in the paper by J.Boardman: https://www.uio.no/studier/emner/matnat/math/MAT9580/v12/undervisningsmateriale/boardman-conditionally-1999.pdf

I am currently confused about the convergence of the zig-zag double complex $D$ mentioned here as one of the answers: https://math.stackexchange.com/questions/1368777/double-complex-with-exact-rows

Firstly Let $C$ be the total complex of $D$. By Theorem 10.1 in Boardman's paper the spectral sequences $E^{\alpha}_2=H(H(D,\delta),d)$ and $E^{\beta}_2=H(H(D,d),\delta)$ converges conditionally to the cohomology $H(\widehat{C})$ where $\widehat{C}$ is the completion of $C$. Furthermore it is clear that $E^{\alpha}$ collapses at page 1 and $E^{\beta}$ collapses at page 2 - so by the remark after Theorem 7.1 we see that $RE_{\infty}=0$ for both spectral sequences. The fact that both spectral sequences collapse imply that the condition for Theorem 8.1 is also satisfied - which means that both of them have $W=0$. Theorem 8.2 then implies that both spectral sequences converge strongly to $H(\widehat{C})$.

Is this a contradiction? Because $E^{\alpha}$ implies that all the filter quotients of $H^n(\widehat{C})$ is zero for all $n$, whereas $E^{\beta}$ tells that that the filter quotients of $H^n(\widehat{C})$ are non-zero.



1 Answer 1


The spectral sequences $E^\alpha$ and $E^\beta$ arise from two different filtrations of $C$, which have different completions, say $\widehat C^\alpha$ and $\widehat C^\beta$. The first converges strongly to $H^*(\widehat C^\alpha) = 0$, and the second converges strongly to $H^*(\widehat C^\beta) \ne 0$. The apparent contradiction arises because you assume that the two completions are the same.

  • $\begingroup$ Thanks very much for your answer. This clears up alot of confusion for me. $\endgroup$
    – Steve
    Commented Jan 29, 2018 at 23:24
  • $\begingroup$ May I ask if it's possible to 'complete' the double complex $D$ in the first place, say $D_1$, so that the completions of the total complex $\widehat{C_1}^\alpha$ and $\widehat{C_1}^\beta$ are the same? and that $H^*(\widehat{C_1}^\alpha)=H^*(\widehat{C_1}^\beta)=H^*(\widehat{C}^\beta)\not=0$? $\endgroup$
    – Steve
    Commented Feb 5, 2018 at 1:04
  • $\begingroup$ Sorry I meant "...so that the cohomology of the completions of the total complexes $\widehat{C_1}^\alpha$ and $\widehat{C_1}^\beta$ are the same?" $\endgroup$
    – Steve
    Commented Feb 5, 2018 at 1:12
  • $\begingroup$ I don't see how to complete the double complex $D$. You could consider a completion of the total complex $C$, at "both ends", but it would not arise from the filtrations giving rise to the two spectral sequences. $\endgroup$ Commented Feb 5, 2018 at 19:59
  • 1
    $\begingroup$ Elements in $C^0 = \bigoplus_p D^{p,-p}$ are finite sums $\sum_p x_p$ with $x_p \in D^{p,-p}$. In the completions we discussed above you allow formal sums where $p$ is either bounded below, or bounded above (depending on the case). By completing at "both ends" I meant that you would allow formal sums, without bounds, amounting to the product $\prod_p D^{p,-p}$. Similarly for $C^1 = \bigoplus_p D^{p,1-p}$, completing to $\prod_p D^{p,1-p}$. $\endgroup$ Commented Feb 5, 2018 at 23:35

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.