Let $S$ and $T$ be Dirac-type operators on vector bundles $E\rightarrow M$ and $F\rightarrow N$ respectively, where $M$ and $N$ are manifolds. Suppose $Q_S$ and $Q_T$ are parametrices for $S$ and $T$. Form the operator $S\#T$ on the external product bundle $E\boxtimes F\rightarrow M\times N$, defined as $$S\#T=S\otimes 1+1\otimes T.$$

Then $S\#T$ is an elliptic operator and has a parametrix $Q_{S\#T}$ so that

$$1-Q_S S,\qquad 1-Q_T T,\qquad 1-Q_{S\#T}(S\#T),$$ $$1-S Q_S,\qquad 1-T Q_T,\qquad 1-(S\#T)Q_{S\#T}$$ have smooth Schwartz kernels.

Question: How can one construct $Q_{S\#T}$ using $Q_S$ and $Q_T$ so that one can relate the Schwartz kernel of, say $1-Q_{S\#T}(S\#T)$, to those of $1-Q_S S$ and $1-Q_T T$?


  • 1
    $\begingroup$ Why is $S\#T$ elliptic? $\endgroup$ – Deane Yang Jun 19 '18 at 19:16
  • $\begingroup$ As a post with similar terminologies please see mathoverflow.net/questions/266734/… $\endgroup$ – Ali Taghavi Jun 19 '18 at 21:14
  • $\begingroup$ @DeaneYang one can ask the same question in the context of fredholm operators on Hilber spaces is $S\#T$ a fredholm operator if both $S$ and $T$ are Fredholm operators on hilbert spaces $H_1,H_2$, respectively? $\endgroup$ – Ali Taghavi Jun 19 '18 at 21:25
  • $\begingroup$ @DeaneYang You're right, it doesn't in general need to be elliptic. Let's specialise to Dirac-type operators. $\endgroup$ – ougoah Jun 20 '18 at 0:42
  • $\begingroup$ I haven't tried to do it myself, but it seems like you could do this by the usual approach using the symbol calculus for pseudodifferential operators. $\endgroup$ – Deane Yang Jun 20 '18 at 2:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.