How one can prove that the Moore-Penrose pseudoinverse can be calculated by using the limit concept as $$H^+=lim_{\delta \to 0^+} (\delta I+H^*H)^{-1}H^*$$, where $H^*$ denotes the Hermitian matrix.

I know this limit exists even if $A^*A$ is not invertible. Moreover, since $(H^*HH^*+\delta H^*)=H^*(HH^*+\delta I)=(H^*H+\delta I) H^*$, and

$(H^*HH^*+\delta H^*)$ and $H^*(HH^*+\delta I)H^*$ have inverses when $\delta > 0$

hence $$H^+=lim_{\delta \to 0^+} (\delta I+H^*H)^{-1}H^*$$ $$=lim_{\delta \to 0^+} H^*(\delta I+HH^*)^{-1}.$$

But, I couldn't go further to prove the limit definition is also Moore-Penrose pseudoinverse.i.e. $lim_{\delta \to 0^+} (\delta I+H^*H)^{-1}H^*=H^+=(H^*H)^{-1}H^*.$

I was wondering if anyone could help me?


You show that the limit gives $H^+$ by showing it satisfies the properties that define $H^+$; e.g. here's a list on Wikipedia.

The specific formula $H^+ = (H^* H)^{-1} H^*$ only applies in special cases.


Your Answer

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

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