I have made some improvements to the LAPACK binding to the HSEQR function, taking into account the suggestions by Karl and Thomas. &nbsp;The use of templated structs to differentiate between complex and double arguments was very helpful. &nbsp;<div>
<br></div><div>Karl suggested using workspaces to allow the use of BLAS3 calls. &nbsp;HSEQR has the ability to calculate the optimal workspace size when called with specific parameters. &nbsp;For this reason, I decided not to use the workspaces, but just asked HSEQR to tell indicate what the size should be. &nbsp;Similar functionality is in GEEV and I tried to follow the example there. &nbsp;Please correct me if my assumption is incorrect.</div>
<div><br></div><div>Finally, I&#39;m still having some troubles and I was hoping someone could help me understand where the problem is. &nbsp;I am trying to pass a pointer to a complex matrix. &nbsp;I tried using traits::complex_ptr(Z) (where Z is the complex matrix), but that only passes a double* instead of traits::complex_d*. &nbsp;</div>
<div><br></div><div>It&#39;s probably best to show this specifically with my code. &nbsp;I have attached my hseqr.hpp file. &nbsp;The calls to hseqr_backend from the Hseqr&lt;2&gt; struct compile just fine, but the calls from the Hseqr&lt;1&gt; struct don&#39;t compile and give &nbsp;the error messages in the file mk.tmp. &nbsp;Can someone help me understand how to fix this?</div>
<div><br></div><div>Thanks again,</div><div>Jeremy</div><div><br></div>