History of Keiper-Li coefficients

   complete: 'no'   rigour: assumed-bound-  rigour details: The computed interval was widened by four units in the last place-    (blur_real_interval). That bound was asserted rather than derived, and the justification-    was not recorded.+  rigour details: Rests on mpmath's stieltjes, which evaluates a contour integral+    by numerical quadrature. mpmath's quadrature returns an error *estimate*, and+    its own documentation says the result may still be inaccurate for some functions;+    the working precision it uses is the requested precision plus ten bits and the+    square root of n, which for the largest n here is about forty. The interval was+    then widened by four units in the last place, a bound asserted rather than derived.+    Recomputing at double precision agrees to about 104 digits, so the hundred written+    are very likely right. Display properties:   number-header: $\lambda_n$