Although I will point out, there is https://en.wikipedia.org/wiki/Structured_program_theorem which is a mathematical statement. Whether to actually follow that result when building software is a matter of engineering taste, but the theorem tells you, at the very least, you can do so.
Even if you try to draw a distinction between computer science and software engineering, most of the people writing those papers were pretty squarely on the computer science side.
But I don't draw that distinction. ALL of them were tenured professors of computer science. ALL of them were published in journals of computer science. There is no valid basis on which you can draw an artificial line between computer science and non-computer science, and put those outside of computer science. Doubly not if your goal is to say that computer science is a subset of mathematics rather than its own thing.
I honestly don't see how that follows. Does the fact that writing LAPACK required software engineering mean that numerical linear algebra is not a part of mathematics?
My basis for grouping them together is because of the subjects under study and research methods used. You can always claim that two different fields are different, but I think making arguments for some relationship is more useful.
AS for your example of LAPACK, "The existence of evening, my dear Boswell, does not mean that day is the same as night." Numerical linear algebra is on a boundary between mathematics and computer science. There is a large boundary between math and computer science. But then again the same can be said of, say, math and physics. But by the time you're worried about how to use caching effectively, and make that work in a distributed computation, you're pretty far on the computer science side.
Also your idea of grouping them because of research methods is problematic at best. For example machine learning research has a highly experimental flavor that does not fit in mathematics at all. Quantum computing contains a lot that is closer to a field of applied physics than of mathematics. And so on.
I guess the difference in that regard would be that (software) engineering doesn't necessarily follow the formal scientific method of formulating a hypothesis and then testing it experimentally. That would, in some sense, make it distinct from science. The same would practically apply to many areas in computer science that are studied experimentally, not just software engineering.
To elaborate on that a little, (software) engineering does build on experience and empiricism, as well as analytical thinking, but in practice it may be more in the form of "lessons learned" than in the form of hypothesis testing.
That doesn't make it any less valuable, or even any less valid as an academic area of research. It just, in some sense, makes it possibly distinct from the sciences.
Of course there are also problems in CS that can actually be studied with the scientific method, but I think amelius might have meant that publications such as Dijkstra's and Knuth's papers on goto would be more in the "lessons learned" category than in the "results from the scientific method" category. They would thus not really make computer science a science even though they're not really in the "CS as a part of math" category either.
My position is that computer science is neither a science nor is it a branch of mathematics.
When you've disposed of those, read the less well-known paper _Programming as Theory Building_ by Peter Naur (see https://pages.cs.wisc.edu/~remzi/Naur.pdf for a link).
Are there computer science papers that could fit within mathematics? Yes. Are there important computer science papers that clearly don't? Also yes.