Partial resultOpenContext 5.12
An improved computer-assisted bound at 0.4996 for all $n$, and a sharp 0.5 value for $n\leq 5$ (part II)
by Rafael Wagner · accepted 19 Nov 2026 · partially AI-assisted
Main result
After a couple of prompts, Claude Fable 5.2 found another advance:
Theorem ( bound): Let be an arbitrary -qubit state, and denote the convex polytope of -qubit stabilizer states. For every integer , it holds that
where is the dimension of the underlying Hilbert space and . Furthermore, for the conjecture is shown to hold, i.e., .
The results have used the strategy present in the notes by Y. Wang. "Partial Results on the Hilbert–Schmidt Inradius of the Multiqubit Stabilizer Polytope", in collaboration with ChatGPT 6 Astra.
How this contribution is organised. This is the second of two parts. The first part proves two results on the spectrum of the operators in the -polytope, which are recalled below without proof. Here they are combined with the Liu–Liu inequality of Ref. [3] to bound the purity of these operators, which proves the theorem above. The numbering of the sections continues that of the first part.
What "[computed]" means. Two steps of the argument are computer-assisted and are marked [computed] below: the case , and the constant for all . In these steps the proof is carried out analytically until the claim is reduced to an explicit inequality between elementary functions of two real variables, and this last inequality is then checked by evaluating the functions numerically on a computer (floating-point arithmetic on a fine grid). This is strong numerical evidence rather than a fully rigorous proof, which would require error-controlled (interval) arithmetic. Everything that is not marked [computed] is proved analytically, in particular the conjecture for and the weaker constant for all . All the inequalities used in the two parts were also tested numerically on elements of for . The argument is AI-generated and has not been peer reviewed.
Preliminary mathematical ideas
Recall that a crucial ingredient in this discussion is the notion of the -polytope from Ref. [4]. For more on -polytopes we refer to Refs. [8, 9, 10, 11, 4]. It can be defined as follows:
Above the relevant operators are always defined relative to -qubit Hilbert spaces of dimension . Let , with eigenvalues , and . Throughout this document we define
With these definitions, it is simple to show that the following equations all hold:
Finally, we recall a Lemma that can be found in Ref. [3].
Lemma (Liu–Liu inequality). For an arbitrary , define , , and . Then,
Note that in various of these advances we have been using that we are interested in asking when
is sufficient as a condition for to be a stabilizer mixture and the conjecture is for all .
Results from the first part
The first part proves the following two statements, which we use without repeating their proofs.
Theorem (spectrum of a witness). Let , with eigenvalues . For all and it holds that and . If or , then has eigenvalues and zeros.
For a witness we write
By the theorem, and , and for we have .
Lemma (the star inequality near ). Let and let have a negative eigenvalue. Then
where .
Purity
We now turn the results of the previous section into bounds on . Let us first recall why this is the quantity we need. For a state and , the Cauchy–Schwarz inequality gives
and the right-hand side is non-negative for every as soon as . Since if and only if for all , we conclude that every upper bound proves the main theorem with . In particular, gives the conjectured constant . On the other hand, for every , because and .
3.1 Three bounds on the third moment
The following lemma collects three elementary upper bounds on the third moment in terms of the extreme eigenvalues and the purity. The third one uses Wilkins' inequality for the skewness of a finite set of real numbers [13].
Lemma (three bounds on the third moment). Let and . Then
If, moreover, has a negative eigenvalue, put
Then and
where
Proof: For the first bound, every eigenvalue of satisfies , and we sum over the eigenvalues. For the other two, let be the eigenvalues of other than and . Since and , we have and , so that and with . The second bound follows from . For the third, Wilkins' inequality states that real numbers with mean satisfy ; with and this is the bound with .
3.2 The conjecture for
Theorem (the conjecture for ). For we have . Consequently, for these every state with belongs to , and the constant cannot be improved [1]. The cases are proved analytically; the case is computer-assisted.
Proof: For we have . Let now , assume that , and suppose that some has purity with . We show that this is impossible. With the Liu–Liu inequality reads
We combine it with the first two bounds on the third moment.
(a) From we get . Hence , that is , and . In particular , so has a negative eigenvalue, and both the lemma of Section 3.1 and the star inequality near apply.
(b) From and we get
The cases and . Since and , the star inequality near requires . For the two brackets are and , so . For they are and , so . Both contradict (a). Hence (from ), and then .
The case . Here , and . By (a), , so the star inequality near (dropping the term ) gives , that is,
In (b) we use , and (by (a), and ), so that . With this gives , that is,
The two bounds on are incompatible for and . Hence .
The case [computed]. For the crude estimates of the previous case are not sufficient, and we use the third bound on the third moment instead of the second. For a fixed pair we have three conditions on . The star inequality near reads , where
Item (a) reads . The Liu–Liu inequality with the bound reads , where
is a convex function of (because is linear in ). Therefore is possible only at pairs with , and
We evaluated the left-hand side on a grid of points of the box , , refined towards . Wherever and , it is at most for (and at most for , in agreement with the analytical proof). Hence no witness in has purity above , and . This step is a floating-point evaluation of an explicit function with a uniform margin; it is not an interval-arithmetic proof.
3.3 The bound for all
For the argument of Section 3.2 stops working: the conditions used there are no longer contradictory, so we cannot conclude that , and we cannot use as an input for the next level. The way out is to carry an upper bound from one level to the next instead of an exact value. The Liu–Liu inequality only needs an upper bound on : if , then, because ,
So from a number we can compute a number , from it a number , and so on. Each step can increase the bound only by an amount of order , and converges, so the bounds stay close to for all . We first make this precise in an elementary form, and then in the sharper form that gives the constant .
Lemma (one level of the recursion, elementary form). Let and let be any number with . Then
and
Proof: Let . By the first bound on the third moment and the theorem on the spectrum, . Inserting this in the Liu–Liu inequality with gives , that is, . Maximising over proves the first claim. For the second, the numerator is decreasing in for and equals at , while the denominator is at least .
Corollary (a bound for all without computer assistance). For every we have . Hence the main theorem holds with by an argument that uses no numerical computation.
Proof: By the cases of the theorem in Section 3.2, which are proved analytically, for . For we apply the lemma repeatedly, starting from :
and .
If one also uses , the same estimate starting from gives and . To reach we replace the elementary step by a sharper one, which uses the star inequality and the Wilkins bound.
Theorem (the bound for all ) [computed]. For every we have . Hence the main theorem holds with .
Proof: Step 1 (conditions that a witness with large purity must satisfy). Let , let be a number with , and suppose that has purity . The Liu–Liu inequality gives . We combine it with the star inequality near and with the first and the third bound on the third moment. This gives three conditions on the triple :
(i) the star inequality near , with as in Section 3.2:
(ii) the Liu–Liu inequality with :
(iii) the Liu–Liu inequality with :
where depends on through .
Moreover, (ii) and force . So has a negative eigenvalue, and by the theorem on the spectrum and .
Step 2 (the function ). For we say that a number is admissible if there is a pair with and such that (i), (ii) and (iii) hold. We define
and if there is no admissible . By Step 1, the purity of every witness in is either at most or admissible, so
Step 3 (how is evaluated). Fix a pair . Conditions (i) and (ii) say that , and is a convex function of (on this box and for one has , so is well defined, and is linear in ). Hence the largest admissible for this pair is if ; it is the unique zero of in if ; and there is no admissible for this pair if and . The number is the maximum of this quantity over the box , . We computed it numerically: a grid of points of the box, followed by six rounds of local refinement around the maximiser. The values in the table below are stable to about under changes of the grid.
Step 4 (the recursion). We put , which is the value of , and
By Step 2, for every . For no purity above is admissible, in agreement with Section 3.2; the first level at which an admissible purity above the input appears is , where . The results are collected in the following table. The bounds on are rounded up, and the resulting constants are rounded down.
| bound on | increase from level | resulting constant | how the bound is obtained | ||
|---|---|---|---|---|---|
| 1 | 2 | 2 | – | 0.5 | direct computation |
| 2 | 4 | 2 | 0 | 0.5 | analytical proof |
| 3 | 8 | 2 | 0 | 0.5 | analytical proof |
| 4 | 16 | 2 | 0 | 0.5 | analytical proof |
| 5 | 32 | 2 | 0 | 0.5 | computed: |
| 6 | 64 | 2.00009 | 0.49997 | computed: | |
| 7 | 128 | 2.00052 | 0.49987 | computed: | |
| 8 | 256 | 2.00091 | 0.49977 | computed: | |
| 9 | 512 | 2.00116 | 0.49971 | computed | |
| 10 | 1024 | 2.00131 | 0.49967 | computed | |
| 11 | 2048 | 2.00139 | 0.49965 | computed | |
| 12 | 4096 | 2.00143 | 0.49964 | computed | |
| 13 to 20 | up to 1048576 | 2.00147 | less than per level | 0.49963 | computed |
| all | – | 2.0015 | less than in total | 0.4996 | tail estimate (Step 5) |
Step 5 (the tail). The computed value at level is . For we use the elementary lemma, which bounds the increase at level by :
Together with the table this proves for every , and .
What the numbers mean. The bound at does not say that a witness with this purity exists. It says that conditions (i), (ii) and (iii) no longer exclude a hypothetical witness in with , and purity . At the next level the Liu–Liu inequality is used with instead of , which leaves a little more room, and so on. The increases soon shrink by a factor of about two per level, and the bounds converge to a number slightly below .
Why the exact value is obtained only up to . Near the star inequality needs , while item (b) needs . These two requirements are incompatible exactly as long as , that is, for . So the induction step closes for and fails, by a very small amount, from on. Closing this gap for all would prove the conjecture.
Permanent link:open-context.jan-bulling.com/contribution/5.12
References
- Zhenhuan Liu, Tobias Haug, Qi Ye, Zi-Wen Liu, and Ingo Roth, Triangle Criterion: a mixed-state magic criterion with applications in distillation and detection, arXiv:2512.16777 (2025)
- Michael Zurel and Jack Davis, Basis-independent stabilizerness and maximally noisy magic states, arXiv:2602.22336 (2026)
- Zhenhuan Liu and Zi-Wen Liu, On the geometry, typicality, and detectability of quantum magic, arXiv:2609.03944 (2026)
- Michael Zurel, Cihan Okay, and Robert Raussendorf, Hidden Variable Model for Universal Quantum Computation with Magic States on Qubits, Phys. Rev. Lett. 125, 260404 (2020)
- Ben W. Reichardt, Quantum universality by state distillation, arXiv:quant-ph/0608085 (2006)
- Hilbert–Schmidt inradius of the multiqubit stabilizer polytope, Quantum Information and Quantum Computation Open Problem Zoo (QIQCOP Zoo), (2026)
- John Watrous, The Theory of Quantum Information, Cambridge University Press (2018)
- Cihan Okay, Michael Zurel, Robert Raussendorf, On the extremal points of the $\Lambda$-polytopes and classical simulation of quantum computation with magic states, Quantum Information and Computation Vol.21 No.13&14, 1533-7146 (2021) doi:10.26421/QIC21.13-14-2
- Michael Zurel, Cihan Okay, Robert Raussendorf, and Arne Heimendahl, Hidden variable model for quantum computation with magic states on qudits of any dimension, Quantum 8, 1323 (2024)
- Michael Zurel, Cihan Okay, and Robert Raussendorf, Simulating Quantum Computation: How Many “Bits” for “It”?, PRX Quantum 5, 030343 (2024)
Discussion
Sign in to join the discussion.
No comments yet. Questions about this result, checks of its steps and pointers are welcome here.