betaFeedback

WorksMathematics

Sum of Powers with Matrix Coefficients

Abstract

Let FF be an algebraically closed field of characteristic zero. We consider polynomial maps with constants ω(x,y)=A1xk+A2ykω(x,y)=A_1x^k+A_2y^k on Mn(F)M_n(F), where A1A_1 is invertible and k≥2k\geq2. We show that the necessary condition for surjectivity obtained by Saini and Singh (Polynomial maps with constants on matrix algebra, Linear Algebra Appl. 752 (2027), 256-277) is not sufficient in any dimension n≥6n\geq6. We also completely characterize surjectivity for n=5n=5 and 66.

Introduction

Let FF be a field and let AA be an associative FF-algebra. A polynomial f∈F⟨x,y,…,xm⟩f\in F\langle x,y,\ldots,x_m\rangle in non-commuting variables defines a map on AA by evaluation. If ff is a polynomial identity for AA, then its image is {0}\{0\}. Otherwise, one may ask which elements of AA belong to its image. If the map is not surjective, can every element of AA be expressed as a sum of a prescribed number of values of ff?

A central direction concerns Waring-type problems: whether every element of an algebra can be expressed as a sum of a prescribed number of polynomial values. Representations of matrices as sums of powers form an important class of such problems; see [2, 3, 4, 9]. Related questions concern the structure of polynomial images themselves, including the L’vov–Kaplansky conjecture on images of multilinear polynomials. For an account of polynomial images and Waring-type problems, we refer to [1].

A further variation is obtained by allowing the coefficients of a polynomial to belong to the algebra itself. These polynomials are called generalized polynomials, or polynomials with constants. In this setting, the image depends on the chosen coefficients as well as on the polynomial expression; see [8, 10, 11, 12]. Moreover, the image need not be invariant under conjugation, since conjugating a value also conjugates the coefficients. The study of such maps arises naturally as analogues of word maps with constants in group theory; see [5, 6, 7].

2020 Mathematics Subject Classification. Primary: 16S50,11P05, Secondary: 16K20.

Key words and phrases. Polynomial maps, Polynomial maps with constant, Matrix algebras.

The author acknowledges financial support from the Indian Institute of Science Education and Research (IISER) Mohali through a Postdoctoral Fellowship.

Panja, Saini and Singh [10] studied polynomial maps with constants on M2(F)M_2(F) over an algebraically closed field. In particular, they characterized the surjectivity of maps of the form A1xk1+A2yk2A_1x^{k_1}+A_2y^{k_2} and proved that the image of A1xy−A2yxA_1xy-A_2yx is a vector space. Subsequently, Saini and Singh [12] considered the maps

ω:Mn(F)×Mn(F)⟶Mn(F),ω(x,y)=A1xk+A2yk.\omega:M_n(F)\times M_n(F)\longrightarrow M_n(F),\qquad\omega(x,y)=A_1x^k+A_2y^k.

where FF is algebraically closed of characteristic zero, k≥2k\ge2, and A1A_1 is invertible. Writing r0=nullity⁡(A2)r_0=\operatorname{nullity}(A_2), they proved that n>k(r0−1)n>k(r_0-1) is necessary for surjectivity and that this condition is also sufficient for n=3,4n=3,4. This raises the question of whether the same inequality characterizes surjectivity in higher orders.

We prove that the necessary inequality is sufficient for n=5n=5, but not for n=6n=6. More generally, for every n≥6n\ge6 and k≥n−1k\ge n-1, we characterize surjectivity in terms of the nullity of the second coefficient. Our main results are the following. Throughout, FF is algebraically closed of characteristic zero.

Theorem 1.1. Let FF be an algebraically closed field of characteristic zero. Let n≥6n\ge6 and k≥n−1k\ge n-1, and A1,A2∈Mn(F)A_1,A_2\in M_n(F) with A1A_1 invertible. Then the map

ω:Mn(F)×Mn(F)⟶Mn(F),ω(x,y)=A1xk+A2yk,\omega:M_n(F)\times M_n(F)\longrightarrow M_n(F),\qquad\omega(x,y)=A_1x^k+A_2y^k,

is surjective if and only if nullity⁡(A2)≤1\operatorname{nullity}(A_2)\le1.

Theorem 1.2. Let FF be an algebraically closed field of characteristic zero. Let n∈{5,6}n\in\{5,6\}, and A1,A2∈Mn(F)A_1,A_2\in M_n(F) with A1A_1 invertible. For an integer k≥2k\ge2, consider the map

ω:Mn(F)×Mn(F)⟶Mn(F),ω(x,y)=A1xk+A2yk.\omega:M_n(F)\times M_n(F)\longrightarrow M_n(F),\qquad\omega(x,y)=A_1x^k+A_2y^k.

Set r0=nullity⁡(A2)r_0=\operatorname{nullity}(A_2). Then:

  1. For n=5n=5, the map ω\omega is surjective if and only if 5>k(r0−1)5>k(r_0-1).

  2. For n=6n=6, the map ω\omega is surjective if and only if 6>k(r0−1)6>k(r_0-1) and (r0,k)≠(2,5)(r_0,k)\ne(2,5).

The paper is organized as follows. § 2 contains the reductions and auxiliary lemmas used throughout. In § 3, we establish the general obstruction and deduce the classification for large exponents. § 4 and § 5 give the classifications in orders five and six. In § 6, we conclude with a question concerning higher orders.

Preliminaries

Throughout, we assume FF is an algebraically closed field of characteristic 0 and k≥2k\ge2. We recall the following reductions [12]. Let A1,A2∈Mn(F)A_1,A_2\in M_n(F) be non-zero elements such that one of the coefficients is invertible. Consider the map

ω:Mn(F)2⟶Mn(F),ω(x,y)=A1xk+A2yk.\omega:M_n(F)^2\longrightarrow M_n(F),\qquad\omega(x,y)=A_1x^k+A_2y^k.

Without loss of generality, we assume A1A_1 is invertible. Define B=A1−1A2∈Mn(F)B=A_1^{-1}A_2\in M_n(F) and r0=nullity⁡(B)=nullity⁡(A2)r_0=\operatorname{nullity}(B)=\operatorname{nullity}(A_2).

Moreover, the map ω\omega is surjective if and only if the map ωB(x,y)=xk+Byk\omega_B(x,y)=x^k+By^k is surjective. Indeed, ω(x,y)=C\omega(x,y)=C if and only if ωB(x,y)=A1−1C\omega_B(x,y)=A_1^{-1}C.

Reduction to Jordan form. Surjectivity of the map ω\omega remains invariant under the action of the automorphism group of PGL⁡n(F)\operatorname{PGL}_n(F). Thus, we may replace BB by its Jordan canonical form

J=(⨁i=1r′Jλi,ni)⊕(⨁s=1r0J0,ms),J=\left(\bigoplus_{i=1}^{r'}J_{\lambda_i,n_i}\right)\oplus\left(\bigoplus_{s=1}^{r_0}J_{0,m_s}\right),

where λi≠0\lambda_i\ne0 and ∑s=1r0ms=n0\sum_{s=1}^{r_0}m_s=n_0. For 1≤s≤r01\le s\le r_0, set ℓs:=n−n0+m1+⋯+ms\ell_s:=n-n_0+m_1+\cdots+m_s, and define σ:{1,…,n}→{1,…,n}\sigma:\{1,\ldots,n\}\to\{1,\ldots,n\} by

σ(i)={i,1≤i≤n−n0,i−#{s:ℓs<i},n−n0<i≤n, i∉{ℓ1,…,ℓr0},n−r0+s,i=ℓs.\sigma(i)= \begin{cases} i, & 1\le i\le n-n_0,\\ i-\#\{s:\ell_s<i\}, & n-n_0<i\le n,\ i\notin\{\ell_1,\ldots,\ell_{r_0}\},\\ n-r_0+s, & i=\ell_s. \end{cases}

associated to JJ and PP be the permutation matrix corresponding to σ\sigma. Consider the partition of CC and JJ obtained under the conjugation of PP:

P−1CP=(C11PC12PC21PC22P)andP−1JP=(J^0),P^{-1}CP= \begin{pmatrix} C_{11}^{P} & C_{12}^{P}\\ C_{21}^{P} & C_{22}^{P} \end{pmatrix} \qquad\text{and}\qquad P^{-1}JP= \begin{pmatrix} \widehat{J}\\ 0 \end{pmatrix},

where C11P∈Mn−r0(F)C_{11}^{P}\in M_{n-r_0}(F), C22P∈Mr0(F)C_{22}^{P}\in M_{r_0}(F) and J^∈M(n−r0)×n(F)\widehat{J}\in M_{(n-r_0)\times n}(F). Note that J^\widehat{J} is a full row rank. We set L1=(C11PC12P)∈Mn−r0×n(F)L_1=\begin{pmatrix}C_{11}^{P}&C_{12}^{P}\end{pmatrix}\in M_{n-r_0\times n}(F) and L2=(C21PC22P)∈Mr0×n(F)L_2=\begin{pmatrix}C_{21}^{P}&C_{22}^{P}\end{pmatrix}\in M_{r_0\times n}(F).

Denote W1⊆FnW_1\subseteq F^n as the subspace spanned by the first n−r0n-r_0 rows of CPC^P and W2⊆FnW_2\subseteq F^n as the subspace spanned by the last r0r_0 rows of CPC^P. In the following sections, we consider CC and BB up to conjugation by PP.

Notation

Throughout, FF denotes an algebraically closed field of characteristic 0, and the split central simple algebra over FF is Mn(F)M_n(F), the algebra of all n×nn\times n matrices. We denote the automorphism group of Mn(F)M_n(F) by PGL⁡n(F)\operatorname{PGL}_n(F), the projective general linear group, which acts on Mn(F)M_n(F) by conjugation.

For a matrix C∈Mn(F)C\in M_n(F), we denote its rows by c1,…,cn∈Fnc_1,\ldots,c_n\in F^n and C=(c1,…,cn)⊤C=(c_1,\ldots,c_n)^\top. For 1≤i≤n1\le i\le n, let ei⊤e_i^\top denote the ii-th standard basis vector of FnF^n. For C1∈Mn1(F)C_1\in M_{n_1}(F) and C2∈Mn2(F)C_2\in M_{n_2}(F), we define their direct sum by

C1⊕C2=(C100C2)∈Mn1+n2(F).C_1\oplus C_2= \begin{pmatrix} C_1&0\\ 0&C_2 \end{pmatrix} \in M_{n_1+n_2}(F).

For B∈Mn(F)B\in M_n(F), we write B∼⨁i=1rJλi,niB\sim\bigoplus_{i=1}^{r}J_{\lambda_i,n_i}, ∑i=1rni=n\sum_{i=1}^{r}n_i=n, to denote a Jordan canonical form of BB, where Jλi,niJ_{\lambda_i,n_i} is the Jordan block of size nin_i corresponding to the eigenvalue λi\lambda_i. We denote by r0(B)r_0(B) the nullity of BB, which turns out to be the number of Jordan blocks corresponding to the eigenvalue 00, and by r′(B)r'(B) the number of Jordan blocks corresponding to nonzero eigenvalues. Thus, r(B)=r0(B)+r′(B)r(B)=r_0(B)+r'(B).

For S∈PGL⁡n(F)S \in\operatorname{PGL}_n(F) and C∈Mn(F)C \in M_n(F), we denote SCS−1SCS^{-1} by CSC^S. For Z∈Ms×t(F)Z \in M_{s\times t}(F), span⁡(Z)\operatorname{span}(Z) is the span of rows of ZZ. We retain the above notation throughout the paper.

Auxiliary Lemmas

We first recall the Jordan form of the power of a nilpotent Jordan block.

Lemma 2.1 (Miller). Let n>k≥2n>k\ge2. Then the kk-th power of the nilpotent Jordan block J0,nJ_{0,n} is conjugate to

(⨁k−mJ0,⌊n/k⌋)⊕(⨁mJ0,⌈n/k⌉),\left(\bigoplus_{k-m} J_{0,\lfloor n/k\rfloor}\right)\oplus\left(\bigoplus_m J_{0,\lceil n/k\rceil}\right),

where m≡n(modk)m\equiv n\pmod{k} and 0≤m≤k−10\le m\le k-1.

Lemma 2.2. For n≥3n\ge3, every rank-one matrix in Mn(F)M_n(F) is a kk-th power for 2≤k≤n−12\le k\le n-1.

Proof. Let ZZ have rank one. If tr⁡(Z)≠0\operatorname{tr}(Z)\ne0, then ZZ is similar to diag⁡(tr⁡(Z),0n−1)\operatorname{diag}(\operatorname{tr}(Z),0_{n-1}) and hence has a kk-th root. Otherwise, ZZ is similar to J0,2⊕0n−2J_{0,2}\oplus0_{n-2}. By Miller’s Theorem 2.1,

J0,2⊕0n−2∼(J0,k+1⊕0n−(k+1))k.J_{0,2}\oplus0_{n-2}\sim\left(J_{0,k+1}\oplus0_{n-(k+1)}\right)^k.

The zero summand is omitted when k=n−1k=n-1. Therefore, Z=XkZ=X^k for some X∈Mn(F)X\in M_n(F). □

Lemma 2.3. For 1≤s≤n1\le s\le n, let R∈Ms×n(F)R\in M_{s\times n}(F). If rank⁡(R)≤n−s\operatorname{rank}(R)\le n-s or rank⁡(R)=s\operatorname{rank}(R)=s, then there exist H∈Mn−s×n(F)H\in M_{n-s\times n}(F) such that the matrix

(HR)=X2for some X∈Mn(F).\begin{pmatrix} H \\ R \end{pmatrix}=X^2\quad\text{for some }X\in M_n(F).

Proof. If rank⁡(R)=s\operatorname{rank}(R)=s, extend its rows to an invertible matrix of order nn and take a square root.

Otherwise, set t=rank⁡R≤n−st=\operatorname{rank}R\le n-s. For t=0t=0, take X=0X=0. Therefore, we take t≠0t\ne0. Choose Q∈GL⁡s(F)Q\in\operatorname{GL}_s(F) such that the last s−ts-t rows of Q−1RQ^{-1}R are zero because rank⁡(R)=t<s\operatorname{rank}(R)=t<s. Let S=In−s⊕QS=I_{n-s}\oplus Q and write

Q−1RS={u1,…,ut,0s−t}T.Q^{-1}RS=\{u_1,\ldots,u_t,0_{s-t}\}^{\mathsf T}.

Since t≤n−st\le n-s, define

Z={u1,…,ut,0(n−s−t)×n,e1T,…,etT,0(s−t)×n}T.Z=\{u_1,\ldots,u_t,0_{(n-s-t)\times n},e_1^{\mathsf T},\ldots,e_t^{\mathsf T},0_{(s-t)\times n}\}^{\mathsf T}.

The last ss rows of Z2Z^2 are Q−1RSQ^{-1}RS. Let X=SZS−1X=SZS^{-1}. Since S=In−s⊕QS=I_{n-s}\oplus Q, the last ss rows of X2=SZ2S−1X^2=SZ^2S^{-1} are Q(Q−1RS)S−1=RQ(Q^{-1}RS)S^{-1}=R. Therefore, X2X^2 has the required form. □

As a corollary, we obtain:

Corollary 2.4. For s≥1s \ge1, let R∈Ms×n(F)R \in M_{s \times n}(F). If rank⁡(R)≤n−s\operatorname{rank}(R) \le n-s or rank⁡(R)=s\operatorname{rank}(R)=s, then there exist H∈Mn−s×n(F)H \in M_{n-s \times n}(F) such that the matrix (RH)=X2\begin{pmatrix} R \\ H \end{pmatrix} = X^2 for some X∈Mn(F)X \in M_n(F).

Proof. Let P=(0s×n−sIsIn−s0n−s×s)P=\begin{pmatrix} 0_{s \times n-s} & I_s \\ I_{n-s} & 0_{n-s \times s} \end{pmatrix}. Since rank⁡(RP)=rank⁡R\operatorname{rank}(RP)=\operatorname{rank}R, by Lemma 2.3 there exists Z∈Mn(F)Z \in M_n(F) such that the last ss rows of Z2Z^2 are RPRP. Then (PZP−1)2=PZ2P−1(PZP^{-1})^2=PZ^2P^{-1} and hence the first ss rows of (PZP−1)2(PZP^{-1})^2 are therefore RPP−1=RRPP^{-1}=R. □

Lemma 2.5. For 1≤s≤n1 \le s \le n and 2≤k≤n−12 \le k \le n-1, let D∈Ms×n(F)D \in M_{s \times n}(F) such that rank⁡(D)\operatorname{rank}(D) is ss. If H∈Ms×n(F)H \in M_{s \times n}(F) satisfying rank⁡(H)∈{0,1,s}\operatorname{rank}(H) \in\{0,1,s\}, then there exists Y∈Mn(F)Y \in M_n(F) such that

DYk=H.DY^k=H.

Proof. If H=0H=0, take Y=0Y=0. Suppose that rank⁡(H)=1\operatorname{rank}(H)=1. Write H=bwH=bw, where b∈Fsb \in F^s is a nonzero column vector and w∈M1×n(F)w \in M_{1 \times n}(F) is nonzero. Since DD has full row rank, there exists q∈Fnq \in F^n such that Dq=bDq=b. In particular, q≠0q \ne0, so qwqw has rank one. By Lemma 2.2, there exists Y∈Mn(F)Y \in M_n(F) such that Yk=qwY^k=qw. Consequently,

DYk=D(qw)=(Dq)w=bw=H.DY^k=D(qw)=(Dq)w=bw=H.

Finally, suppose that rank⁡(H)=s\operatorname{rank}(H)=s. Extend the rows of DD and HH to bases of FnF^n, obtaining invertible matrices

S=(DD0),T=(HH0).S=\begin{pmatrix} D \\ D_0 \end{pmatrix}, \qquad T=\begin{pmatrix} H \\ H_0 \end{pmatrix}.

With E=(Is 0)E=(I_s\ 0), we have D=ESD=ES and H=ETH=ET. The invertible matrix S−1TS^{-1}T has a kk-th root Y∈Mn(F)Y \in M_n(F). Hence

DYk=ES(S−1T)=ET=H.DY^k=ES(S^{-1}T)=ET=H.

When s=ns=n, the additional rows are omitted. □

Lemma 2.6. Let 1≤r<n1 \le r<n, and set s=n−rs=n-r. Let R∈Mr×n(F)R \in M_{r \times n}(F) have rank rr. For every L∈Ms×n(F)L \in M_{s \times n}(F), there exists L′∈Ms×n(F)L' \in M_{s \times n}(F) such that N=(L′R)N=\begin{pmatrix} L' \\ R \end{pmatrix} is invertible and

rank⁡(L−L′)=s.\operatorname{rank}(L-L')=s.

Proof. Extend the rows of RR to a basis of FnF^n, obtaining an invertible matrix

N0=(R0R).N_0=\begin{pmatrix} R_0 \\ R \end{pmatrix}.

Set E=(Is  0)E=(I_s\ \ 0) and, for t∈Ft\in F, define

N(t)=(R0+tER).N(t)=\begin{pmatrix} R_0+tE \\ R \end{pmatrix}.

Let DD be the matrix formed by the first ss columns of L−R0L-R_0. Consider

f(t)=det⁡N(t),g(t)=det⁡(D−tIs).f(t)=\det N(t),\qquad g(t)=\det(D-tI_s).

The polynomial ff is nonzero because f(0)=det⁡N0≠0f(0)=\det N_0\ne0, while gg has degree ss and leading coefficient (−1)s(-1)^s. Since FF is infinite, there exists t0∈Ft_0\in F such that

f(t0)g(t0)≠0.f(t_0)g(t_0)\ne0.

Taking L′=R0+t0EL'=R_0+t_0E, we obtain an invertible matrix N=N(t0)N=N(t_0). Moreover, the first ss columns of L−L′L-L' form the invertible matrix D−t0IsD-t_0I_s. Thus rank⁡(L−L′)=s\operatorname{rank}(L-L')=s.

Non-Surjectivity for large exponents

In [12], the authors proved that, when A1A_1 is invertible, the inequality n>k(r0−1)n>k(r_0-1), where r0=nullity⁡(A2)r_0=\operatorname{nullity}(A_2), is necessary for the surjectivity of ω(x,y)=A1xk+A2yk\omega(x,y)=A_1x^k+A_2y^k and is also sufficient for n=3,4n=3,4. We show below that this condition is not sufficient in general by constructing counterexamples in every order n≥6n\ge6.

Proposition 3.1. Let FF be an algebraically closed field of characteristic zero, let n≥6n\ge6, and let A∈M(n−2)×n(F)A\in M_{(n-2)\times n}(F) have rank n−2n-2. Set

B=(A02×n).B=\begin{pmatrix} A \\ 0_{2\times n} \end{pmatrix}.

Then the map Φ(X,Y)=Xn−1+BYn−1\Phi(X,Y)=X^{n-1}+BY^{n-1} is not surjective.

Proof. Set p=n−2p=n-2.

We first observe that if Zn−1Z^{n-1} has a nontrivial Jordan block at zero, then ZZ is similar to J0,nJ_{0,n}, see Theorem 2.1. Indeed, every zero Jordan block of size at most n−1n-1 vanishes upon taking its (n−1)(n-1)-st power. Consequently,

rank⁡(Zn−1)=1.\operatorname{rank}(Z^{n-1})=1.

Since rank⁡A=p≥4\operatorname{rank} A = p \ge4, choose G∈GL⁡p(F)G \in\operatorname{GL}_p(F) such that the first two rows v1,v2v_1,v_2 of GAGA are linearly independent modulo FenTF e_n^{\mathsf{T}}. Denote the remaining rows by v3,…,vpv_3,\ldots,v_p, and define

R=G−1(00v1v20(p−4)×n),C=(RenT0).R = G^{-1} \begin{pmatrix} 0\\ 0\\ v_1\\ v_2\\ 0_{(p-4)\times n} \end{pmatrix}, \qquad C = \begin{pmatrix} R\\ e_n^{\mathsf{T}}\\ 0 \end{pmatrix}.

The block 0(p−4)×n0_{(p-4)\times n} in the definition of RR is omitted when p=4p=4. We show that CC is not in the image of Φ\Phi.

Suppose that

C=M+BN,M=Xn−1,N=Yn−1.C=M+BN,\qquad M=X^{n-1},\qquad N=Y^{n-1}.

The last two rows give

en−1TM=enT,enTM=0.e_{n-1}^{\mathsf{T}}M=e_n^{\mathsf{T}},\qquad e_n^{\mathsf{T}}M=0.

Thus MM has a nontrivial Jordan block at zero, and hence rank⁡M=1\operatorname{rank} M=1. Since its (n−1)(n-1)-st row is enTe_n^{\mathsf{T}}, we may write

M=(aenTenT0)M= \begin{pmatrix} a e_n^{\mathsf{T}}\\ e_n^{\mathsf{T}}\\ 0 \end{pmatrix}

for some column vector a∈Fpa\in F^p.

Set α=Ga\alpha=Ga. Multiplying the top-row equation R=aenT+ANR=a e_n^{\mathsf{T}}+AN on the left by GG gives

v1N=−α1enT,v2N=−α2enT,v_1N=-\alpha_1e_n^{\mathsf{T}},\qquad v_2N=-\alpha_2e_n^{\mathsf{T}},
v3N=v1−α3enT,v4N=v2−α4enT.v_3N=v_1-\alpha_3e_n^{\mathsf{T}},\qquad v_4N=v_2-\alpha_4e_n^{\mathsf{T}}.

The vectors {v3N,v4N}\{v_3N,v_4N\} are linearly independent because v1,v2v_1,v_2 are independent modulo FenTF e_n^{\mathsf{T}}. Hence rank⁡N≥2\operatorname{rank}N\ge2.

We next find a nonzero vector z∈span⁡(N)z\in\operatorname{span}(N) satisfying zN=0zN=0. If (α1,α2)≠(0,0)(\alpha_1,\alpha_2)\ne(0,0), the displayed identities imply that enT,v1,v2e_n^{\mathsf{T}},v_1,v_2 belong to span⁡(N)\operatorname{span}(N). Therefore

z=α2v1−α1v2z=\alpha_2v_1-\alpha_1v_2

is nonzero, belongs to span⁡(N)\operatorname{span}(N), and satisfies zN=0zN=0.

If α1=α2=0\alpha_1=\alpha_2=0, choose (b1,b2)≠(0,0)(b_1,b_2)\ne(0,0) such that b1α3+b2α4=0b_1\alpha_3+b_2\alpha_4=0. Then

z=b1(v1−α3enT)+b2(v2−α4enT)=b1v1+b2v2z=b_1(v_1-\alpha_3e_n^{\mathsf{T}})+b_2(v_2-\alpha_4e_n^{\mathsf{T}})=b_1v_1+b_2v_2

is a nonzero element of span⁡(N)\operatorname{span}(N), and zN=0zN=0.

In either case, there exists a row vector ww such that wN=z≠0wN=z\ne0 and wN2=0wN^2=0. Thus NN has a nontrivial Jordan block at zero. Since N=Yn−1N=Y^{n-1}, the initial observation gives rank N=1N=1, a contradiction. □

As a corollary, we obtain the following:

Corollary 3.2. Let n≥6n\ge6, and let B∈Mn(F)B\in M_n(F) have nullity two. Then the map

(X,Y)⟼Xn−1+BYn−1(X,Y)\longmapsto X^{n-1}+BY^{n-1}

is not surjective.

Proof. By similarity, BB may be replaced by

(A02×n)\begin{pmatrix} A \\ 0_{2\times n} \end{pmatrix}

where AA has rank n−2n-2. The result follows from Proposition 3.1. □

Proof of Theorem 1.1 Let r0=nullity⁡(A2)r_0=\operatorname{nullity}(A_2). If r0≤1r_0\le1, surjectivity follows from [12 Theorem 1.2(1)]. Suppose that r0≥2r_0\ge2. If k≥nk\ge n, then k(r0−1)≥nk(r_0-1)\ge n, so ω\omega is not surjective by [12 Theorem 1.2(2)].

It remains to consider k=n−1k=n-1. If r0≥3r_0\ge3, then k(r0−1)≥2(n−1)≥nk(r_0-1)\ge2(n-1)\ge n, and the same necessary condition gives non-surjectivity. If r0=2r_0=2, set B=A1−1A2B=A_1^{-1}A_2. Since nullity⁡(B)=2\operatorname{nullity}(B)=2, Corollary 3.2 shows that (X,Y)↦Xn−1+BYn−1(X,Y)\mapsto X^{n-1}+BY^{n-1} is not surjective. Multiplication by A1A_1 gives the required conclusion.

The case n=5n=5

In this section, we prove that the necessary condition obtained in [12] is also sufficient in order five.

Proposition 4.1. Let B∈M5(F)B\in M_5(F) be a non-zero element and k≥2k\ge2. Define the map

ω:M5(F)×M5(F)⟶M5(F),ω(x,y)=xk+Byk.\omega:M_5(F)\times M_5(F)\longrightarrow M_5(F),\qquad\omega(x,y)=x^k+By^k.

The map ω\omega is surjective if and only if 5>k(r0−1)5>k(r_0-1).

Proof. The nullity of BB is denoted by r0r_0. For r0≤1r_0\le1, surjectivity follows from [12 Theorem 1.2(1)]. For r0≥2r_0\ge2, the inequality 5>k(r0−1)5>k(r_0-1) is necessary by [12 Theorem 1.2(2)]. Thus, it remains to prove that ω\omega is surjective for 2≤r0≤32\le r_0\le3 and 2≤k≤42\le k\le4 such that k(r0−1)<5k(r_0-1)<5, which is precisely,

(r0,k)∈{(2,2),(3,2),(2,3),(2,4)}.(r_0,k)\in\{(2,2),(3,2),(2,3),(2,4)\}.

Denote s=5−r0s=5-r_0. By the reduction in § 2, we may assume that

B=(J^0),J^∈Ms×5(F),rank⁡J^=s.B=\begin{pmatrix}\widehat{J}\\0\end{pmatrix},\qquad\widehat{J}\in M_{s\times5}(F),\qquad\operatorname{rank}\widehat{J}=s.

Let C∈M5(F)C\in M_5(F) be arbitrary element such that

C=(L1L2),L1∈Ms×5(F) and L2∈Mr0×5(F).C=\begin{pmatrix}L_1\\L_2\end{pmatrix},\qquad L_1\in M_{s\times5}(F)\text{ and }L_2\in M_{r_0\times5}(F).

Case 1: k=2k=2. Since r0≤3r_0\leq3, every possible rank of L2L_2 satisfies one of the hypotheses of Lemma 2.3. Hence there exists L1′∈Ms×5(F)L'_1\in M_{s\times5}(F) such that

(L1′L2)=X2\begin{pmatrix}L'_1\\L_2\end{pmatrix}=X^2

for some X∈M5(F)X\in M_5(F). Set H=L1−L1′H=L_1-L'_1. Choose U∈GL5(F)U\in GL_5(F) satisfying

J^U=(0s×r0  Is)=E.\widehat{J}U=(0_{s\times r_0}\ \ I_s)=E.

Since s≤3s\leq3, again by Lemma 2.3, there exists Z∈M5(F)Z\in M_5(F) such that EZ2=HUEZ^2=HU. Taking Y=UZU−1Y=UZU^{-1}, we obtain

J^Y2=J^UZ2U−1=EZ2U−1=H.\widehat{J}Y^2=\widehat{J}UZ^2U^{-1}=EZ^2U^{-1}=H.

Therefore X2+BY2=CX^2+BY^2=C.

Case 2: r0=2r_0=2, k∈{3,4}k\in\{3,4\}. By Lemma 2.5, it suffices to construct a kk-th power matrix

N=(L1′L2)N=\begin{pmatrix}L'_1\\L_2\end{pmatrix}

such that the rank of L1−L1′L_1-L'_1 is at most one or ss. Indeed, writing N=XkN=X^k, the lemma gives Y∈Mn(F)Y\in M_n(F) satisfying J^Yk=L1−L1′\widehat{J}Y^k=L_1-L'_1, and therefore Xk+BYk=CX^k+BY^k=C.

Set E=(I3  0)E=(I_3\ \ 0). If rank⁡(L2)=2\operatorname{rank}(L_2)=2, then Lemma 2.6 gives L1′∈Ms×n(F)L'_1\in M_{s\times n}(F) such that

N=(L1′L2)N=\begin{pmatrix}L'_1\\L_2\end{pmatrix}

is invertible and rank⁡(L1−L1′)=s\operatorname{rank}(L_1-L'_1)=s. Thus NN is a kk-th power with the required rank.

If L2=0L_2=0, take N(t)=tI3⊕02N(t)=tI_3\oplus0_2. Clearly, N=XkN=X^k for X=tkI3⊕02X=\sqrt[k]{t}I_3\oplus0_2. The determinant of the first three columns of L1−tEL_1-tE has leading term −t3-t^3, so a suitable nonzero t0t_0 makes the matrix C−NC-N have rank three.

It remains to consider rank⁡L2=1\operatorname{rank}L_2=1. Recall that L2=(C21PC22P)L_2=\begin{pmatrix}C^P_{21}&C^P_{22}\end{pmatrix}. If rank⁡C21P=1\operatorname{rank}C^P_{21}=1, then C∈Image⁡(ω)C\in\operatorname{Image}(\omega) by [12 Proposition 4.5] as dim⁡(W2)=rank⁡(C21P)\dim(W_2)=\operatorname{rank}(C^P_{21}). We may therefore assume that C21P=0C^P_{21}=0 and $\operatorname{rank}C^P_{22}=1. \end_mid} Let Q∈GL2(F)Q \in GL_2(F) such that QC22PQ−1QC_{22}^{P}Q^{-1} is the Jordan canonical form of C22PC_{22}^{P}. Simultaneous conjugation by S:=I3⊕QS := I_3 \oplus Q preserves the zero lower rows of BB and the rank of J^\hat{J}. Putting C22PC_{22}^{P} in Jordan canonical form, reduces L2L_2 to one of

(μe4T0),(e5T0),μ≠0.\begin{pmatrix} \mu e_4^{\mathsf T} \\ 0 \end{pmatrix}, \qquad\begin{pmatrix} e_5^{\mathsf T} \\ 0 \end{pmatrix}, \qquad\mu\ne0.

For the first form, take N(t)=tI3⊕(μ)⊕0N(t) = tI_3 \oplus(\mu) \oplus0, which is a kk-th power. A suitable nonzero t0t_0 makes H=L1−t0EH = L_1 - t_0E have rank three.

For the second form, if rank⁡L1≠2\operatorname{rank} L_1 \ne2, take N=E45N = E_{45}. By Lemma 2.2, NN is a kk-th power, and CS−NC^S - N is of the form

(L102×5),\begin{pmatrix} L_1 \\ 0_{2 \times5} \end{pmatrix},

having rank at most one or three.

Finally, suppose rank⁡L1=2\operatorname{rank} L_1 = 2. Let c1,c2,c3c_1,c_2,c_3 be its rows, and set

W1=span⁡{c1,c2,c3}.W_1 = \operatorname{span}\{c_1,c_2,c_3\}.

If e5T∈W1e_5^{\mathsf T} \in W_1, choose a basis {e5T,w}\{e_5^{\mathsf T},w\} of W1W_1 and write ci=aie5T+biwc_i = a_i e_5^{\mathsf T} + b_iw for 1≤i≤31 \le i \le3. Take

N=(a1a2a310)e5T.N = \begin{pmatrix} a_1\\ a_2\\ a_3\\ 1\\ 0 \end{pmatrix} e_5^{\mathsf T}.

Then NN and C−NC-N both have rank at most one.

If e5T∉W1e_5^{\mathsf T} \notin W_1, choose independent rows cj,cℓc_j,c_\ell, and let ii be the remaining index. Take

N=(e4+ei)e5T.N = (e_4+e_i)e_5^{\mathsf T}.

The first three rows of C−NC-N are cj,cℓc_j,c_\ell and ci−e5Tc_i-e_5^{\mathsf T}, in their original order. They are linearly independent because e5T∉W1e_5^{\mathsf T} \notin W_1. Hence the matrix C−NC-N has rank three.

In both subcases, NN has rank one and is a kk-th power by Lemma 2.2. Undoing the simultaneous conjugations completes the proof. □

Proof of Theorem (1)

As a corollary, we present the proof of Theorem 1.2(1):

4.1. Proof of Theorem 1.2(1). Since A1A_1 is invertible, let B=A1−1A2B=A_1^{-1}A_2. The nullity(B)(B) = nullity(A2)(A_2). Therefore, if C=xk+BykC=x^k+By^k, then A1C=A1xk+A2ykA_1C=A_1x^k+A_2y^k. Thus, ωB\omega_B is surjective if and only if ω\omega is surjective. Hence, ω\omega is surjective if and only if 5>k(r0−1)5>k(r_0-1).

The case n=6

As proved in § 3, the condition n>k(r−1)n>k(r-1) is necessary but not sufficient for surjectivity. Indeed, for n=6n=6, it holds when r=2r=2 and k=5k=5, whereas the corresponding map is not surjective, as shown in Proposition 3.1. In this section, we prove that for n=6n=6 this is the only exception and obtain the following classification.

We first show that every square-zero nilpotent matrix of rank two in order six has a kk-th root for 2≤k≤42 \leq k \leq4.

Lemma 5.1. Let N∈M6(F)N \in M_6(F) such that rank⁡(N)=2\operatorname{rank}(N)=2 and N2=0N^2=0. Then there exist X∈M6(F)X \in M_6(F) such that N=XkN=X^k for k=2,3,4k=2,3,4.

Proof. The Jordan canonical form of NN is J0,2⊕J0,2⊕02J_{0,2}\oplus J_{0,2}\oplus0_2. For k=2k=2, take X=(J0,4⊕02)X=(J_{0,4}\oplus0_2), for k=3k=3 take X=(J0,5⊕0)X=(J_{0,5}\oplus0) and X=J0,6X=J_{0,6} for k=4k=4. In each case, by Theorem 2.1, XkX^k is similar to NN. □

Proposition 5.2. Let B∈M6(F)B \in M_6(F) be a non-zero element, and k≥2k \geq2 be an integer. The map ω(x,y)=xk+Byk\omega(x,y)=x^k+By^k is surjective on M6(F)M_6(F) if and only if 6>k(r0−1)6>k(r_0-1) and (r0,k)≠(2,5)(r_0,k)\ne(2,5), where r0=nullity⁡(B)r_0=\operatorname{nullity}(B).

Proof. For r0≤1r_0\leq1, surjectivity follows from [12 Theorem 1.2(1)]. For r0≥2r_0\geq2, the condition 6>k(r0−1)6>k(r_0-1) is necessary by [12 Theorem 1.2(2)]. The pairs satisfying this condition are precisely

(r0,k)∈{(2,2),(2,3),(2,4),(2,5),(3,2)}.(r_0,k)\in\{(2,2),(2,3),(2,4),(2,5),(3,2)\}.

For (r0,k)=(2,5)(r_0,k)=(2,5), the map ω\omega is not surjective by Proposition 3.1. Thus, we prove the surjectivity for the remaining pairs.

Let s=6−r0s=6-r_0. By the reduction in § 2, we may assume that B=(J^0)B=\begin{pmatrix}\widehat{J}\\0\end{pmatrix}, where J^∈Ms×6(F)\widehat{J}\in M_{s\times6}(F) and rank⁡J^=s\operatorname{rank}\widehat{J}=s.

Let C∈M6(F)C\in M_6(F) be arbitrary, and we write C=(L1L2)C=\begin{pmatrix}L_1\\L_2\end{pmatrix}, where L1∈Ms×6(F)L_1\in M_{s\times6}(F) and L2∈Mr0×6(F)L_2\in M_{r_0\times6}(F).

Case 1: (r0,k)=(3,2)(r_0,k)=(3,2). By Lemma 2.3, there exists X∈M6(F)X\in M_6(F) such that X2=(L1′L2)X^2=\begin{pmatrix}L'_1\\L_2\end{pmatrix} for some L1′∈M3×6(F)L'_1\in M_{3\times6}(F). Set H=L1−L1′H=L_1-L'_1. Choose U∈GL⁡6(F)U\in\operatorname{GL}_6(F) satisfying

J^U=(03×3 I3)=E.\widehat{J}U=(0_{3\times3}\ I_3)=E.

Since rank⁡(HU)≤3\operatorname{rank}(HU)\leq3, by Lemma 2.3, there exists Z∈M6(F)Z\in M_6(F) such that EZ2=HUEZ^2=HU. Taking Y=UZU−1Y=UZU^{-1} gives

J^Y2=J^UZ2U−1=H,\widehat{J}Y^2=\widehat{J}UZ^2U^{-1}=H,

and hence X2+BY2=CX^2+BY^2=C.

Case 2: r0=2r_0 = 2, 2≤k≤42 \le k \le4. By Lemma 2.5, it suffices to construct a kk-th power

N=(L1′L2)N = \begin{pmatrix} L'_1 \\ L_2 \end{pmatrix}

such that

rank⁡(L1−L1′)∈{0,1,s}.\operatorname{rank}(L_1 - L'_1) \in\{0,1,s\}.

Indeed, writing N=XkN = X^k, the lemma gives Y∈Mn(F)Y \in M_n(F) satisfying J^Yk=L1−L1′\hat{J}Y^k = L_1 - L'_1, and therefore Xk+BYk=CX^k + BY^k = C.

Set E=(I4  0)E = (I_4\ \ 0). If rank⁡(L2)=2\operatorname{rank}(L_2) = 2, then Lemma 2.6 gives L1′∈Ms×n(F)L'_1 \in M_{s \times n}(F) such that

N=(L1′L2)N = \begin{pmatrix} L'_1 \\ L_2 \end{pmatrix}

is invertible and rank⁡(L1−L1′)=s\operatorname{rank}(L_1 - L'_1) = s. Thus NN is a kk-th power with the required rank.

If L2=0L_2 = 0, take N(t)=(tI4⊕02)=(tkI4⊕02)kN(t) = (tI_4 \oplus0_2) = (\sqrt[k]{t}I_4 \oplus0_2)^k. A suitable nonzero t0t_0 makes L1−t0EL_1 - t_0E have rank four.

It remains to consider rank⁡(L2)=1\operatorname{rank}(L_2) = 1. Recall Section 2, L2=(C21P  C22P)L_2 = \left(C^P_{21}\ \ C^P_{22}\right).

If rank⁡C21P=1\operatorname{rank} C^P_{21} = 1, then CC belongs to the image by [12 Proposition 4.5]. We may therefore assume that C21P=0C^P_{21} = 0 and rank⁡C22P=1\operatorname{rank} C^P_{22} = 1.

A simultaneous conjugation by I4⊕QI_4 \oplus Q, with Q∈GL⁡2(F)Q \in\operatorname{GL}_2(F), preserves the zero lower rows of BB and the rank of J^\hat{J}. Therefore, consider C22PC^P_{22} in Jordan canonical form and thus L2L_2 reduces to one of

(μe5T0),(e6T0),μ≠0.\begin{pmatrix} \mu e_5^{\mathsf T} \\ 0 \end{pmatrix}, \qquad \begin{pmatrix} e_6^{\mathsf T} \\ 0 \end{pmatrix}, \qquad\mu\ne0.

For the first form, take N(t)=tI4⊕(μ)⊕0=XkN(t) = tI_4 \oplus(\mu) \oplus0 = X^k for X=tkI4⊕(μk)⊕0X = \sqrt[k]{t}I_4 \oplus(\sqrt[k]{\mu}) \oplus0. We can choose t0t_0 such that H=L1−t0EH = L_1 - t_0E has rank four.

Now, we consider the second form, C=(L1e6T0)C = \begin{pmatrix} L_1 \\ e_6^{\mathsf T} \\ 0 \end{pmatrix}. If rank⁡L1∈{0,1,4}\operatorname{rank} L_1 \in\{0,1,4\}, take N=E56N = E_{56}, which is a kk-th power by Lemma 2.2 and C−N=(L10)C - N = \begin{pmatrix} L_1 \\ 0 \end{pmatrix}.

Suppose that rank L1=2L_1=2. If e6T∈W1e_6^{\mathsf{T}}\in W_1, choose a basis {e6T,w}\{e_6^{\mathsf{T}},w\} of W1W_1 and write the rows of L1L_1 as ci=aie6T+biwc_i=a_i e_6^{\mathsf{T}}+b_iw for 1≤i≤41\le i\le4. Then

(L1L2)=(a1a2a3a410)e6T+(b1b2b3b400)w=N+M.\begin{pmatrix} L_1 \\ L_2 \end{pmatrix} = \begin{pmatrix} a_1 \\ a_2 \\ a_3 \\ a_4 \\ 1 \\ 0 \end{pmatrix}e_6^{\mathsf{T}} + \begin{pmatrix} b_1 \\ b_2 \\ b_3 \\ b_4 \\ 0 \\ 0 \end{pmatrix}w = N+M.

Thus, CC is the sum of two rank-one matrices. The matrix N=XkN=X^k for some XX by Theorem 2.2 and C−NC-N is again a rank-one matrix.

If e6T∉W1e_6^{\mathsf{T}}\notin W_1. Since rank⁡(L1)=2\operatorname{rank}(L_1)=2, there exist Q∈GL4(F)Q\in GL_4(F) such that QL1=(u1u200)QL_1=\begin{pmatrix}u_1\\u_2\\0\\0\end{pmatrix}. Let S=Q⊕I2S=Q\oplus I_2. Simultaneous conjugation of C=Xk+JykC=X^k+Jy^k by SS gives

(QL1e6T0)=Xk+(QJJ~S−102×6)yk.\begin{pmatrix} QL_1 \\ e_6^{\mathsf{T}} \\ 0 \end{pmatrix} = X^k+ \begin{pmatrix} QJ\widetilde{J}S^{-1} \\ 0_{2\times6} \end{pmatrix}y^k.

where QL1S−1=(u1S−1u2S−100)QL_1S^{-1}=\begin{pmatrix}u_1S^{-1}\\u_2S^{-1}\\0\\0\end{pmatrix}. The conjugation by SS preserves the last two rows of CC, the form of BB, and the condition e6T∉W1e_6^{\mathsf{T}}\notin W_1.

Consider V=span⁡{e1T,e2T,e3T−e5T}V=\operatorname{span}\{e_1^{\mathsf{T}},e_2^{\mathsf{T}},e_3^{\mathsf{T}}-e_5^{\mathsf{T}}\}. Here W1W_1 denotes the row space of the transformed top block QL1S−1QL_1S^{-1}. Both VV and W1+Fe6TW_1+Fe_6^{\mathsf{T}} have dimension three. However, e6T∈W1+Fe6Te_6^{\mathsf{T}}\in W_1+Fe_6^{\mathsf{T}} whereas e6T∉Ve_6^{\mathsf{T}}\notin V. Hence these subspaces are distinct, and we may choose w∈V∖(W1+Fe6T)w\in V\setminus(W_1+Fe_6^{\mathsf{T}}). Let N=(0,0,e6T,w,e6T,0)TN=(0,0,e_6^{\mathsf{T}},w,e_6^{\mathsf{T}},0)^{\mathsf{T}}. Since w∈Vw\in V, we have w=(a,b,c,0,−c,0)w=(a,b,c,0,-c,0), we obtain wN=0wN=0. Also e6TN=0e_6^{\mathsf{T}}N=0, so N2=0N^2=0. The row space of NN is spanned by ww and e6Te_6^{\mathsf{T}}, so rank⁡N=2\operatorname{rank}N=2. By Lemma 5.1, NN is a kk-th power.

Thus, we have

CS−N=(u1S−1u2S−1−e6T−w02×6).C^S-N= \begin{pmatrix} u_1S^{-1}\\ u_2S^{-1}\\ -e_6^{\mathsf{T}}\\ -w\\ 0_{2\times6} \end{pmatrix}.

Proof of Theorem (2)

The first two rows span W1W_1. Since e6T∉W1e_6^{\mathsf T} \notin W_1 and w∉W1+Fe6Tw \notin W_1 + F e_6^{\mathsf T}, the first four rows are linearly independent. Thus rank⁡(CS−N)=4\operatorname{rank}(C^S-N)=4.

Finally, suppose that rank⁡L1=3\operatorname{rank} L_1=3. If e6T∉W1e_6^{\mathsf T} \notin W_1, choose three independent rows of L1L_1, and let ii be the remaining index. Taking N=(e5+ei)e6TN=(e_5+e_i)e_6^{\mathsf T} gives a rank-one kk-th power such that C−NC-N has rank four.

If e6T∈W1e_6^{\mathsf T} \in W_1, a similar construction as above allows us to assume that L1=(u1u2u30)L_1=\begin{pmatrix}u_1\\u_2\\u_3\\0\end{pmatrix} where u1,u2,u3u_1,u_2,u_3 are independent. Let V=span⁡{e1T,e2T,e3T}V=\operatorname{span}\{e_1^{\mathsf T},e_2^{\mathsf T},e_3^{\mathsf T}\}. Since VV and W1W_1 have dimension three and e6T∈W1∖Ve_6^{\mathsf T}\in W_1\setminus V, choose w∈V∖W1w\in V\setminus W_1. Define N=(0,0,0,w,e6T,0)TN=(0,0,0,w,e_6^{\mathsf T},0)^{\mathsf T}. Now, wN=0wN=0 and e6TN=0e_6^{\mathsf T}N=0, and thus N2=0N^2=0. Moreover, rank⁡N=2\operatorname{rank}N=2, so NN is a kk-th power by Lemma 5.1. The matrix C−NC-N has rank four, with the last two rows being zero.

In every case, we have constructed N=XkN=X^k with the prescribed last two rows and rank⁡(L1−L1′)∈{0,1,4}\operatorname{rank}(L_1-L'_1)\in\{0,1,4\}.

Lemma 2.5 therefore gives Y∈M6(F)Y\in M_6(F) satisfying J^Yk=L1−L1′\widehat{J}Y^k=L_1-L'_1. Hence Xk+BYk=CX^k+BY^k=C. □

As a corollary, we obtain:

5.1. Proof of Theorem 1.2(2). Since A1A_1 is invertible, let B=A1−1A2B=A_1^{-1}A_2. The nullity(B)(B) = nullity(A2)(A_2). Therefore, if C=Xk+BYkC=X^k+BY^k, then A1C=A1Xk+A2YkA_1C=A_1X^k+A_2Y^k. Hence, ω\omega is surjective if and only if 6>k(r0−1)6>k(r_0-1) and (r0,k)≠(2,5)(r_0,k)\ne(2,5).

Concluding remarks

The results above complete the surjectivity classification in orders five and six and show that the necessary condition n>k(r0−1)n>k(r_0-1) is not sufficient in any order n≥6n\ge6. For coefficients of nullity two, the case k=n−1k=n-1 is settled by the general obstruction. This leaves the following question for smaller exponents.

Question 6.1. Let n≥7n\ge7 and 2≤k≤n−22\le k\le n-2. For which matrices B∈Mn(F)B\in M_n(F) of nullity two is the map

ωB:Mn(F)×Mn(F)⟶Mn(F),ωB(X,Y)=Xk+BYk,\omega_B:M_n(F)\times M_n(F)\longrightarrow M_n(F),\qquad\omega_B(X,Y)=X^k+BY^k,

surjective?

References

Indian Institute of Science Education and Research Mohali, Knowledge City, Sector 81, Punjab 140306, India

Email address: [email protected]

References

  1. [1]M. Brešar and C. Martínez. Waring Problems Across Algebra. 2026. arXiv: 2603. 07033.
  2. [2]M. Brešar and P. Šemrl. The Waring problem for matrix algebras. In: Israel J. Math. 253.1 (2023), pp. 381–405. ISSN: 0021-2172,1565-8511. DOI: 10 . 1007/ s11856-022-2366-7.arxiv.org/abs/2103.10708
  3. [3]M. Brešar and P. Šemrl. The Waring problem for matrix algebras, II. In: Bull. Lond. Math. Soc. 55.4 (2023), pp. 1880–1889. DOI: 10.1112/blms.12825.DOI
  4. [4]I. G. Gargate and T. C. de Mello. Images of multilinear polynomials on n × n upper triangular matrices over infinite fields. In: Israel J. Math. 252.1 (2022), pp. 337–354. DOI: 10.1007/s11856-022-2350-2.DOI
  5. [5]F. A. Gnutov. On the image of a word map with constants of a simple algebraic group II. In: Zap. Nauchn. Sem. S.-Peterburg. Otdel. Mat. Inst. Steklov. (POMI) 492 (2020). MR4223709, pp. 75–93.
  6. [6]N. Gordeev, B. Kunyavskiĭ, and E. Plotkin. Word maps, word maps with constants and representation varieties of one-relator groups. In: J. Algebra 500 (2018), pp. 390–424. DOI: 10.1016/j.jalgebra.2017.03.016.DOI
  7. [7]M. Larsen and A. Shalev. Word maps and Waring type problems. In: J. Amer. Math. Soc. 22 (2009), pp. 437–466. DOI: 10.1090/S0894-0347-08-00615-2.DOI
  8. [8]T. C. D. Mello and P. Saini. Generalized Commutator on Octonion Algebra. 2026. arXiv: 2610.04466 [math.RA].arxiv.org/abs/2610.04466
  9. [9]S. Panja and S. Prasad. The image of polynomials and Waring type problems on upper triangular matrix algebras. In: J. Algebra 631 (2023), pp. 148–193. DOI: 10.1016/j.jalgebra.2023.04.027.DOI
  10. [10]S. Panja, P. Saini, and A. Singh. Images of polynomial maps with constants. In: Mathematika 71 (2025), Paper No. e70031 (17 pages). DOI: 10.1112/mtk.70031.DOI
  11. [11]S. Panja, P. Saini, and A. Singh. Polynomial maps with constants on split octonion algebras. In: Comm. Algebra 54 (2026), pp. 579–594. DOI: 10.1080/00927872. 2025.2531559.
  12. [12]P. Saini and A. Singh. Polynomial maps with constants on matrix algebra. In: Linear Algebra and its Applications 752 (2027), pp. 256–277. ISSN: 0024-3795. DOI: 10.1016/j.laa.2026.08.016.DOI

Paper details

vlnc:W_qkm69m5g
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY-SA 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents