TheoremBase

Induction on the number m of equations: a zero first column gives the solution e1e_1; otherwise a row with nonzero first entry eliminates the first unknown from the other rows, and a nonzero solution of the smaller system is extended by back-substitution.

Proof

Each result cited is universally quantified over the data in its own statement. R\mathbb{R} is the ordered field of the setting of Metric Spaces: Balls, the Metric Topology, and the Real and Complex Numbers as Metric Spaces, sums are as in Sums and Products over a Finite Set and over an Interval §sums, and negatives and differences in R\mathbb{R} as in Negatives, Differences, Reciprocals and Quotients §negative. For p∈Np\in\mathbb{N} and u∈Rpu\in\mathbb{R}^{p}, by Euclidean Space Rn\mathbb{R}^n: n-Tuples of Real Numbers with Their Componentwise Operations §equality and Euclidean Space Rn\mathbb{R}^n: n-Tuples of Real Numbers with Their Componentwise Operations §origin, u=0u=0 if and only if uj=0u_{j}=0 for every j∈[p]j\in[p]; so u≠0u\neq0 if and only if uj≠0u_{j}\neq0 for some j∈[p]j\in[p].

Step 1 (a zero first column). Let m,n∈Nm,n\in\mathbb{N} and A∈Mm×n(R)A\in\mathcal{M}_{m\times n}(\mathbb{R}) with Ai1=0A_{i1}=0 for every i∈[m]i\in[m]; here 1∈[n]1\in[n] by Arithmetic and Order of the Natural Numbers §least. For the standard basis vector e1∈Rne_{1}\in\mathbb{R}^{n} of Identity Matrix and Standard Basis Vectors §basis, Algebraic Rules for Matrix Products, Matrix-Vector Products and Transposes §column gives (Ae1)i=Ai1=0(Ae_{1})_{i}=A_{i1}=0 for every i∈[m]i\in[m], so Ae1=0Ae_{1}=0; and (e1)1=δ11=1≠0(e_{1})_{1}=\delta_{11}=1\neq0 by Identity Matrix and Standard Basis Vectors §identity, so e1≠0e_{1}\neq0.

Step 2 (eliminating the first unknown). Let m,n′∈Nm,n'\in\mathbb{N}, n=n′+1n=n'+1, A∈Mm×n(R)A\in\mathcal{M}_{m\times n}(\mathbb{R}) and r∈[m]r\in[m] with Ar1≠0A_{r1}\neq0, and let cc be the reciprocal Ar1−1A_{r1}^{-1} of Ar1≠0A_{r1}\neq0 in the field R\mathbb{R}, by Negatives, Differences, Reciprocals and Quotients §reciprocal, so Ar1c=1A_{r1}c=1. By Bijections between Intervals: the Gap Map and the Reversal §gap (with n′n' for nn and j=1j=1), g1g_{1} is a bijection from [n′][n'] onto [n]∖{1}[n]\setminus\{1\}; by Basic Properties of Functions: Equality, Composition, Identity, Inverse and Restriction §inverse its inverse is a map h:[n]∖{1}→[n′]h:[n]\setminus\{1\}\to[n'], and h(g1(k))=kh(g_{1}(k))=k for k∈[n′]k\in[n'] by Basic Properties of Functions: Equality, Composition, Identity, Inverse and Restriction §injective-inverse, g1g_{1} being injective. Let y∈Rn′y\in\mathbb{R}^{n'}, for i∈[m]i\in[m] put Si=∑k=1n′Ai,g1(k)ykS_{i}=\sum_{k=1}^{n'}A_{i,g_{1}(k)}y_{k}, and let t=−(c Sr)t=-(c\,S_{r}). By Maps Defined by Cases §cases, on the set [n][n] with the predicative property j=1j=1 and the expressions tt and y(h(j))y(h(j)) (the latter used properly for j≠1j\neq1, since then j∈[n]∖{1}j\in[n]\setminus\{1\}), there is exactly one map x:[n]→Rx:[n]\to\mathbb{R} with x1=tx_{1}=t and xj=yh(j)x_{j}=y_{h(j)} for j∈[n]j\in[n], j≠1j\neq1. Thus x∈Rnx\in\mathbb{R}^{n} by Euclidean Space Rn\mathbb{R}^n: n-Tuples of Real Numbers with Their Componentwise Operations §membership, and xg1(k)=ykx_{g_{1}(k)}=y_{k} for k∈[n′]k\in[n'], as g1(k)≠1g_{1}(k)\neq1. By Matrix-Vector Product §product and Iterated Operations over an Interval: Extracting One Term and Reversing the Order §extraction (with n′n' for nn, j=1j=1, and the terms AijxjA_{ij}x_{j}), for every i∈[m]i\in[m],

(∗)(Ax)i=Si+Ai1t=Si−(Ai1c)Sr,(\ast)\qquad (Ax)_{i}=S_{i}+A_{i1}t=S_{i}-(A_{i1}c)S_{r},

the last step by Rules of Arithmetic in a Commutative Ring: Zero, Signs and Squares, and No Zero Divisors in a Field §signs and the difference of Negatives, Differences, Reciprocals and Quotients §negative. (i) For i=ri=r, (∗)(\ast) and Ar1c=1A_{r1}c=1 give (Ax)r=Sr−Sr=0(Ax)_{r}=S_{r}-S_{r}=0.

(ii) If y≠0y\neq0, say yk≠0y_{k}\neq0 with k∈[n′]k\in[n'], then xg1(k)=yk≠0x_{g_{1}(k)}=y_{k}\neq0, so x≠0x\neq0.

(iii) Let moreover m=m′+1m=m'+1 with m′∈Nm'\in\mathbb{N}. By Bijections between Intervals: the Gap Map and the Reversal §gap, grg_{r} is a bijection from [m′][m'] onto [m]∖{r}[m]\setminus\{r\}. By Maps and Relations Given by Formulas §binary, there is exactly one map A′:[m′]×[n′]→RA':[m']\times[n']\to\mathbb{R}, that is, A′∈Mm′×n′(R)A'\in\mathcal{M}_{m'\times n'}(\mathbb{R}), with

Ai′k′=Agr(i′),g1(k)−Agr(i′),1 c Ar,g1(k)(i′∈[m′], k∈[n′])A'_{i'k}=A_{g_{r}(i'),g_{1}(k)}-A_{g_{r}(i'),1}\,c\,A_{r,g_{1}(k)}\qquad(i'\in[m'],\ k\in[n'])

(gr(i′)∈[m]g_{r}(i')\in[m] and g1(k)∈[n]g_{1}(k)\in[n] by Bijections between Intervals: the Gap Map and the Reversal §gap, so these entries exist; Ai′k′=A′((i′,k))A'_{i'k}=A'((i',k)) by Real Matrices: the Set of Real m x n Matrices with Their Entrywise Operations §entries). For i′∈[m′]i'\in[m'] and i=gr(i′)i=g_{r}(i'), (∗)(\ast), Finite Sums in a Commutative Ring and in an Ordered Field: Distributivity, Differences, Telescoping, Constant Terms, Comparison and the Triangle Inequality §distributive (for the commutative ring R\mathbb{R}) and Finite Sums in a Commutative Ring and in an Ordered Field: Distributivity, Differences, Telescoping, Constant Terms, Comparison and the Triangle Inequality §difference give (Ax)i=∑k=1n′(Ai,g1(k)yk−Ai1cAr,g1(k)yk)=∑k=1n′Ai′k′yk=(A′y)i′(Ax)_{i}=\sum_{k=1}^{n'}\bigl(A_{i,g_{1}(k)}y_{k}-A_{i1}cA_{r,g_{1}(k)}y_{k}\bigr)=\sum_{k=1}^{n'}A'_{i'k}y_{k}=(A'y)_{i'}, the second equality termwise by the rule (x−y)z=xz−yz(x-y)z=xz-yz of Rules of Arithmetic in a Commutative Ring: Zero, Signs and Squares, and No Zero Divisors in a Field §signs, the last by Matrix-Vector Product §product.

Step 3 (the class for the induction). For m∈Nm\in\mathbb{N} let P(m)P(m) be the statement: for every n∈Nn\in\mathbb{N} with m<nm<n and every A∈R[m]×[n]A\in\mathbb{R}^{[m]\times[n]} there is x∈R[n]x\in\mathbb{R}^{[n]} with x(j)≠0x(j)\neq0 for some j∈[n]j\in[n] such that, for every i∈[m]i\in[m] and every map a:[n]→Ra:[n]\to\mathbb{R} with a(j)=A((i,j))⋅x(j)a(j)=A((i,j))\cdot x(j) for all j∈[n]j\in[n], ∑k=1nak=0\sum_{k=1}^{n}a_{k}=0. For given AA, xx and ii exactly one such map aa exists, by Maps and Relations Given by Formulas §map; hence, by Real Matrices: the Set of Real m x n Matrices with Their Entrywise Operations §matrices, Real Matrices: the Set of Real m x n Matrices with Their Entrywise Operations §entries, Euclidean Space Rn\mathbb{R}^n: n-Tuples of Real Numbers with Their Componentwise Operations §space, Matrix-Vector Product §product and the description of u≠0u\neq0 above, P(m)P(m) says exactly that the lemma holds for this mm and all nn and AA. Written out, P(m)P(m) quantifies over the sets nn, AA, xx, ii, jj and aa only: the condition that aa is a map [n]→R[n]\to\mathbb{R} is predicative by Predicativity under Renaming of Set Variables by Variables, of Formulas with Ordered Pairs, and of Formulas for Relations, Functions, Maps, Injectivity, Surjectivity and Bijectivity §map; the values A((i,j))A((i,j)), x(j)x(j), a(j)a(j) and the product in R\mathbb{R} enter predicatively by Predicative Substitution of Values of Functions and Binary Operations, and Predicative Formulas for Negatives, Differences, Absolute Values, Distances, Reciprocals and Quotients in an Ordered Field §value-substitution and Predicative Substitution of Values of Functions and Binary Operations, and Predicative Formulas for Negatives, Differences, Absolute Values, Distances, Reciprocals and Quotients in an Ordered Field §operation-substitution, applied successively for the nested term A((i,j))⋅x(j)A((i,j))\cdot x(j), first to the product and then to its operands (equivalently, by Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols §elimination); and ∑k=1nak=0\sum_{k=1}^{n}a_{k}=0 is ∃w (it(w,n,a)∧w=0)\exists w\,(\mathrm{it}(w,n,a)\wedge w=0) with the predicative formula it\mathrm{it} of Iterated Operations Are Given by a Predicative Formula §predicative for the addition of R\mathbb{R} (the sum over {1,…,n}\{1,\dots,n\} agreeing with the iterated operation by Sums and Products over a Finite Set and over an Interval §intervals), its parameters being R\mathbb{R}, the addition of R\mathbb{R}, N\mathbb{N} and the order and addition of N0\mathbb{N}_{0}. The remaining symbols ([m][m], [n][n], [m]×[n][m]\times[n], R[m]×[n]\mathbb{R}^{[m]\times[n]}, R[n]\mathbb{R}^{[n]}, 00, <<) are defined set symbols or class terms used properly. So P(m)P(m) is predicative by Predicativity of Expressions with Abbreviations, and Symbol-Free Forms of the Basic Defined Set Symbols §closure, and The Class Comprehension Theorem for Predicative Formulas §comprehension gives a class ZZ such that, for every set mm, m∈Zm\in Z if and only if m∈Nm\in\mathbb{N} and P(m)P(m). Then Z⊆NZ\subseteq\mathbb{N}.

In Steps 4 and 5, n∈Nn\in\mathbb{N} is given with ℓ<n\ell<n, where ℓ=1\ell=1 in Step 4 and ℓ=m+1\ell=m+1 in Step 5. As 1≤ℓ1\le\ell by Arithmetic and Order of the Natural Numbers §least, Arithmetic and Order of the Natural Numbers §partial-order gives 1<n1<n, so n≠1n\neq1 by Arithmetic and Order of the Natural Numbers §trichotomy, and n=n′+1n=n'+1 for some n′∈Nn'\in\mathbb{N} by Arithmetic and Order of the Natural Numbers §predecessor.

Step 4 (1∈Z1\in Z). Let m=1m=1 and A∈M1×n(R)A\in\mathcal{M}_{1\times n}(\mathbb{R}). Here [1]={1}[1]=\{1\}, by Arithmetic and Order of the Natural Numbers §partial-order. If A11=0A_{11}=0, Step 1 gives x=e1x=e_{1}. Otherwise take r=1r=1 and, in Step 2, y=e1∈Rn′y=e_{1}\in\mathbb{R}^{n'}, which is nonzero as in Step 1 (1∈[n′]1\in[n'] by Arithmetic and Order of the Natural Numbers §least); the resulting x∈Rnx\in\mathbb{R}^{n} satisfies x≠0x\neq0 by (ii) and (Ax)1=0(Ax)_{1}=0 by (i), so Ax=0Ax=0 as [1]={1}[1]=\{1\}. Hence P(1)P(1), and 1∈Z1\in Z.

Step 5 (m∈Zm\in Z implies m+1∈Zm+1\in Z). Let m∈Zm\in Z, now with m+1<nm+1<n, and let A∈M(m+1)×n(R)A\in\mathcal{M}_{(m+1)\times n}(\mathbb{R}). From m+1<n′+1m+1<n'+1, Arithmetic and Order of the Natural Numbers §order gives m<n′m<n'. If Ai1=0A_{i1}=0 for every i∈[m+1]i\in[m+1], Step 1 gives x=e1x=e_{1}. Otherwise choose r∈[m+1]r\in[m+1] with Ar1≠0A_{r1}\neq0 and form A′∈Mm×n′(R)A'\in\mathcal{M}_{m\times n'}(\mathbb{R}) as in (iii), with m′=mm'=m. Since m∈Zm\in Z and m<n′m<n', P(m)P(m) gives y∈Rn′y\in\mathbb{R}^{n'} with y≠0y\neq0 and A′y=0A'y=0. Let xx be the point of Step 2 for this yy. Then x≠0x\neq0 by (ii); (Ax)r=0(Ax)_{r}=0 by (i); and (Ax)gr(i′)=(A′y)i′=0(Ax)_{g_{r}(i')}=(A'y)_{i'}=0 for every i′∈[m]i'\in[m] by (iii), where gr(i′)∈[m+1]g_{r}(i')\in[m+1] by Bijections between Intervals: the Gap Map and the Reversal §gap. As grg_{r} maps [m][m] onto [m+1]∖{r}[m+1]\setminus\{r\}, every i∈[m+1]i\in[m+1] is rr or gr(i′)g_{r}(i') for some i′∈[m]i'\in[m], so (Ax)i=0(Ax)_{i}=0 for every i∈[m+1]i\in[m+1] and Ax=0Ax=0. Hence P(m+1)P(m+1), and m+1∈Zm+1\in Z.

By Steps 4 and 5, Arithmetic and Order of the Natural Numbers §induction gives Z=NZ=\mathbb{N}. For the given m<nm<n and AA, P(m)P(m) yields x∈Rnx\in\mathbb{R}^{n} with x≠0x\neq0 and Ax=0Ax=0, as claimed.

Citations

Loading…

Dependencies

Uses0

Loading…

Comments

Log in to comment.

Loading…