Error estimation in numerical solution of equations and systems of equations

9 years ago

Abstract

In [7], [8], M. Urabe studies the numerical convergence and error estimation in the case of operatorial equation solution by means of iteration methods. Urabe’s results refer to operatorial equations in complete metric spaces, while as application the numerical convergence of Newton’s method in Banach spaces is studied. Using Urabe’s results, M. Fujii [1] studies the same problems for Steffensen’s method and the chord method applied to equations with real functions. In [6], Urabe’s method is applied to a large class of iteration methods with arbitrary convergence order. We propose further down to extend Urabe’s results to the case of the Gauss-Seidel method for systems of equations in metric spaces.

Authors

Ion Păvăloiu
(Tiberiu Popoviciu Institute of Numerical Analysis)

Keywords

systems of equations in metric spaces; error estimations; Gauss-Seidel method

PDF

Scanned paper: on the journal website.

PDF-LaTeX version of the paper.

Cite this paper as:

I. Păvăloiu, Error estimation in numerical solution of equations and systems of equations, Rev. Anal. Numér. Théor. Approx., 21 (1992) no. 2, pp. 153-165.

About this paper

Print ISSN

1222-9024

Online ISSN

2457-8126

References

[1] Fujii, M., Remarks to accelerated iterative processes for numerical solution of equations, J. Sci. Hiroshima Univ. Ser. A-I, 27 (1963), 97–118.

[2] Lanncaster, P., Error for the Newton-Raphson method, Numerische Mathematik, 9, 1, (1966), 55–68.

[3] Ostrowski, A., The round of stability of iterations, Z.A.M.M. 47, (1967), 77–81

[4] Pavaloiu, I., Estimation des erreurs dans la resolution numerique des systemes d’equations dans des espaces metriques. Seminar of Functional Analysis and Numerical Methods. Preprint Nr. 1, (1987), 121–129.

[5] Pavaloiu, I., Delimitations des erreurs dans la resolution numerique des systemes d’equations, Research Seminars. Seminar on Mathematical analysis. Preprint Nr. 7, (1988), 167–178.

[6] Pavaloiu, I., Introducere in teoria aproximarii solutiilor ecuatiilor. Ed. Dacia, Cluj- Napoca, (1976).

[7] Urabe, M., Convergence of numerical iteration in solution of equations. J. Sci. Hiroshima, Univ.Ser. A, 19 (1956), 479–489.

[8] Urabe, M., Error estimation in numerical solution of equations by iteration processes, J. Sci. Hiroshima Univ. Ser. A-I, 26, (1962), 77–91.

[9] Varga, R.S., Matrix Iterative Analysis, Englewood Cliffs H.J. Prentice Hall (1962).

Paper (preprint) in HTML form

/*====================================================================== Core CSS for LaTeXML documents converted to (X)HTML */ /* Generic Page layout */ .ltx_page_header, .ltx_page_footer { font-size:0.8em; } .ltx_page_header *[rel~="prev"], .ltx_page_footer *[rel~="prev"] { float:left; } .ltx_page_header *[rel~="up"], .ltx_page_footer *[rel~="up"] { display:block; text-align:center; } .ltx_page_header *[rel~="next"], .ltx_page_footer *[rel~="next"] { float:right; } /* What was I trying for here; need more selective rule! .ltx_page_header .ltx_ref, .ltx_page_footer .ltx_ref { margin:0 1em; } */ .ltx_page_header li { padding:0.1em 0.2em 0.1em 1em;} /* Main content */ .ltx_page_content { clear:both; } .ltx_page_header { border-bottom:1px solid; margin-bottom:5px; } .ltx_page_footer { clear:both; border-top:1px solid; margin-top:5px; } .ltx_page_header:after, .ltx_page_footer:after, .ltx_page_content:after { content:"."; display:block; height:0; clear:both; visibility:hidden; } .ltx_page_footer:before { content:"."; display:block; height:0; clear:both; visibility:hidden; } .ltx_page_logo { font-size:80%; margin-top: 5px; clear:both; float:right; } .ltx_page_logo a { font-variant: small-caps; } .ltx_page_logo img { vertical-align:-3px; } /* if shown */ .ltx_page_navbar li { white-space:nowrap; display:block; overflow:hidden; } /* If ref got turned into span, it's "this section"*/ .ltx_page_navbar li span.ltx_ref { white-space:normal; overflow:visible; } /* Ought to be easily removable/overridable? */ .ltx_pagination.ltx_role_newpage { height:2em; } /*====================================================================== Document Structure; Titles & Frontmatter */ /* undo bold here to remove the browser's native h# styling, at let all other styles override it (with more specific rules)*/ .ltx_title { font-size:100%; font-weight:normal; } /* Hack to simulate run-in! put class="ltx_runin" on a title or tag for it to run-into the following text. */ .ltx_runin { display:inline; } .ltx_runin:after { content:" "; } .ltx_runin + .ltx_para, .ltx_runin + .ltx_para p, .ltx_runin + p { display:inline; } .ltx_outdent { margin-left: -2em; } /* .ltx_chapter_title, etc should be in ltx-article.css etc. */ .ltx_page_main { margin:0px; padding:1em 3em 1em 2em; } .ltx_tocentry { list-style-type:none; } /* support for common author block layouts.*/ /* add class ltx_authors_1line to get authors in single line with pop-up affiliation, etc. */ .ltx_authors_1line .ltx_creator, .ltx_authors_1line .ltx_author_before, .ltx_authors_1line .ltx_author_after { display:inline;} .ltx_authors_1line .ltx_author_notes { display:inline-block; } .ltx_authors_1line .ltx_author_notes:before { content:"*"; color:blue;} .ltx_authors_1line .ltx_author_notes span { display:none; } .ltx_authors_1line .ltx_author_notes:hover span { display:block; position:absolute; z-index:10; background:white; text-align:left; border: 1px solid black; border-radius: 0 5px 5px 5px; box-shadow: 5px 5px 10px gray; } /* add class=ltx_authors_multiline to get authors & affliations on separate lines*/ .ltx_authors_multiline .ltx_creator, .ltx_authors_multiline .ltx_author_before, .ltx_authors_multiline .ltx_author_after, .ltx_authors_multiline .ltx_author_notes, .ltx_authors_multiline .ltx_author_notes .ltx_contact { display:block; } /*====================================================================== Para level */ .ltx_float { margin: 1ex 3em 1ex 3em; } td.ltx_subfigure, td.ltx_subtable, td.ltx_subfloat { width:50%; } /* theorems, figure, tables, floats captions.. */ /*====================================================================== Blocks, Lists, Floats */ .ltx_p, .ltx_quote, .ltx_block, .ltx_para { display: block; } /* alignment within blocks */ .ltx_align_left { text-align:left; } .ltx_align_right { text-align:right; } .ltx_align_center { text-align:center; } .ltx_align_justify { text-align:justify; } .ltx_align_top { vertical-align:top; } .ltx_align_bottom { vertical-align:bottom; } .ltx_align_middle { vertical-align:middle; } .ltx_align_baseline { vertical-align:baseline; } .ltx_align_floatleft { float:left; } .ltx_align_floatright { float:right; } .ltx_td.ltx_align_left, .ltx_th.ltx_align_left, .ltx_td.ltx_align_right, .ltx_th.ltx_align_right, .ltx_td.ltx_align_center, .ltx_th.ltx_align_center { white-space:nowrap; } .ltx_td.ltx_align_left.ltx_wrap, .ltx_th.ltx_align_left.ltx_wrap, .ltx_td.ltx_align_right.ltx_wrap, .ltx_th.ltx_align_right.ltx_wrap, .ltx_td.ltx_align_center.ltx_wrap, .ltx_th.ltx_align_center.ltx_wrap, .ltx_td.ltx_align_justify, .ltx_th.ltx_align_justify { white-space:normal; } .ltx_tabular .ltx_tabular { width:100%; } .ltx_inline-block { display:inline-block; } /* equations in non-aligned mode (not normally used) */ .ltx_eqn_div { display:block; width:95%; text-align:center; } /* equations in aligned mode (aligning tags, etc as well as equations) */ .ltx_eqn_table { display:table; width:100%; border-collapse:collapse; } .ltx_eqn_row { display:table-row; } .ltx_eqn_cell { display:table-cell; width:auto; } /* Padding between column pairs in ams align */ table.ltx_eqn_align tr.ltx_equation td.ltx_align_left + td.ltx_align_right, table.ltx_eqn_align tr.ltx_equation td.ltx_align_left + td.ltx_align_center, table.ltx_eqn_align tr.ltx_equation td.ltx_align_center + td.ltx_align_right, table.ltx_eqn_align tr.ltx_equation td.ltx_align_center + td.ltx_align_center { padding-left:3em; } table.ltx_eqn_eqnarray tr.ltx_eqn_lefteqn + tr td.ltx_align_right { min-width:2em; } .ltx_eqn_eqno { max-width:0em; overflow:visible; white-space: nowrap; } .ltx_eqn_eqno.ltx_align_right .ltx_tag { float:right; } .ltx_eqn_center_padleft, .ltx_eqn_center_padright { width:50%; min-width:2em;} .ltx_eqn_left_padleft, .ltx_eqn_right_padright { min-width:2em; } .ltx_eqn_left_padright, .ltx_eqn_right_padleft { width:100%; } /* Various lists */ .ltx_itemize, .ltx_enumerate, .ltx_description { display:block; } .ltx_itemize .ltx_item, .ltx_enumerate .ltx_item { display: list-item; } /* Position the tag to look like a normal item bullet. */ li.ltx_item > .ltx_tag { display:inline-block; margin-left:-1.5em; min-width:1.5em; text-align:right; } .ltx_item .ltx_tag + .ltx_para, .ltx_item .ltx_tag + .ltx_para .ltx_p { display:inline; } /* NOTE: Need to try harder to get runin appearance? */ dl.ltx_description dt { margin-right:0.5em; float:left; font-weight:bold; font-size:95%; } dl.ltx_description dd { margin-left:5em; } dl.ltx_description dl.ltx_description dd { margin-left:3em; } /* Theorems */ .ltx_theorem {margin:1em 0em 1em 0em; } .ltx_title_theorem { font-size:100%; } /* Bibliographies */ .ltx_bibliography dt { margin-right:0.5em; float:left; } .ltx_bibliography dd { margin-left:3em; } /*.ltx_biblist { list-style-type:none; }*/ .ltx_bibitem { list-style-type:none; } .ltx_bibitem .ltx_tag { font-weight:bold; margin-left:-2em; width:3em; } /*.bibitem-tag + div { display:inline; }*/ .ltx_bib_title { font-style:italic; } .ltx_bib_article .bib-title { font-style:normal !important; } .ltx_bib_journal { font-style:italic; } .ltx_bib_volume { font-weight:bold; } /* Indices */ .ltx_indexlist li { list-style-type:none; } .ltx_indexlist { margin-left:1em; padding-left:1em;} /* Listings */ .ltx_listing { display:block; margin: 1ex 3em 1ex 0em; overflow-x:auto; text-align: left; } .ltx_float .ltx_listing { margin: 0; } .ltx_listingline { white-space:nowrap; min-height:1em; } .ltx_lst_numbers_left .ltx_listingline .ltx_tag { background-color:transparent; margin-left:-3em; width:2.5em; position:absolute; text-align:right; } .ltx_lst_numbers_right .ltx_listingline .ltx_tag { background-color:transparent; width:2.5em; position:absolute; right:3em; text-align:right; } /* position:absolute; left:0em; max-width:0em; text-align:right; } */ .ltx_parbox {text-indent:0em; } /* NOTE that it is CRITICAL to put position:relative outside & absolute inside!! I wish I understood why! Outer box establishes resulting size, neutralizes any outer positioning, etc; inner establishes position of stuff to be rotated */ .ltx_transformed_outer { position:relative; bottom:0pt;left:0pt; overflow:visible; } .ltx_transformed_inner { display:block; position:absolute;bottom:0pt;left:0pt; } .ltx_transformed_inner > .ltx_p {text-indent:0em; margin:0; padding:0; } /* If simulating a table (html5), try to get rowspan to work...sorta? */ span.ltx_rowspan { position:absolute; top:0; bottom:0; } /* by default, p doesn't indent */ .ltx_p { text-indent:0em; white-space:normal; } /* explicit control of indentation (on ltx_para) */ .ltx_indent > .ltx_p:first-child { text-indent:2em!important; } .ltx_noindent > .ltx_p:first-child { text-indent:0em!important; } /*====================================================================== Columns */ .ltx_page_column1 { width:44%; float:left; } /* IE uses % of wrong container*/ .ltx_page_column2 { width:44%; float:right; } .ltx_page_columns > .ltx_page_column1 { width:48%; float:left; } .ltx_page_columns > .ltx_page_column2 { width:48%; float:right; } .ltx_page_columns:after { content:"."; display:block; height:0; clear:both; visibility:hidden; } /*====================================================================== Borders and such */ .ltx_tabular { display:inline-table; border-collapse:collapse; } .ltx_tabular.ltx_centering { display:table; } .ltx_thead, .ltx_tfoot, .ltx_tbody { display:table-row-group; } .ltx_tr { display:table-row; } .ltx_td, .ltx_th { display:table-cell; } .ltx_framed { border:1px solid black;} .ltx_tabular .ltx_td, .ltx_tabular .ltx_th { padding:0.1em 0.5em; } /* regular lines */ .ltx_border_t { border-top:1px solid black; } .ltx_border_r { border-right:1px solid black; } .ltx_border_b { border-bottom:1px solid black; } .ltx_border_l { border-left:1px solid black; } /* double lines */ .ltx_border_tt { border-top:3px double black; } .ltx_border_rr { border-right:3px double black; } .ltx_border_bb { border-bottom:3px double black; } .ltx_border_ll { border-left:3px double black; } /* Light lines */ .ltx_border_T { border-top:1px solid gray; } .ltx_border_R { border-right:1px solid gray; } .ltx_border_B { border-bottom:1px solid gray; } .ltx_border_L { border-left:1px solid gray; } /* Framing */ .ltx_framed_rectangle { border-style:solid; border-width:1px; } .ltx_framed_top { border-top-style:solid; border-top-width:1px; } .ltx_framed_left { border-left-style:solid; border-left-width:1px; } .ltx_framed_right { border-right-style:solid; border-right-width:1px; } .ltx_framed_bottom, .ltx_framed_underline { border-bottom-style:solid; border-bottom-width:1px; } .ltx_framed_topbottom { border-top-style:solid; border-top-width:1px; border-bottom-style:solid; border-bottom-width:1px; } .ltx_framed_leftright { border-left-style:solid; border-left-width:1px; border-right-style:solid; border-right-width:1px; } /*====================================================================== Misc */ /* .ltx_verbatim*/ .ltx_verbatim { text-align:left; } /*====================================================================== Meta stuff, footnotes */ .ltx_note_content { display:none; } /*right:5%; */ .ltx_note_content { max-width: 70%; font-size:90%; left:15%; text-align:left; background-color: white; padding: 0.5em 1em 0.5em 1.5em; border: 1px solid black; border-radius: 0 5px 5px 5px; box-shadow: 5px 5px 10px gray; } .ltx_note_mark { color:blue; } .ltx_note_type { font-weight: bold; } .ltx_note { display:inline-block; text-indent:0; } /* So we establish containing block */ .ltx_note_content .ltx_note_mark { position:absolute; left:0.2em; top:-0.1em; } .ltx_note:hover .ltx_note_content, .ltx_note .ltx_note_content:hover { display:block; position:absolute; z-index:10; } .ltx_ERROR { color:red; } .ltx_rdf { display:none; } .ltx_missing { color:red;} .ltx_nounicode { color:red; } /*====================================================================== SVG (pgf/tikz ?) basics */ /* Stuff appearing in svg:foreignObject */ .ltx_svg_fog foreignObject { margin:0; padding:0; overflow:visible; } .ltx_svg_fog foreignObject > p { margin:0; padding:0; display:block; } /*.ltx_svg_fog foreignObject > p { margin:0; padding:0; display:block; white-space:nowrap; }*/ /*====================================================================== Low-level Basics */ /* Note that LaTeX(ML)'s font model doesn't map quite exactly to CSS's */ /* Font Families => font-family */ .ltx_font_serif { font-family: serif; } .ltx_font_sansserif { font-family: sans-serif; } .ltx_font_typewriter { font-family: monospace; } /* dingbats should be converted to unicode? */ /* Math font families handled within math: script, symbol, fraktur, blackboard ? */ /* Font Series => font-weight */ .ltx_font_bold { font-weight: bold; } .ltx_font_medium { font-weight: normal; } /* Font Shapes => font-style or font-variant */ .ltx_font_italic { font-style: italic; font-variant:normal; } .ltx_font_upright { font-style: normal; font-variant:normal; } .ltx_font_slanted { font-style: oblique; font-variant:normal; } .ltx_font_smallcaps { font-variant: small-caps; font-style:normal; } .ltx_font_oldstyle { font-variant: oldstyle-nums; /* experimental css3 ? Doesn't seem to work!*/ font-style:normal; -moz-font-feature-settings: "onum"; -ms-font-feature-settings: "onum"; -webkit-font-feature-settings: "onum"; font-variant-numeric: oldstyle-nums; } .ltx_font_mathcaligraphic { font-family: "Lucida Calligraphy", "Zapf Chancery","URW Chancery L"; } /* .ltx_font_mathscript { ? } */ cite { font-style: normal; } .ltx_red { color:red; } /*.ltx_centering { text-align:center; margin:auto; }*/ /*.ltx_inline-block.ltx_centering,*/ /* Hmm.... is this right in general? */ .ltx_centering { display:block; margin:auto; text-align:center; } /* Dubious stuff */ .ltx_hflipped { display:inline-block; -moz-transform: scaleX(-1); -o-transform: scaleX(-1); -webkit-transform: scaleX(-1); transform: scaleX(-1); filter: FlipH; -ms-fliter: "FlipH"; } .ltx_vflipped { display:inline-block; -moz-transform: scaleY(-1); -o-transform: scaleY(-1); -webkit-transform: scaleY(-1); transform: scaleY(-1); filter: FlipV; -ms-fliter: "FlipV"; } /* .ltx_phantom handled in xslt */ .ltx_title_document { font-size:170%; text-align:center; margin:0.5em 0 0.5em 0; } .ltx_authors, .ltx_role_author { text-align:center; margin:0.5em 0 0.5em 0; } .ltx_role_author .ltx_personname { font-size: 120%; } .ltx_date { text-align:center; font-size: 120%; margin:0.5em 0 0.5em 0; } .ltx_subtitle { text-align:center; font-size: 120%; padding-left:0.2em; margin-left:-0.5em; } .ltx_title_abstract { text-align:center; font-size: 100%; font-weight:bold; } .ltx_abstract { margin-left:4em; margin-right:4em; } .ltx_title_acknowledgements, .ltx_title_keywords, .ltx_title_classification { text-align:left; font-size: 100%; font-weight:bold; margin:0.5 0 0 0; } .ltx_appendix, .ltx_section, .ltx_subsection, .ltx_subsubsection { margin-top:1.5em; } .ltx_paragraph, .ltx_subparagraph { margin-top:1.0em; } .ltx_title_appendix, .ltx_title_section, .ltx_title_bibliography { font-size:140%; font-weight:bold; margin-bottom:1em; } .ltx_title_subsection { font-size:120%; font-weight:bold; margin-bottom:1em; } .ltx_title_subsubsection { font-size:100%; font-weight:bold; margin-bottom:1em; } /* Paragraph & Subparagraph titles should be runin! */ .ltx_title_paragraph { font-size:100%; font-weight:bold; display:inline; margin-right:1em; } .ltx_paragraph .ltx_title, .ltx_paragraph .ltx_title + .ltx_para, .ltx_paragraph .ltx_title + .ltx_para > .ltx_p { display:inline; } .ltx_title_subparagraph { font-size:100%; font-weight:bold; display:inline; margin-left:2em; margin-right:1em; } .ltx_subparagraph .ltx_title, .ltx_subparagraph .ltx_title + .ltx_para, .ltx_subparagraph .ltx_title + .ltx_para > .ltx_p { display:inline; } .ltx_figure { text-align:center; margin:auto; margin:0.5em; } .ltx_table { text-align:center; margin:auto; margin:0.5em; } /* experimental: flex model, to center short captions, justify long ones */ .ltx_table, .ltx_figure { display:flex; flex-direction:column; align-items:center; } .ltx_table .ltx_caption, .ltx_figure .ltx_caption { text-align:justify; } /* first p in para gets indented */ .ltx_para > .ltx_p:first-child { text-indent:2em; } /* except the initial in a section */ section > .ltx_title +.ltx_para > .ltx_p, section > .ltx_title +.ltx_date +.ltx_para > .ltx_p {text-indent:0em; } .ltx_title_abstract + .ltx_p {text-indent:2em; } .ltx_itemize, .ltx_enumerate { margin-left:1em; } .ltx_theorem { margin-top:0.5em; margin-bottom:0.5em; } .ltx_theorem .ltx_title { margin-bottom:0.1em; } .ltx_theorem .ltx_title + .ltx_para, .ltx_theorem .ltx_title + .ltx_para .ltx_p, .ltx_theorem .ltx_title + .ltx_p { margin-top:0em; } Error estimation in numerical solution of equations and systems of equations

Error estimation in numerical solution
of equations and systems of equations

Ion Păvăloiu
(Cluj-Napoca)

In [7], [8] M. Urabe studies the numerical convergence and error estimation in the case of operatorial equation solution by means of iteration methods. Urabe’s results refer to operatorial equations in complete metric spaces, while as application the numerical convergence of Newton’s method in Banach spaces is studied. Using Urabe’s results, M. Fujii [1] studies the same problems for Steffensen’s method and the chord method applied to equations with real functions. In [6] Urabe’s method is applied to a large class of iteration methods with arbitrary convergence order.

We propose further down to extend Urabe’s results to the case of the Gauss-Seidel method for systems of equations in metric spaces.

1.

For a unitary exposition of the problem, we shall firstly present the ideas on which Urabe’s main results are based.

Let $(E, ρ)$ be a metric space and let $F \subset E$ a complete subset of $E$ . Consider the equation:

(1)

x = T (x)

where $T : F \to E$ .

The following fixed point theorem is well known:

Theorem 1.1.

If the following conditions:

(i₁).

$ρ (T (x_{1}), T (x_{2})) \leq K ρ (x_{1}, x_{2})$ for every $x_{1}, x_{2} \in F$ , where $0 < K < 1$ ;
(ii₁).

there exists at least one element $x_{0} \in F$ such that $x_{1} = T (x_{0}) \in F;$
(iii ${}_{1}) .$

the set $S = {x \in E | ρ (x, x_{1}) \leq \frac{K}{1 - K} ρ (x_{1}, x_{0})} \subseteq F$ ,

is fulfilled, then the following properties hold:

(j₁).

the sequence ${(x_{n})}_{n \geq 0}$ , generated by the successive approximations method:

(2) $x_{n + 1} = T (x_{n}), n = 0, 1, \dots,$

where $x_{0}$ fulfils condition $(i i_{1})$ , is convergent, and if $\bar{x} = lim_{n \to \infty} x_{n},$ then $\bar{x}$ is the solution of equation (1);
(jj₁).

$\bar{x}$ is the unique solution of equation (1) from the set $S$ ;
(jjj₁).

the following inequality holds:

(3) $ρ (\bar{x}, x_{n}) \leq \frac{K^{n}}{1 - K} ρ (x_{1}, x_{0}) .$

The numerical solutions of equations (1) by means of the successive approximations method oblige us to consider instead of $T$ another mapping $T^{*} : F \to E$ which approximates $T$ . Equation (1) is replaced by an approximant equation of the form:

(4)

x = T^{*} (x) .

We shall suppose that for a given $ε > 0$ the mappings $T$ and $T^{*}$ fulfil the condition:

(5)

ρ (T^{*} (x), T (x)) \leq ε, for every x \in F .

Consider thus the iterative method:

(6)

ξ_{n + 1} = T^{*} (ξ_{n}), n = 0, 1, \dots, ξ_{0} = x_{0} .

As to the sequence ${(ξ_{n})}_{n \geq 0}$ , M. Urabe proved the following theorem:

Theorem 1.2.

If the mapping $T$ verifies the hypotheses of Theorem 1.1, the mappings $T$ and $T^{*}$ fulfil condition (5), and the set

S^{*} = {x \in E : ρ (x, ξ_{1}) \leq \frac{K}{1 - K} ρ (ξ_{1}, ξ_{0}) + 2 δ} \subseteq F,

where $δ = \frac{1}{1 - K}$ , then the elements of the sequence ${(ξ_{n})}_{n \geq 0}$ generated by (6) are continued into the set $S^{*}$ , $ρ (x_{n}, ξ_{n}) \leq δ$ for every $n = 0, 1, \dots$ , where ${(x_{n})}_{n \geq 0}$ is the sequence generated by (2), and the solution $\bar{x}$ of equation (1) belongs to the set

S = {x \in E : ρ (x, ξ_{1}) \leq ρ (ξ_{0}, ξ_{1}) + δ} .

Condition (5) imposed to the mapping $T^{*}$ does not lead to the conclusion that the sequence ${(ξ_{n})}_{n \geq 0}$ is convergent; that is why if we suppose that the element $ξ_{n}$ which approximates the solution $\bar{x}$ of (1) is determined with a condition of the form:

(7)

ρ (ξ_{n + 1}, ξ_{n}) \leq η

where $η > 0$ is a given real number, then $η$ cannot be chosen arbitrarily small. In [7] it is shown that, if $η > \frac{2 ε}{1 - K},$ then there exists a natural number $n^{'} \in ℕ$ such that inequality (7) is fulfilled for every $n > n^{'}$ .

Urabe shows that, if $η > \frac{2 ε}{1 - K}$ and $ξ_{n}$ is determined by condition (7), then the following inequality holds:

(8)

ρ (\bar{x}, ξ_{n + 1}) \leq \frac{ε + K η}{1 - K} .

Another situation which often occurs in the numerical solution of equations by successive approximations is that in which the sequence ${(ξ_{n})}_{n \geq 0}$ becomes periodic, that is, there exist two natural numbers $m$ and $n^{''}$ such that:

(9)

ξ_{n} = ξ_{n + m},

for every $n > n^{''}$ . In this case the error estimation is given by the following theorem:

Theorem 1.3.

If the mapping $T$ fulfils the hypotheses of Theorem 1.1, and if the elements of the sequence ${(ξ_{n})}_{\geq 0}$ verify the equalities (9), then for every $n > n^{''}$ the following inequality holds:

(10)

ρ (\bar{x}, ξ_{n}) \leq \frac{ε}{1 - K} .

2.

In what follows, starting from the ideas exposed in the previous Section, we shall attempt to obtain delimitations for error in the case of a Gauss-Seidel-type method for the solution of a system of two equations in complete metric spaces.

Denote by $(X_{i}, ρ_{i}), i = 1, 2$ two complete metric spaces, and let $X = X_{1} \times X_{2}$ the cartesian product of the spaces $X_{1}$ and $X_{2}$ . Consider two mappings, $F_{1} : X \to X_{1}$ and $F_{2} : X \to X_{2}$ , which appear in the following system of equations:

(11)		$x_{1}$	$= F_{1} (x_{1}, x_{2}),$
	$x_{2}$	$= F_{2} (x_{1}, x_{2}) .$

In order to solve system (11), we shall adopt the following Gauss-Seidel-type method:

(12)		$x_{1}^{(n + 1)}$	$= F_{1} (x_{1}^{(n)}, x_{2}^{(n)}),$
	$x_{2}^{(n + 1)}$	$= F_{2} (x_{1}^{(n + 1)}, x_{2}^{(n)}), n = 0, 1, \dots; x_{1}^{(0)}, x_{2}^{(0)} \in X$

In [4], [5], [6] we studied the convergence of the sequences ${(x_{1}^{(n)})}_{n \geq 0}$ and ${(x_{2}^{(n)})}_{n \geq 0}$ generated by (12) with the assumption that the mappings $F_{1}$ and $F_{2}$ fulfil Lipschitz-type conditions on the whole space $X$ . But if

for the numerical solution of system (11) we consider, as previously, the approached mappings $F_{1}^{*}$ and $F_{2}^{*}$ , and impose them to verify conditions of type (5) on the whole space $X$ , then such conditions can restrict considerably their sphere of applicability, especially when the space $X$ is unbounded. For this reason it becomes necessary to study the convergence of the sequences ${(x_{1}^{(n)})}_{n \geq 0}$ and ${(x_{2}^{(n)})}_{n \geq 0}$ with the hypothesis that the mappings $F_{1}$ , and $F_{2}$ fulfil Lipschitz-type conditions on a set $D = D_{1} \times D_{2}$ , where $D_{1} \subset X_{1}$ and $D_{2} \subset X_{2}$ are bounded sets.

Consider two sequences of real numbers, ${(f_{n})}_{n \geq 0}$ and ${(g_{n})}_{n \geq 0}$ , whose terms fulfil the conditions:

(13)		$f_{n}$	$\leq α f_{n - 1} + β g_{n - 1},$
	$g_{n}$	$\leq a f_{n} + b g_{n - 1}, n = 1, 2, \dots,$

where $α, β, a, b$ are nonnegative real numbers, while $f_{n} \geq 0$ and $g_{n} \geq 0$ for every $n = 0, 1, \dots$

We associate to inequalities (13) the following system of equations with the unknowns $k$ and $h$ :

(14)		$α + β h$	$= k h,$
	$a k + b$	$= k h .$

In [6] we showed that if $α, β, a, b$ verify the relations:

(15)		$α + b + a β$	$< 2,$
	$(1 - α) (1 - b) - a β$	$> 0$
	$b > 0, α$	$> 0,$

then the system (14) has the real solutions $(h_{i}, k_{i}),$ $i = 1, 2$ for which $0 < h_{i} k_{i} < 1, i = 1, 2,$ and one of these solutions has both components positive. Let $h_{1} > 0$ and $k_{1} > 0$ be the solution with both components positive; then the elements of the sequences ${(f_{n})}_{n \geq 0}$ and ${(g_{n})}_{n \geq 0}$ verify the relations:

(16)		$f_{n}$	$\leq c h_{1}^{n - 1} k_{1}^{n - 1},$
	$g_{n}$	$\leq c h_{1}^{n} k_{1}^{n - 1}, n = 1, 2, \dots,$

where $c = \max {α f_{0} + β g_{0}, \frac{a f_{1} + b g_{0}}{h_{1}}} .$

If we write $p_{1} = h_{1} k_{1},$ then one sees immediately that $p_{1}$ is one of the roots of the equation

(17)

p^{2} - (b + β a + α) p + b α = 0 .

Let $d_{1} > 0$ be a real number chosen such that the sets:

(18)		$S_{1}$	$= {x \in X_{1} : ρ_{1} (x, x_{1}^{0}) \leq \frac{d_{1}}{(1 - p_{1})}};$
	$S_{2}$	$= {x \in X_{2} : ρ_{2} (x, x_{2}^{0}) \leq \frac{d_{1} h_{1}}{(1 - p_{1})}},$

verify the relations $S_{1} \subseteq D_{1}$ and $S_{2} \subseteq D_{2}$ .

With the above specifications we can state the following theorem:

Theorem 2.1.

If the mappings $F_{1}$ and $F_{2}$ fulfil the conditions:

(i₂).

$ρ_{1} (F_{1} (x_{1}, y_{1}), F_{1} (x_{2}, y_{2})) \leq α ρ_{1} (x_{1}, x_{2}) + β ρ_{2} (y_{1}, y_{2})$ ;

$ρ_{2} (F_{2} (x_{1}, y_{1}), F_{2} (x_{2}, y_{2})) \leq a ρ_{1} (x_{1}, x_{2}) + b ρ_{2} (y_{1}, y_{2})$ ,

for every $(x_{1}, y_{1}), (x_{2}, y_{2}) \in D$ ;
(ii₂).

the numbers $α, β, a, b$ fulfil conditions (15);
(iii₂).

the elements $x_{1}^{(1)}$ and $x_{2}^{(1)}$ of the sequences ${(x_{1}^{(n)})}_{n \geq 0}$ , ${(x_{2}^{(n)})}_{n \geq 0}$ generated by (12) verify the conditions

$ρ_{1} (x_{1}^{(0)}, x_{1}^{(1)}) \leq d_{1}, ρ_{2} (x_{2}^{(0)}, x_{2}^{(1)}) \leq d_{1} h_{1},$

then the following properties hold:
(j₂).

the sequences ${(x_{1}^{(n)})}_{n \geq 0}$ and ${(x_{2}^{(n)})}_{n \geq 0}$ generated by (12) are convergent;
(jj₂).

if we write ${\bar{x}}_{1} = lim_{n \to \infty} x_{1}^{(n)}$ and $\bar{x} = lim_{n \to \infty} x_{2}^{(n)}$ , then $({\bar{x}}_{1}, {\bar{x}}_{2}) \in S = S_{1} \times S_{2}$ , and $({\bar{x}}_{1}, {\bar{x}}_{2})$ is the unique solution of the system (11) contained into the set $S$ ;
(jjj₂).

the following inequalities hold:

(19) $ρ_{1} ({\bar{x}}_{1}, x_{1}^{(n)})$ $\leq \frac{d_{1} p_{1}^{n}}{1 - p_{1}};$

$ρ_{2} ({\bar{x}}_{2}, x_{2}^{(n)})$ $\leq \frac{d_{1} h_{1} p_{1}^{n}}{1 - p_{1}} .$

Proof.

From $(i i i_{2})$ follows $x_{1}^{(1)} \in S_{1}$ and $x_{2}^{(1)} \in S_{2}$ . With this, with (12), and with the hypothesis $(i_{2})$ , we have:

	$ρ_{1} (x_{1}^{(2)}, x_{1}^{(1)})$	$\leq α ρ_{1} (x_{1}^{(1)}, x_{1}^{(0)}) + β ρ_{2} (x_{2}^{(1)}, x_{2}^{(0)}) \leq α d_{1} + β d_{1} h_{1}$
		$= d_{1} (α + β h_{1}) = d_{1} p_{1};$
	$ρ_{2} (x_{2}^{(2)}, x_{2}^{(1)})$	$\leq a ρ_{1} (x_{1}^{(2)}, x_{1}^{(1)}) + b ρ_{2} (x_{2}^{(1)}, x_{2}^{(0)}) S \leq a d_{1} p_{1} + b d_{1} h_{1}$
		$= a d_{1} h_{1} k_{1} + b d_{1} h_{1} = d_{1} h_{1} (a k_{1} + b) = d_{1} p_{1} h_{1} .$

Using the above inequalities and the hypothesis $(i i i_{2})$ , we have:

	$ρ_{1} (x_{1}^{(2)}, x_{1}^{(0)})$	$\leq ρ_{1} (x_{1}^{(2)}, x_{1}^{(1)}) + ρ_{1} (x_{1}^{(1)}, x_{1}^{(0)}) \leq d_{1} + d_{1} p_{1} \leq \frac{d_{1}}{1 - p_{1}};$
	$ρ_{2} (x_{2}^{(2)}, x_{2}^{(0)})$	$\leq ρ_{2} (x_{2}^{(2)}, x_{2}^{(1)}) + ρ_{2} (x_{2}^{(1)}, x_{2}^{(0)}) \leq d_{1} p_{1} h_{1} + d_{1} h_{1} \leq \frac{d_{1} h_{1}}{1 - p} .$

From these inequalities it follows that $x_{1}^{(2)} \in S_{1}$ and $x_{2}^{(2)} \in S_{2}$ .

Suppose now that $x_{1}^{(i)} \in S_{1}, x_{2}^{(i)} \in S_{2}$ for every $i = 1, 2, \dots, k$ , and $ρ_{1} (x_{1}^{(i)}, x_{1}^{(i - 1)}) \leq d_{1} p_{1}^{i - 1}$ , $ρ_{2} (x_{2}^{(i)}, x_{2}^{(i - 1)}) \leq d_{1} h_{1} p_{1}^{i - 1}$ for $1 = 1, 2, \dots, k$ . Using these hypotheses, and $(i_{2})$ together with (12), we deduce:

	$ρ_{1} (x_{1}^{(k + 1)}, x_{1}^{(k)})$	$\leq α ρ_{1} (x_{1}^{(k)}, x_{1}^{(k - 1)}) + β ρ_{2} (x_{2}^{(k)}, x_{2}^{(k - 1)})$
		$\leq d_{1} p_{1}^{k - 1} (α + β h_{1}) = d_{1} p_{1}^{k} .$

Analogously, and taking also into account the above inequality we deduce:

ρ_{2} (x_{2}^{(k + 1)}, x_{2}^{(k)}) \leq d_{1} h_{1} p_{1}^{k} .

From the above inequalities it easily results that $x_{1}^{(k + 1)} \in S_{1}$ and $x_{2}^{(k + 1)} \in S_{2}$ .

The previously proved relations and the induction principle show that the following relations hold:

	$ρ_{1} (x_{1}^{(n + 1)}, x_{1}^{(n)})$	$\leq d_{1} p_{1}^{n},$
	$ρ_{2} (x_{2}^{(n + 1)}, x_{2}^{(n)})$	$\leq d_{1} h_{1} p_{1}^{n},$
	$x_{1}^{(n)} \in S_{1}, x_{2}^{(n)}$	$\in S_{2}, for every n \in ℕ .$

By virtue of the last relations we deduce that for every $n, s \in ℕ$ the following inequalities hold:

	$ρ_{1} (x_{1}^{(n + s)}, x_{1}^{(n)})$	$\leq \frac{d_{1} p_{1}^{n}}{1 - p_{1}};$
	$ρ_{2} (x_{2}^{(n + s)}, x_{2}^{(n)})$	$\leq \frac{d_{1} h_{1} p_{1}^{n}}{1 - p_{1}},$

from which, taking into account the fact that $0 < p_{1} < 1$ , it follows that the sequence ${(x_{1}^{(n)})}_{n \geq 0}$ and ${(x_{2}^{(n)})}_{n \geq 0}$ are fundamental.

Using this remark and the completeness of the spaces $X_{1}$ and $X_{2}$ , it results that $lim_{n \to \infty} x_{1}^{(n)} = {\bar{x}}_{1}$ and $lim_{n \to \infty} x_{2}^{(n)} = {\bar{x}}_{2}$ do exist, and the following inequalities hold:

	$ρ_{1} ({\bar{x}}_{1}, x_{1}^{(n)})$	$\leq \frac{d_{1} p_{1}^{n}}{1 - p_{1}},$
	$ρ_{2} ({\bar{x}}_{2}, x_{2}^{(n)})$	$\leq \frac{d_{1} h_{1} p_{1}^{n}}{1 - p_{1}} .$

One deduces easily that ${\bar{x}}_{1}$ and ${\bar{x}}_{2}$ form the slution of the system (11) and $\bar{x} \in S_{1}$ , ${\bar{x}}_{2} \in S_{2}$ .

The uniqueness of the solution $({\bar{x}}_{1}, {\bar{x}}_{2})$ in $S = S_{1} \times S_{2}$ is verified by reductio ad absurdum, taking into account the fact that

0 < \frac{β a}{(1 - α) (1 - b)} < 1 .

∎

Consider now two mappings, $F_{1}^{*} : D \to X_{1}$ and $F_{2}^{*} : D \to X_{2}$ , where $D = D_{1} \times D_{2}$ . Suppose that the mappings $F_{1}, F_{2} . F_{1}^{*}, F_{2}^{*}$ verify the relations

(20)		$ρ_{1} (F_{1} (u, v), F_{1}^{*} (u, v))$	$\leq δ_{1},$
	$ρ_{2} (F_{2} (u, v), F_{2}^{*} (u, v))$	$\leq δ_{2},$

for every $(u, v) \in D$ , where $δ_{1} > 0, δ_{2} > 0$ are given numbers.

In order to solve the system (11), consider now the approximate procedure:

(21)		$ξ_{1}^{(n + 1)}$	$= F_{1}^{*} (ξ_{1}^{(n)}, ξ_{2}^{(n)}),$
	$ξ_{2}^{(n + 1)}$	$= F_{2}^{*} (ξ_{1}^{(n + 1)}, ξ_{2}^{(n)}), n = 0, 1 \dots,$

where $ξ_{1}^{(0)} = x_{1}^{(0)}, ξ_{2}^{(0)} = x_{2}^{(0)} .$

Write:

(22)		$θ_{1}$	$= \frac{β δ_{2} + (1 - b) δ_{1}}{(1 - α) (1 - b) - a β},$
	$θ_{2}$	$= \frac{(1 - α) δ_{2} + a δ_{1}}{(1 - α) (1 - b) - a β},$

and consider the sets:

(23)		$S_{1}^{*}$	$= {x \in X_{1} : ρ_{1} (x, x_{1}^{(0)}) \leq d_{1} + \frac{d_{1}}{1 - p_{1}} + θ_{1}},$
	$S_{2}^{*}$	$= {x \in X_{2} : ρ_{2} (x, x_{2}^{(0)}) \leq d_{1} h_{1} + \frac{d_{1} h_{1}}{1 - p} + θ_{2}} .$

The following theorem holds:

Theorem 2.2.

If the hypotheses of Theorem 2.1 and the additional conditions:

(i₃).

the mappings $F_{1}, F_{2}, F_{1}^{*}, F_{2}^{*}$ fulfil relations (20);
(ii₃).

$S_{1}^{*} \subseteq D_{1}, S_{2}^{*} \subseteq D_{2}$

are verified, then for every real numbers $ε_{1}$ and $ε_{2}$ which verify the relations $ε_{1} > 2 θ_{1}, ε_{2} > 2 θ_{2}$ there exists a natural number $n^{'} \in ℕ$ such that for every

$n > n^{'}$ the inequalities $ρ_{1} (ξ_{1}^{(n + 1)}, ξ_{1}^{(n)}) < ε_{1}$ and $ρ_{2} (ξ_{2}^{(n + 1)}, ξ_{2}^{(n)}) < ε_{2}$ hold, and the additional properties hold, two:

(j₃).

$ξ_{1}^{(n)} \in S_{1}^{*}, ξ_{2}^{(n)} \in S_{2}^{*}$ for every $n = 0, 1, \dots$ ;

(jj₃).

the following inequalities hold:

(24)		$ρ_{1} ({\bar{x}}_{1}, ξ_{1}^{(n + 1)})$	$\leq \frac{β (a ε_{1} + b ε_{2}) + α ε_{1} (1 - b)}{(1 - α) (1 - b) - a β} + \frac{ε_{1}}{2},$
	$ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n + 1)})$	$\leq \frac{a (α ε_{1} + β ε_{2}) + b ε_{2} (1 - α)}{(1 - α) (1 - b) - a β} + \frac{ε_{2}}{2}$

for every $n > n^{'}$ , where $({\bar{x}}_{1}, {\bar{x}}_{2})$ is the solution of system (11).

Proof.

We show that the relations $(j_{3})$ hold. Indeed, by (21) and (20) it results:

ρ_{1} (x_{1}^{(1)}, ξ_{1}^{(1)}) = ρ_{1} (F_{1} (x_{1}^{(0)}, x_{2}^{(0)}), F_{1}^{*} (ξ_{1}^{(0)}, ξ_{2}^{(0)})) \leq δ_{1},

since we assumed that $x_{1}^{(0)} = ξ_{1}^{(0)}$ and $x_{2}^{(0)} = ξ_{2}^{(0)}$ .

Taking into account the hypothesis $(i i i_{2})$ of Theorem 2.1, we shall have:

ρ_{1} (ξ_{1}^{(1)}, x_{1}^{(0)}) \leq ρ_{1} (ξ_{1}^{(1)}, x_{1}^{(1)}) + ρ_{1} (x_{1}^{(1)}, x_{1}^{(0)}) \leq δ_{1} + d_{1} < d_{1} + \frac{d_{1}}{1 - p_{1}} + θ_{1}

since from (22) it follows $δ_{1} < θ_{1}$ . From the last inequality it follows $ξ_{1}^{(1)} \in S_{1}^{*}$ .

Analogously we have:

	$ρ_{2} (x_{2}^{(1)}, ξ_{2}^{(1)})$	$\leq ρ_{2} (F_{2} (x_{1}^{(1)}, x_{2}^{(0)}), F_{2}^{*} (ξ_{2}^{(1)}, ξ_{2}^{(0)}))$
		$\leq δ_{2} + a ρ_{1} (ξ_{1}^{(1)}, x_{1}^{(1)}) + b ρ_{2} (ξ_{2}^{(0)}, x_{2}^{(0)}) \leq δ_{2} + a δ_{1},$

from which, taking into account $(i i i_{2})$ , it follows:

ρ_{2} (ξ_{2}^{(1)}, x_{2}^{(0)}) \leq ρ_{2} (ξ_{2}^{(1)}, x_{2}^{(1)}) + ρ_{2} (x_{2}^{(1)}, x_{2}^{(0)}) \leq δ_{2} + a δ_{1} + d_{1} h_{1},

but one can easily verify that $δ_{2} + a δ_{1} \leq θ_{2}$ , and hence:

ρ_{2} (ξ_{2}^{(1)}, x_{2}^{(0)}) \leq h_{1} d_{1} + \frac{h_{1} d_{1}}{1 - p_{1}} + θ_{2},

therefore $ξ_{2}^{(1)} \in S_{2}^{*}$ .

Suppose now that $ξ_{1}^{(n - 1)} \in S_{1}^{*}$ and $ξ_{2}^{(n - 1)} \in S_{2}^{*}$ for an $n \geq 2$ . Then we have:

	$ρ_{1} (ξ_{1}^{(n)}, x_{1}^{(n)})$	$= ρ_{1} (F_{1}^{*} (ξ_{1}^{(n - 1)}, ξ_{2}^{(n - 1)}), F_{1} (x_{1}^{(n - 1)}, x_{2}^{(n - 1)}))$
		$\leq α ρ_{1} (x_{1}^{(n - 1)}, ξ_{1}^{(n - 1)}) + β ρ_{2} (x_{2}^{(n - 1)}, ξ_{2}^{(n - 1)}) + δ_{1},$

	$ρ_{2} (ξ_{2}^{(n)}, x_{2}^{(n)})$	$= ρ_{2} (F_{2}^{*} (ξ_{1}^{(n)}, ξ_{2}^{(n - 1)}), F_{2} (x_{1}^{(n)}, x_{2}^{(n - 1)}))$
		$\leq a ρ_{1} (x_{1}^{(n)}, ξ_{1}^{(n)}) + b ρ_{2} (x_{2}^{(n - 1)}, ξ_{2}^{(n - 1)}) + δ_{2} .$

Starting from the above relations, we deduce immediately:

(25)		$ρ_{1} (ξ_{1}^{(n)}, x_{1}^{(n)})$	$\leq d_{1} p_{1}^{n - 1} + θ_{1},$
	$ρ_{2} (ξ_{2}^{(n)}, x_{2}^{(n)})$	$\leq d_{1} h_{1} p_{1}^{n - 1} + θ_{2},$

from which one easily obtains:

	$ρ_{1} (ξ_{1}^{(n)}, x_{1}^{(0)})$	$\leq ρ_{1} (ξ_{1}^{(n)}, x_{1}^{(n)}) + ρ_{1} (x_{1}^{(n)}, x_{1}^{(0)}) \leq d_{1} + \frac{d_{1}}{1 - p_{1}} + θ_{1},$
	$ρ_{2} (ξ_{2}^{(n)}, x_{2}^{(0)})$	$\leq ρ_{2} (ξ_{2}^{(n)}, x_{2}^{(n)}) + ρ_{2} (x_{2}^{(n)}, x_{2}^{(0)}) \leq d_{1} h_{1} + \frac{d_{1} h_{1}}{1 - p_{1}} + θ_{2},$

that is, $ξ_{1}^{(n)} \in S_{1}^{*}$ and $ξ_{2}^{(n)} \in S_{2}^{*}$ .

From the above results it follows:

	$ρ_{1} (ξ_{1}^{(n + 1)}, ξ_{1}^{(n)})$	$\leq α ρ_{1} (ξ_{1}^{(n)}, ξ_{1}^{(n - 1)}) + β ρ_{2} (ξ_{2}^{(n)}, ξ_{2}^{(n - 1)}) + 2 δ_{1},$
	$ρ_{2} (ξ_{2}^{(n + 1)}, ξ_{2}^{(n)})$	$\leq a ρ_{1} (ξ_{1}^{(n + 1)}, ξ_{1}^{(n)}) + b ρ_{2} (ξ_{2}^{(n)}, ξ_{2}^{(n - 1)}) + 2 δ_{2},$

from which one deduces immediately by induction the inequalities:

(26)		$ρ_{1} (ξ_{1}^{(n + 1)}, ξ_{1}^{(n)})$	$\leq d_{1} p_{1}^{n - 1} + 2 θ_{1},$
	$ρ_{2} (ξ_{2}^{(n + 1)}, ξ_{2}^{(n)})$	$\leq d_{1} h_{1} p_{1}^{n - 1} + 2 θ_{2} .$

From relations (26) it follows that, if $ε_{1} > 2 θ_{1}$ and $ε_{2} > 2 θ_{2}$ , then there exists a natural number $n^{'} \in ℕ$ such that for $n > n^{'}$ the relations $ρ_{1} (ξ_{1}^{(n + 1)}, ξ_{1}^{(n)})$ $\leq ε_{1}$ and $ρ_{2} (ξ_{2}^{(n + 1)}, ξ_{2}^{(n)}) \leq ε_{2}$ hold, namely the approximating iterative procedure (21) can be stopped when the distance between two successive iterations becomes smaller than a given number.

Suppose that $ε_{1}$ and $ε_{2}$ were chosen such that for $n > n^{'}$ the inequalities $ρ_{1} (ξ_{1}^{(n + 1)}, ξ_{1}^{(n)}) < ε_{1}$ and $ρ_{2} (ξ_{2}^{(n + 1)}, ξ_{2}^{(n)}) < ε_{2}$ are verified. Then, for $n > n^{'}$ , we have:

	$ρ_{1} ({\bar{x}}_{1}, ξ_{1}^{(n + 1)})$	$\leq α ρ_{1} ({\bar{x}}_{1}, ξ_{1}^{(n)}) + β ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n)}) + δ_{1},$
	$ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n + 1)})$	$\leq a ρ_{1} ({\bar{x}}_{1}, ξ_{1}^{(n + 1)}) + b ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n)}) + δ_{2},$

from which it follows:

	$(1 - α) ρ_{1} ({\bar{x}}_{1}, ξ_{2}^{(n + 1)})$	$\leq α ε_{1} + β ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n)}) + δ_{1},$
	$(1 - b) ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n + 1)})$	$\leq b ε_{2} + a ρ_{1} ({\bar{x}}_{1}, ξ_{1}^{(n + 1)}) + δ_{2},$

namely:

ρ_{2} ({\bar{x}}_{2}, ξ_{2}^{(n + 1)}) \leq \frac{a (α ε_{1} + β ε_{2}) + b (1 - α) ε_{2}}{(1 - α) (1 - b) - a β} + \frac{ε_{2}}{2},

and using this inequality we have:

ρ_{1} ({\bar{x}}_{1}, ξ_{1}^{(n + 1)}) \leq \frac{β (a ε_{1} + b ε_{2}) + α (1 - b) ε_{1}}{(1 - α) (1 - b) - a β} + \frac{ε_{1}}{2} .

∎

3.

We present further down an application of Theorem 2.1 For this purpose, consider the linear system:

(27)

x = A x + b,

where $b^{T} \in ℝ^{n},$ $A \in ℳ_{n} (ℝ),$ and $x^{T} \in ℝ^{n} .$

In order to solve system (27), we shall use a method exposed by R. Varga in [9].

Decompose the matrix $A$ into four blocks of matrices:

M_{1} \in ℳ_{s, s} (ℝ), M_{2} \in ℳ_{s, n - s} (ℝ), M_{3} \in ℳ_{n - s, s} (ℝ), M_{4} \in ℳ_{n - s, n - s} (ℝ),

where $1 \leq s < n$ . The matrix $A$ will then have the form:

A = (\begin{matrix} M_{1} & M_{2} \\ M_{3} & M_{4} \end{matrix}) .

Write $x = (\binom{u}{v})$ and $b = (\binom{b^{'}}{b^{''}}),$ with $u^{T} \in ℝ^{s}$ $b^{', T} \in ℝ^{s}, v^{T} \in ℝ^{n - s}, b^{'', T} \in ℝ^{n - s}$ .

With these notations system (27) will acquire the form:

(28)		$u$	$= M_{1} u + M_{2} v + b^{'},$
	$v$	$= M_{3} u + M_{4} v + b^{''} .$

In order to solve the system (28), we apply the Gauss-Seidel method, that is:

(29)		$u_{i}$	$= M_{1} u_{i - 1} + M_{2} v_{i - 1} + b^{'},$
	$v_{i}$	$= M_{3} u_{i} + M_{4} v_{i - 1} + b^{''}, (u_{0}, v_{0}) \in ℝ^{s} \times ℝ^{n - s}, i = 1, 2, \dots$

If we put in Theorem 2.1 $X_{1} = ℝ^{s}, X_{2} = ℝ^{n - s}, α = ‖ M_{1} ‖, β = ‖ M_{2} ‖, a = ‖ M_{3} ‖, b = ‖ M_{4} ‖$ , where the above norms are of the same kind and are every time considered on the metrics corresponding spaces, then as a consequence of this theorem, we can state the following theorem:

Theorem 3.1.

If the inequalities:

(30)		$‖ M_{1} ‖ + ‖ M_{4} ‖ + ‖ M_{2} ‖ ‖ M_{3} ‖$	$< 2,$
	$(1 - ‖ M_{1} ‖) (1 - ‖ M_{4} ‖)$	$> ‖ M_{3} ‖ \cdot ‖ M_{2} ‖$

hold, then the system (27) has only one solution $\bar{x} = (\bar{u}, \bar{v}) \in ℝ^{s} \times ℝ^{n - s}$ which is obtained as limit of the sequences ${(u_{n})}_{n \geq 0}$ and ${(v_{n})}_{n \geq 0}$ generated by the iterative procedure (29).

Also from Theorem 2.1 one deduces that $(\bar{u}, \bar{v}) \in {\hat{S}}_{1} \times {\hat{S}}_{2},$ where

{\hat{S}}_{1} = {u \in ℝ^{s} : ‖ u - u_{0} ‖ \leq \frac{{\hat{d}}_{1}}{1 - {\hat{p}}_{1}}} and {\hat{S}}_{2} = {v \in ℝ^{n - s} : ‖ v - v_{0} ‖ \geq \frac{{\hat{d}}_{1} {\hat{h}}_{1}}{1 - {\hat{p}}_{1}}},

${\hat{d}}_{1}$ is a positive number for which $‖ u_{1} - u_{0} ‖ \leq {\hat{d}}_{1}$ and $‖ v_{1} - v_{0} ‖ \leq {\hat{d}}_{1} {\hat{h}}_{1}, {\hat{p}}_{1} = {\hat{h}}_{1} {\hat{k}}_{1}$ , while $({\hat{h}}_{1}, {\hat{k}}_{1})$ is the solution with positive components of the system of equations:

(31)		$‖ M_{1} ‖ + ‖ M_{2} ‖ h$	$= h k,$
	$‖ M_{3} ‖ k + ‖ M_{4} ‖$	$= h k .$

Let now $M_{1}^{*} \in ℳ_{s, s} (ℝ), M_{2}^{*} \in ℳ_{s, n - s} (ℝ), M_{3}^{*} \in ℳ_{n - s, s} (ℝ)$ and $M_{4}^{*} \in ℳ_{n - s, n - s} (ℝ)$ be four matrices for which:

‖ M_{i} - M_{i}^{*} ‖ \leq ε, i = 1, 2, 3, 4, ε > 0,

and let $b^{', *} \in ℝ$ , $b^{'', *} \in ℝ^{n - s}$ for which we also have $‖ b^{'} - b^{', *} ‖ < ε$ and $‖ b^{''} - b^{'', *} ‖ < ε$ . Thus, if we consider instead of the procedure (29) the approximate procedure:

(32)		$ξ_{1}^{(i)}$	$= M_{1}^{} ξ_{1}^{(i - 1)} + M_{2}^{} ξ_{2}^{(i - 1)} + b^{', *},$
	$ξ_{2}^{(i)}$	$= M_{3}^{} ξ_{1}^{(i)} + M_{4}^{} ξ_{2}^{(i - 1)} + b^{'', *}, i = 1, 2, \dots,$
	$ξ_{1}^{(0)}$	$= u_{0}, ξ_{2}^{(0)} = v_{0},$

and put into (29) $u_{0} = {\bar{θ}}_{1}, v_{0} = {\bar{θ}}_{2},$ where ${\bar{θ}}_{1}$ is the null vector from $ℝ^{s}$ and ${\bar{θ}}_{2}$ is the null vector from $ℝ^{n - s}$ , it results $u_{1} = b^{'}$ and $v_{1} = b^{''} + M_{3} b^{'}$ and we may consider $d_{1} = \max$ ${‖ b^{'} ‖, ‖ b^{''} ‖ / {\hat{h}}_{1}} .$

If we write:

	$F_{1} (u, v)$	$= M_{1} u + M_{2} v + b^{'},$
	$F_{2} (u, v)$	$= M_{3} u + M_{4} v + b^{''},$
	$F_{1}^{*} (u, v)$	$= M_{1}^{} u + M_{2}^{} v + b^{', *},$
	$F_{2}^{*} (u, v)$	$= M_{3}^{} u + M_{4}^{} v + b^{'', *},$

and take into account the above hypotheses, we shall have:

(33)

‖ F_{i} (u, v) - F_{i}^{*} (u, v) ‖ \leq ε (‖ u ‖ + ‖ v ‖ + 1), i = 1, 2

and if

1 - ε \frac{2 + a + β - b - α}{(1 - α) (1 - b) - a β} > 0,

then, denoting

δ = \frac{(\frac{2 - \hat{p}}{1 - \hat{p}} {\hat{d}}_{1} (1 + h_{1}) + 1)}{1 - ε \frac{2 + a + β - b - α}{(1 - α) (1 - b) - a β}},

it follows from (33) that:

‖ F_{i} (u, v) - F_{i}^{*} (u, v) ‖ \leq δ, i = 1, 2,

for $(u, v) \in {\hat{S}}_{1}^{*} \times {\hat{S}}_{2}^{*},$ ${\hat{S}}_{1}^{*}$ and ${\hat{S}}_{2}^{*}$ being the sets:

	${\hat{S}}_{1}^{*}$	$= {u \in R^{s} : ‖ u ‖ \leq {\hat{d}}_{1} \frac{2 - {\hat{p}}_{1}}{1 - {\hat{p}}_{1}} + {\hat{θ}}_{1}}$
	${\hat{S}}_{2}^{*}$	$= {v \in R^{n - s} : ‖ v ‖ \leq {\hat{d}}_{1} {\hat{h}}_{1} \frac{2 - {\hat{p}}_{1}}{1 - {\hat{p}}_{1}} + {\hat{θ}}_{2}}$

where:

	${\hat{θ}}_{1}$	$= δ \frac{1 + β - b}{(1 - α) (1 - b) - a β},$
	${\hat{θ}}_{2}$	$= \frac{1 + a - α}{(1 - α) (1 - b) - a β}$

Taking all this into account, if $\hat{ε} > 2$ $\max {{\hat{θ}}_{1}, {\hat{θ}}_{2}},$ then there exists $\hat{n} \in ℕ$ such that, for $n > \hat{n}, ‖ ξ_{i}^{(n + 1)} - ξ^{(n)} ‖ < \hat{ε}, i = 1, 2$ , and $ξ_{1}^{(n)} \in {\hat{S}}_{1}^{*}, ξ_{2}^{(n)} \in {\hat{S}}_{2}^{*},$ where ${(ξ_{1}^{(n)})}_{n \geq 0}$ and ${(ξ_{2}^{(n)})}_{n \geq 0}$ are the sequences generated by means of the approximating procedure (32).

Using the conclusions of Theorem 2.2, the following error estimations hold:

	$‖ \bar{u} - ξ_{1}^{(n + 1)} ‖$	$\leq \hat{ε} [\frac{a α + a β + b - b α}{(1 - α) (1 - b) - a β} + \frac{1}{2}],$
	$‖ \bar{v} - ξ_{2}^{(n + 1)} ‖$	$\leq \hat{ε} [\frac{a β + b β + α - α b}{(1 - α) (1 - b) - a β} + \frac{1}{2},]$

where, as we already specified, $α = ‖ M_{1} ‖, β = ‖ M_{2} ‖, a = ‖ M_{3} ‖, b = ‖ M_{4} ‖ .$

References

[1] Fujii, M., Remarks to accelerated iterative processes for numerical solution of equations, J. Sci. Hiroshima Univ. Ser. A-I, 27 (1963), 97–118.
[2] Lanncaster, P., Error for the Newton-Raphson method, Numerische Mathematik, 9, 1, (1966), 55–68.
[3] Ostrowski, A., The round of stability of iterations, Z.A.M.M. 47, (1967), 77–81
[4] ^†^†margin: $\leftarrow$ clickable Păvăloiu, I., Estimation des erreurs dans la résolution numérique des systèmes d’équations dans des espaces métriques. Seminar of Functional Analysis and Numerical Methods. Preprint Nr. 1, (1987), 121–129.
[5] ^†^†margin: $\leftarrow$ clickable Păvăloiu, I., Délimitations des erreurs dans la résolution numérique des systèmes d’équations, Research Seminars. Seminar on Mathematical analysis. Preprint Nr. 7, (1988), 167–178.
[6] Păvăloiu, I., ^†^†margin: $\leftarrow$ clickable Introducere în teoria aproximării soluţiilor ecuaţiilor. Ed. Dacia, Cluj-Napoca, (1976).
[7] Urabe, M., Convergence of numerical iteration in solution of equations. J. Sci. Hiroshima, Univ.Ser. A, 19 (1956), 479–489.
[8] Urabe, M., Error estimation in numerical solution of equations by iteration processes, J. Sci. Hiroshima Univ. Ser. A-I, 26, (1962), 77–91.
[9] Varga, R.S., Matrix Iterative Analysis, Englewood Cliffs H.J. Prentice Hall (1962).

Received 1.XII.1991

Institutul de Calcul

C.P. 68 Oficiul Poştal 1

3400 Cluj-Napoca

Romania

1992

On a competition issue

July 24, 2025

Professor Dimitrie D. Stancu at his 65ᵗʰ birthday

26 Aug at 10:46 am

Selections associated to McShane’s extension theorem for Lipschitz functions

May 9, 2023

(19)		$ρ_{1} ({\bar{x}}_{1}, x_{1}^{(n)})$	$\leq \frac{d_{1} p_{1}^{n}}{1 - p_{1}};$
	$ρ_{2} ({\bar{x}}_{2}, x_{2}^{(n)})$	$\leq \frac{d_{1} h_{1} p_{1}^{n}}{1 - p_{1}} .$

Error estimation in numerical solution of equations and systems of equations

Abstract

Authors

Keywords

PDF

Cite this paper as:

About this paper

Journal

Publisher Name

Article on the journal website

Print ISSN

Online ISSN

References

Paper (preprint) in HTML form

Error estimation in numerical solution of equations and systems of equations

1.

Theorem 1.1.

Theorem 1.2.

Theorem 1.3.

2.

Theorem 2.1.

Proof.

Theorem 2.2.

Proof.

3.

Theorem 3.1.

References

Related Posts

Error estimation in numerical solution
of equations and systems of equations