# Fileset

[icam_prc.pdf](https://mdr.nims.go.jp/filesets/1232c032-69c7-40ef-94d1-a3a55df303f7/download)

## Creator

[TODOROKI, Shin-ichi](https://orcid.org/0000-0003-3986-1900), INOUE, Satoru

## Rights



## Other metadata

[Multi-dimensional data management by virtual sample library written in object-oriented script language Ruby](https://mdr.nims.go.jp/datasets/88cdbc74-6ff9-4fd0-80c3-20341702ca00)

## Fulltext

Multi-dimensional data management by virtual sample library written in object-oriented script language RubyTransactions of the Materials Research Society of Japan29 [1] 293-296 (2004)Multi-dimensional data management by virtual sample library written inobject-oriented script languageRubyShin-ichi Todoroki∗ and Satoru InoueAdvanced Materials Laboratory, National Institute for Material Science1-1 Namiki, Tsukuba, Ibaraki 305-0044, JapanFax: 81-29-854-9060, e-mail: TODOROKI.Shin-ichi at nims.go.jpVirtual sample library (VSL) is developed for accessing multi-dimensional data acquired from severalmeasurements on a single combinatorial sample library. Management of such data is generally veryhard because each data is stored independently in various format. The data is structuralized andstandardized by VSL, which has a hierarchy structure whose top layer has the same geometry ofthe combinatorial sample library. Each data is stored into the VSL systematically according to theircoordinates in the library, the name of the measurements, and the dimension of the data. Thus, VSLcan provide any data for analysis and visualization by specifying their identifying information.An actual example is demonstrated for the case of one-dimensional combinatorial glass sample li-brary. Some tellurite glass libraries containing Er and F are annealed in a temperature-gradient fur-nace and the decay curves of 1.5µm-fluorescence of Er3+ are recorded along the libraries in order tofind annealing conditions for precipitation of Er-containing fluoride crystals. Their fluorescence spec-tra, fluorescence lifetime, state of precipitates as a function of annealing conditions (7-dimensionaldata) are plotted in two figures through VSL.Key words: multi-dimensional data, virtual sample library, object-oriented, informatics1. INTRODUCTIONCombinatorial technology brings about significant ad-vantages to our research activities provided we can af-ford to analyze large amount of data obtained. Whenwe are to make several kinds of measurements on onecombinatorially-integrated sample library, the obtaineddata become multi-dimensional and quite hard to dealwith by ordinary spreadsheet programs. For example,suppose that we have a 2-dimensional(2D) sample arrayon which three kinds of measurements are performed, andeach measurement gives us data in scalar, 2D, and 3D for-mat per pixel, respectively. This situation is illustrated inFig. 1. Then we obtain 3D, 4D and 5D data from thesemeasurements.Since the coordinates in the sample array are temporaryparameters for us, they should be converted to appropriatephysical quantities (e.g. composition, annealing temper-ature, etc...) according to the fabrication condition of thesample library. Moreover, when we need to see the cor-relation between these measurements, we have to mergeand re-compile the whole data so that the individual datameasured at the same pixel are related.These time-consuming editing jobs would be reducedif we could treat the whole data in one format, in whicheach data is linked with the corresponding coordinate andphysical quantity. This paper demonstrates that such amulti-dimensional data management is possible through“virtual sample library”(VSL), which is a data mediumused in software programs for analysis and/or visualiza-tion.???Position  (x, y)2.19Sample libraryMeasurement AMeasurement BMeasurement CfA(x, y)fB(x, y, λ)fC(x, y, λ, t)λλtFig. 1: Illustration showing a situation of treating multi-dimensional data obtained from several measurements ona combinatorially-integrated sample library (see text).2. WORKING HYPOTHESISFirst of all, let us discuss what kind of format is fea-sible for us to store the whole multi-dimensional dataand access any of them easily. Considering the coordi-nates in the library are common information among thedata obtained from each measurement, it is reasonable tostore the data under the individual coordinates hierarchi-cally. After specifying an arbitrary position in the library,we naturally notice what kind of measurements are per-formed there or what is the fabrication condition there.Finally, we can access data for the specified measurementor fabrication condition.One example of this hierarchy structure is visualized293294 Multi-dimensional data management by virtual sample library written in object-oriented script language Rubyas “pull-down menu” style in Fig. 2, where an 1D sam-ple array is used as a model. In the bottom of the win-dow, a menu is located showing the coordinates in the li-brary, from 40 to 60. After the item of “41” is selected, asub-menu appears showing what kinds of data are stored.Next, the item of “Fluorescence Spectrum” is chosen andanother sub-menu is opened to show this consists of twoitems. Finally, by choosing “Wavelength” item, a seriesof wavelength values are shown at “Value” box just abovethe main menu.In this way, such a hierarchy structure can store thewhole multi-dimensional data systematically and gives usan easy way to access individual data. Let us call thisstructure, “virtual sample library” (VSL), since the top ofthe hierarchy corresponds the shape of the sample library.The next thing we have to consider is how we con-struct this VSL and how we use it for data analysis andvisualization. Although there are many ways to realize it,we decided to make it as an original software written byobject-oriented script language, calledRuby[1, 2]. Thisis because object-oriented languages have abstract ex-pressive power essential for treating complicated data[3]and Rubywas developed for reducing the work of pro-grammers and allowing them concentrated on substantialmatters[1].3. METHODS AND PROCEDURESFor constructing VSL, a set of data is used which wascollected from 1D sample arrays of Er3+-F-co-doped tel-lurite glass[4]. The samples of about 30 cm in length wereprepared by sucking the glass melt into a pre-heated glasscapillary in a vertical temperature-gradient furnace andwere annealed in the furnace and/or quenched with dif-ferent conditions; (0) no annealing, (1) being annealedfor 5min, and (2) another successive annealing at a differ-ent position in the furnace for 5min. Since the annealingtemperature is a function of the height in the furnace, 1stand 2nd annealing temperatures for the sample (1) and (2)are calculated according to the temperature profile in thefurnace. Appearance of these glass samples are judged byhuman eye, as “transparent”, “white” or “opaque”. Time-resolved fluorescence intensity of Er3+ ions (1.5µm band,excitation: 977nm) were collected along the sample li-braries in every 1mm.The collected data of about 400MB in size are con-verted to obtain positional dependence of heat treat-ment temperatures, appearance, fluorescence spectra (CWFig. 2: An illustration showing hierarchy structure of vir-tual sample library (see text).Table 1: List of the 7-dimensional data used in this study(see text).. Some of these are plotted in Fig. 3 and 4.# Data Fig. 3 Fig. 41 Position,x/mm ✓ ✓2 1st annealing temp.,T1/◦C ✓ ✓3 2nd annealing temp.,T2/◦C ✓4 Appearance of glass segment ✓5 Fluorescence lifetime,τ /msec ✓Fluorescence spectra (CW component)6 Intensity (a.u.) ✓7 Wavelength,λ/nm ✓component), and lifetime of 1.533µm fluorescence. Sumof their dimension is 7, which is listed in Table 1, and thetotal size of data file is about 2.7MB in text format.The VSL is composed of three parts; (a) the definitionof the top of the hierarchy, (b) the procedure of loadingexperimental data and storing them under the top of thehierarchy, and (c) the routine for visualization.The first part corresponds to the definition of an arrayof data-containers whose configuration is the same as theactual sample libraries. In the second part, experimentaldata are extracted from the files and each of them is linkedwith the corresponding data-container according to theircoordinates in the sample library. After finishing theseprocedures, individual data can be accessed by tracingthe tree structure; for example, the series of wavelengthvalues in Fig. 2, which is pointed by the black arrow, isreferred by theRubysentence displayed in the “Code”box, which is located at the second line in the window.The word “vl ” at the beginning of the sentence meansa variable corresponding to a VSL and being defined inadvance by anotherRubysentence shown at the first lineof the window, where three data files are specified in theparenthesis to construct the VSL.Visualization is performed by calling an optionalgraphics library, Ruby/PGPLOT[5].4. RESULTSThe appearance of the glass segments as a function ofthe position in the library and/or the annealing temper-atures are plotted in Fig. 3. The glass segments namedAn(n = 0, 1, 2) and Bn(n = 0, 1) are located in the samepositions in the three sample libraries, respectively. Theannealing conditions for these segments are summarizedShin-ichi Todoroki et al. Transactions of the Materials Research Society of Japan29 [1] 293-296 (2004) 295Fig. 3: Annealing condition of the sample libraries andappearance of the annealed glass segments inside. (0)with no annealing treatment for reference. (1) with 1stannealing for 5min only. (2) with two successive heattreatments, each for 5min. Thick black line: completelytransparent segment, thin black line: white, and mediumgray line: opaque(see text). An and Bn indicate specificglass segments. These notations are also used in Fig. 4.by the following.(melt)↓A01st annealfor 5min−−−−−−−−−−−→T1 = 470 ↓A12nd annealfor 5min−−−−−−−−−−−→T2 = 550 ↓A2(1)(melt)↓B01st annealfor 5min−−−−−−−−−−−→T1 = 550 ↓B1(2)As the heat treatment proceeded, crystallization oc-curred and the portion of transparent segments in the li-brary decreased.The remainder of the multi-dimensional data are shownin Fig. 4. Three contour charts in gray scale located atthe top left are the positional dependence of fluorescencespectra, whose sectional views at positions of A and Bare shown at the bottom right in thick and broken line, re-spectively. While no positional dependence are observedfor the libraries of (0) and (1), the glass segments aroundA2 shows some spectral broadening. At the same time,an increase of fluorescence lifetime is observed at aroundA2 as shown at the top right of Fig. 4, where the blackpoints correspond to the library (2) and the gray dots tothe library (0) and (1).Consequently, the segment of A2 shows the highest flu-orescence lifetime. This is probably because Er3+ ionsmoved from the oxide glass matrix to the precipitatedfluorine-rich phase during the 2nd heat treatment[4]. Thisphenomenon does not occur if the 1st heat treatment of470◦C is omitted such as B1(see Eq. 2).Fig. 4: Fluorescence spectra of Er3+ and lifetime of1.533nm fluorescence plotted along the library or as afunction of the 1st annealing temperature,T1(see text).Excitation wavelength is 977nm.5. DISCUSSIONIn order to make the benefit of VSL clear, let us considerthe process of making Fig. 3 and 4 without using VSL.The experimental data used here are listed in Table 1.The annealing temperatures (#2 and #3 in Table 1) arefunctions of the position in the sample library (#1) andare calculated from the temperature gradient of the fur-nace and the annealing position of the glass capillary inthe furnace.In general, the temperature gradient of the furnace isnot linear. Therefore, the conversion from #1 to #2 and/or#3 is not so simple. For this reason, the intervals of tickmarks along thex-axis in Fig. 3 andT1-axis in Fig. 4 arenot fixed and the shape of the library (2) plotted in Fig. 3is not straight. This means that this conversion is neededwhenever we plot a graph in which both thex-axis andT1-axis are included.Without using VSL, Fig. 3 and 4 can be plotted onlyafter merging #1–#4 data and #1, #2, #5–#7 data (see Ta-ble 1), respectively, which are stored separately in differ-ent format. Moreover, the conversion ofx → T1, T2 isneeded every time we merge them. One of the advantagesof VSL is to avoid this merging process, which is neededonly once when the VSL is constructed.6. CONCLUSIONSConcept of virtual sample library (VSL) is proposed formanaging multi-dimensional data obtained from severalmeasurements on one combinatorially integrated samplelibrary. VSL is a data medium for storing data hierarchi-cally and accessing data intuitively. It reduces the load ofmerging experimental data which are separately stored indifferent formats. This is realized with the aid of abstractexpressive power of object-oriented language. Plotting 7-dimensional data in two figures is demonstrated.296 Multi-dimensional data management by virtual sample library written in object-oriented script language RubyReferences[1] “Ruby Home Page.” (http://www.ruby-lang.org ).[2] D. Thomas and A. Hunt,Programing Ruby, The Pragmatic Programmer’s Guide. Addison-Wesley, 2001. (ISBN0-201-71089-7).[3] C. Laird, “Open source in the lab,”IBM developerWorksTM, Oct. 2002.(http://www-106.ibm.com/developerworks/linux/library/l-oslab/or http://www-6.ibm.com/jp/developerworks/linux/030110/j l-oslab.html in Japanese).[4] S. Todoroki and S. Inoue, “Combinatorial fluorescence lifetime measuring system for developing Er-doped transparentglass ceramics,”Appl. Surface Sci., 2003. in print.[5] M. Tanaka, “Ruby/pgplot.”http://www.ir.isas.ac.jp/˜masa/ruby/pgplot/index.html .AcknowledgmentsThe first author express his sincere thanks to YukihiroMatsumoto (Ruby) and Masahiro Tanaka (NArray &Ruby/PGPLOT) for providing their excellent software forfree.(Recieved October 12, 2003; Accepted November 19, 2003)http://www.ruby-lang.org�http://www-106.ibm.com/developerworks/linux/library/l-oslab/�http://www-6.ibm.com/jp/developerworks/linux/030110/j_l-oslab.html�