GemIdent – The Gem Identification Domain Expert

GemIdent is a gem identification software and data base used in gemology.It should not be confused with a similarly named image recognition program. GemIdent is derived off MinIdent.

GemIdent represents information about gems in a statistical manner and identifies unknown gems with a fuzzy-like matching system.

People. GemIdent was developed between 1989 and 1990 by David Pierre Leibovitz and Heideh Omoumi.

Related Publications:

Omoumi, H. (1990) GemIdent: a data base for gems and some applications of the electron microprobe in gem characterisation. MSc. Thesis, University of Alberta.

Heideh Omoumi

Heideh Omoumi (Heida Mani) is Director of Markets & Industry at Vale. In 1990, she received a Masters of Science (MSc) degree in Mineralogy at the University of Alberta where she collaborated with David Pierre Leibovitz.

Collaborations:

Joint Publications:

Smith, D. G. W., Omoumi, H., & Leibovitz, D. P. (1989) The MinIdent database: some recent development. 28th International Geological Congress. Abstracts 3: 138-139 [doi: 10.13140/RG.2.1.5092.7842]

Links:

The MinIdent database: some recent development

Smith, D. G. W., Omoumi, H., & Leibovitz, D. P. (1989) The MinIdent database: some recent development. 28th International Geological Congress. Extended Abstract, 3: 138-139. [doi: 10.13140/RG.2.1.5092.7842] (pdf)

MinIdent-PCAbstract: The MinIdent database and the software for mineral identification (Smith &: Leibovitz 1984, 1986; Smith, 1986) have been successfully ported to a PC from the Amdahl mainframe on which they were developed. The “compiled” data base (used in mineral identification) plus necessary management programs can be accommodated on a 30 Mbyte hard disc. Developments presently being undertaken, will further reduce these storage requirements.

Since the publication by Smith &: Leibovitz (1986). many new data have been added and information now exists for some 4.200 mineral varieties. species. series. groups etc. Literature and data-entry errors are being identified using tests for self-consistency. and progressively eliminated.

In particular. that part of the database dealing with un-named minerals has been greatly expanded, and reorganised by year of first description. Data for un-named minerals are presently scattered throughout 100 years of earth science literature and range from vague descriptions of hand-specimen properties to complete modern analyses. The lists compiled by Hey (1962. 1963) were made the starting point. and then a wide range of journal and other literature sources used to obtain additional information and to bring the list up to date. Only those minerals for which numerical data are available have been included. Presently. nearly 600 un-named minerals appearing in the literature have been included. Once complete, this subset of the database will provide a unique resource and will allow users attempting to identify unknowns to compare their data with those for all previously described minerals and not only with those species that have received names. The use of the mineral identification software permits a numerical estimate to be obtained of the similarity between an unknown and previously described species. The immediate availability of a compilation of literature data for the latter provides a convenient indication of what other parameters might be obtained for a more unambiguous identification. The possibility also exists of adding a further category of data – for phases which have been obtained as the products of experimental work but which have not so far been found occurring naturally.

Another addition to MinIdent is a substantial list of discredited mineral names and synonyms. Entries are also included for minerals which although of dubious authenticity have not been unequivocally discredited and therefore remain in the database. At present there are about 1500 entries in the list, each of which includes a brief explanatory comment, the synonym (where applicable) and source reference(s). The scheme for naming rare earth-bearing minerals which was recently approved by the lMA has been fully implemented and, as far as possible, data associated with each rare earth variant of that species, appropriately assigned. However, the paucity of complete and reliable information on the concentrations of individual rare earths, continues to pose a problem. The re cent IMA decision to return to the original spelling of many non-English names has been implemented, although present software constraints preclude the inclusion of diacritical marks in such names.

Classification of minerals has been possible since the inception of ~dent. This facility has now been expanded so that the following divisions can be recognised where appropriate: variety, sub-species, species, series, sub-group, group, super-group, family, class and type. The top level “type” (e.g., silicates, oxides, sulphides, etc.) has been chosen to avoid ambiguity or overlap with other levels of classification previously used in the literature. Much progress has been achieved particularly with respect to rock-forming and more common minerals. For example, the full IMA amphibole classification scheme has been implemented (Goble &: Smith, 1988), and that for pyroxenes is presently being undertaken. However, much remains to be done and progress is hindered by the absence of a gene rally accepted and definitive classification scheme for minerals.

Other changes include the up-dating of the JCPDS PDF number and the entry or re-entry of the d-values for the five most intense lines from original literature sources. The algorithm that was specially developed for MinIdent to identify minerals on a very limited number of d-values has proved to work very well on pure phases. It is not intended for use with mixtures of phases.

Finally, a sub-set facility has been implemented which allows any group of minerals of interest to be selected from the database, and subsequently only these to be considered in the MATCH/IDENTIFY procedures. Such subsets are entirely flexible and could include categories such as “silicates”. “meteorite minerals” – or sets of minerals for instructional purposes. The use of subsets greatly reduces computational time for identification. It could prove extremely useful for the automation of mineral identification in combination with the analytical and image analysis capabilities of modern microbeam instruments.

References:

Goble, R.J. &: Smith, D.G.W. (1988): MinIdent: An application to the identification and
classification of amphiboles. Mineral. Petrol. v.38, p.213-227.

Hey. M.H. (1962): Chemical Index of Minerals. Brit. Mus. Nat. Hist. (London), 728 pp.

Hey, M.H. (1963): Appendix to Chemical Index of Minerals. Brit. Mus. Nat. Hist. (London) 135pp.

Smith, D.G.W. (1986): Automation of mineral identification from electron microprobe analyses. In: “Microbeam Analysis – 1986” (A. D. Romig &: W. F. Chambers. Eds.) San Francisco Press, San Francisco, U.S.A.

Smith, D.G.W. &: Leibovitz, D.P. (1986): MinIdent: A data base for minerals and a computer pro gram for their identification.

Smith D.G.W. &: Leibovitz, D.P. (1984): A computer based system for identification of minerals on the basis of composition and other properties. 27th Internat. Geol. Congr .• Moscow (1984) Abstracts v.5. p.169.

Links:

MinIdent – A Data Base for Minerals and a FORTRAN 77 Program for Their Identification – A Reference Manual

Smith, D. G. W., & Leibovitz, D. P. (1987) MinIdent – A Data Base for Minerals and a FORTRAN 77 Program for Their Identification – A Reference Manual, xiv-127. Department of Geology, University of Alberta, Edmonton, Alberta, Canada. [doi:10.13140/RG.2.1.1815.9841] (pdf)

Smith & Leibovitz (1987) MinIdent Reference Manual- A Data Base for Minerals and a FORTRAN 77 Program for Their IdentificationAbstract: Minldent is an interactive mineral identification and mineral data base management program written in FORTRAN 77. The data base contains compositional, optical and other parameters describing more than 3700 minerals. The data base management aspect of the program will not be used by the general user. It contains facilities to modify the data base through additions, deletions, etc. The normal usage consists of the following steps:

  1. entering data for a mineral to be identified (the unknown) or entering search criteria.
  2. identifying the mineral, or matching minerals meeting the search criteria.
  3. displaying data for specified, identified or matched minerals.

MinIdent-PCThis manual assumes the reader is already familiar to some extent with Minldent. It contains all the terms that may be explained via the ? / HELP / EXPLAIN commands.

Links:

MinIdent: A data base for minerals and a computer program for their identification

Smith, D. G. W., & Leibovitz, D. P. (1986) MinIdent: A data base for minerals and a computer program for their identification. Canadian Mineralogist. 24(4): 695-708. (pdf)

MinIdent-PCAbstract: MinIdent is a program for interactive mineral identification and mineral data base management, rewritten in FORTRAN 77, Data have been stored for about 4000 mineral groups, species and varieties. These data comprise: composition, optical properties, symmetry, cell dimensions, density, Vickers and Mohs hardness, d-values and relative intensities for the strongest five X-ray powder-diffraction lines, the PDF number, polymorphs if any, occurrences, localities, year first described and sources of the data. As yet, however, not all minerals have data stored for all properties. The program will generate a list of minerals whose properties, lie within input ranges for an unidentified mineral or display and rank twenty possible identities for an unknown. It can also be used to tabulate chosen properties of matched minerals or to tabulate minerals (in the data base) that have certain specified properties. Tests using data for known species to simulate unidentified minerals show high reliability, given accurate input information, and surprising success even with qualitative input data. The MinIdent software currently uses about 400 kbytes of memory, and the data base used in mineral identification uses a further l0 Mbytes. Running time for a typical identification procedure ranges from about 0.05 to 3.0 seconds of CPU time on the AMDAHL 580/FF mainframe computer, on which the program has been developed. The cycle time of this computer is about 23 ns.

Current MinIdent-Win software available at www.micronex.ca.

Links:

 

MinIdent – A Data Base for Minerals and a Computer Program for Their Identification

Smith, D. G. W., & Leibovitz, D. P. (1986) MinIdent – A Data Base for Minerals and a Computer Program for Their Identification. Program with Abstracts GAC, MAC, CGU-AGC, AMC, UCG: Joint Annual Meeting, May 19-21, 1986, Carleton University, Ottawa. Abstracts 11: 129. [doi10.13140/RG.2.1.1667.5048] (pdf)

MinIdent-PCAbstract: MinIdent is an interactive mineral identification and mineral data base management program, now rewritten in FORTRAN 77. Data have been stored for about 4000 mineral groups, species and varieties. These data include composition, optical properties in transmitted and reflected light, symmetry, unit cell dimensions, densities, Vickers and Mohs hardness, d-values and relative intensities of the 5 strongest X-ray powder-diffraction lines, JCPDS numbers, any polymorphs, occurrences, localities, year first described and sources of the data. However, not all minerals yet have data stored for all these fields.

The program can be used to generate a list of minerals having properties within within the ranges input for an unidentified mineral or can be made to display and rank the twenty most likely identities for an unknown. The program can also be used to tabulate chosen properties of matched minerals, or to tabulate minerals in the data base that have certain specified properties. Alternatively, all analytical and other data stored for a particular mineral can be displayed.

Tests using data for known minerals to simulate unknowns indicate a high degree of reliability given accurate input information, and a surprising success rate even when input data are qualitative in character.

The MinIdent identification and data base management software uses about 400 kbytes of memory and the data base used in mineral identification currently uses less than 4 Mbytes. Running times for typical identification procedures range between about 0.5 and 3.0 seconds of CPU time on the AMDAHL 580/FF mainframe computer on which the program has been developed. The cycle time of this computer is about 23 ns. MinIdent can be accessed globally via data communications networks such as DATAPAC, TELENET and TYMNET.

Application of the MinIdent data base and software are envisaged wherever earth scientists are faced with the task of mineral identification. Such areas of specialization include petrology (igneous, metamorphic and sedimentary), economic geology (ore mineralogy, mineral exploration and mineral beneficiation), geochemistry, meteorites and crystallography.

Links:

MinIdent User’s Manual. A FORTRAN 77 program for mineral identification

Smith, D. G. W., & Leibovitz, D. P. (1986) MinIdent User’s Manual. A FORTRAN 77 program for mineral identification, pp. vii-88. Computing Services, University of Alberta, Edmonton, Alberta, Canada. [doi: 10.13140/RG.2.1.2733.4882] (pdf)

Smith & Leibovitz (1986) MinIdent User's Manual- A FORTRAN 77 Program for Mineral IdentificationAbstract: MinIdent is a mineral identification software used in mineralogy. The original Command-line interface (CLI) program was written in FORTRAN and ran on a mainframe computer. It was later ported to a PC. This manual forms the user’s guide (UG) for the original version. The current version of the MinIdent-Win software has a graphical user interface (GUI) and is available at www.micronex.ca.

Links: