Modeling languages and Condor: metacomputing for optimization

A generic framework is postulated for utilizing the computational resources provided by a metacomputer to concurrently solve a large number of optimization problems generated by a modeling language. An example of the framework using the Condor resource manager and the AMPL and GAMS modeling languages is provided. A mixed integer programming formulation of a feature selection problem from machine learning is used to test the mechanism developed. Due to this application’s computational requirements, the ability to perform optimizations in parallel is necessary in order to obtain results within a reasonable amount of time. Details about the simple and easy to use tool and implementation are presented so that other modelers with applications generating many independent mathematical programs can take advantage of it to significantly reduce solution times. Received: October 28, 1998 / Accepted: December 01, 1999?Published online June 8, 2000     

Feasible descent algorithms for mixed complementarity problems

In this paper we consider a general algorithmic framework for solving nonlinear mixed complementarity problems. The main features of this framework are: (a) it is well-defined for an arbitrary mixed complementarity problem, (b) it generates only feasible iterates, (c) it has a strong global convergence theory, and (d) it is locally fast convergent under standard regularity assumptions. This framework is applied to the PATH solver in order to show viability of the approach. Numerical results for an appropriate modification of the PATH solver indicate that this framework leads to substantial computational improvements. Received April 9, 1998 / Revised version received November 23, 1998?Published online March 16, 1999     

Radiative stabilization of trimethylsilyl adduct ions

Ketones and phenol react with trimethylsilyl ions to form adduct ions by radiatively or collisionally stabilized addition reactions, in contrast to aliphatic alcohols and ethers, which react with trimethylsilyl ions to form adduct ions by a rapid two-step process. Secondorder rate constants for the addition of trimethylsilyl ions to acetone were independent of pressure from 3×10^?7 to 50×10^?7 tort at room temperature; consequently, the adduct ions, [M+73]⁺, are formed primarily by radiatively stabilized addition in these ion cyclotron resonance experiments.     

Studies on fluorescein-III The acid strengths of fluorescein as shown by potentiometric titration

Potentiometric back-titration of yellow solid fluorescein (H(2)Fl) and of red solid fluorescein in alkali with acid yielded titration curves that were practically identical in shape and position. The end-points at pH 8.5, 5.40 and 3.3 corresponded, respectively, to titration of the excess of standard alkali, and the successive protonations Fl(2-) + H(+) = HFl(-) and HFl(-) + H(+) = H(2)Fl. The pH at the mid-point of the first protonation yielded a value of 6.36 for pK(HFl) (ionic strength 0.10). Because of precipitation of yellow fluorescein during the second protonation step, a value for pK(H(2)Fl) could not be obtained. The total concentration of fluorescein at the first appearance of the precipitate fell on the curve for the solubility of yellow fluorescein as a function of pH. The titrations and the pK values found for the three acid groups of protonated fluorescein (H(3)Fl(+)) have been interpreted on the basis that in water fluorescein exists in only one structural form the yellow zwitterion. Similar back-titrations of alkalinized solutions of yellow or red fluorescein in 50% aqueous ethanol showed that in this medium fluorescein is present in only one form, presumably the quinonoid structure, with much weaker apparent acid functions, pK'(1) = 6.38 and PK'(2) = 7.16 (ionic strength 0.10).     

Stereochemical effects in the mass spectra of 2-hydroxy-, 5-hydroxy- and 2,5-dihydroxyprotoadamantanes

Electron ionization and methane and isobutane chemical ionization mass spectra have been obtained for the four configurationally isomeric 2,5-protoadamantanediols and the corresponding model monoalcohols. Although the electron impact (both 70 and 15 eV) and methane chemical ionization mass spectra of the 2,5-protoadamantanediols are distinct, the variations are not large and in general they are difficult to rationalize in terms of specific structural features. In contrast, the isobutane chemical ionization mass spectra of these diols show significant differences which can easily be correlated with sterochemical orientations of the substituents.     

A new spot test for aliphatic aldehydes and ketones

A new spot test for aliphatic aldehydes and ketones has been developed utilizing free-radical formation. Aliphatic aldehydes are detected in the range of 0.2–200 μg while aliphatic ketones are detected from 100–2000 μg. Benzaldehyde was the only other compound tested that gave a positive result.     

Investigation of the mechanism of n-butane oxidation on vanadium phosphorus oxide catalysts: evidence from isotopic labeling studies

The selective oxidation of n-butane to maleic acid catalyzed by vanadium phosphates (VPO) is one of the most complex partial oxidation reactions used in industry today. Numerous reaction mechanisms have been proposed in the literature, many of which have butenes, butadiene, and furan as reaction intermediates. We have developed an experimental protocol to study the mechanism of this reaction in which (13)C-isotopically labeled n-butane is flowed over a catalyst bed and the reaction products are analyzed using (13)C NMR spectroscopy. This protocol approximates the conditions found in an industrial reactor without requiring an exorbitant amount of isotopically labeled material. When [1,4-(13)C]n-butane reacted on VPO catalysts to produce maleic acid and butadiene, the isotopic labels were observed in both the 1,4 and 2,3 positions of butadiene and maleic acid. The ratio of label scrambling was typically 1:20 for the 2,3:1,4 positions in maleic acid. For butadiene, the ratio of label scrambling was consistently much higher, at 2:3 for the 2,3:1,4 positions. Because of the discrepancy in the amount of label scrambling between maleic acid and butadiene, butadiene is unlikely to be the primary reaction intermediate for the conversion of n-butane to maleic anhydride under typical industrial conditions. Ethylene was always observed as a side product for n-butane oxidation on VPO catalysts. Fully (13)C-labeled butane produced about 5-13 times as much isotopically labeled ethylene as did [1,4-(13)C]butane, indicating that ethylene was produced mainly from the two methylene carbons of n-butane. When the reaction was run under conditions which minimize total oxidation products such as CO and CO(2), the amounts of ethylene and carbon oxides produced from fully (13)C-labeled butane were almost equal. This strongly suggests that the total oxidation of n-butane on VPO catalysts involves the oxidation and abstraction of the two methyl groups of n-butane, and the two methylene groups of n-butane form ethylene. An organometallic mechanism is proposed to explain these results.     

Kinetic modeling study of the laser-induced plasma plume of cyclotrimethylenetrinitramine (RDX)

A kinetic model of a laser-induced breakdown spectroscopy (LIBS) plume of cyclotrimethylenetrinitramine (RDX) was developed for the analysis of processes responsible for the LIBS signature of explosives. Air and argon were considered as buffer gases. The model includes a set of processes involving ion chemistry, as well as excitation, ionization, and other processes affecting neutral and ion species. Modeling results show that the overall reaction process can be considered a two-stage process. The first stage corresponds to a fast approach to a quasi-stationary state, while the second stage corresponds to the change of quasi-stationary species concentrations due to the change in temperature. As a result of the two-stage process, the initial mechanism of explosive decomposition is not important in determining its signature in the LIBS measurement time window (1–30 μs). The main processes responsible for generation of excited states for the LIBS emission are electron-excitation impact processes. A mechanism for the appearance of a double peak of the C₂ species concentration in the RDX plasma plume was suggested. Double-peak behavior of the C₂ species was previously experimentally observed during laser ablation of graphite.     

Energy and substituent effects on gaseous ionic decompositions: Electron and charge exchange ionization of benzophenones and benzils

The correlation with substituent constants reported previously for [YØCO]⁺/[ØCO]⁺ ratios in the electron ionization mass spectra of substituted benzophenones and acetophenones has also been observed in the electron ionization spectra of substituted benzils. The [YØCO]⁺/[ØCO]⁺ ratio for the substituted benzils varied with energy of the ionizing electrons according to predictions from a simple kinetic and thermochemical analysis. [YØCO]⁺/[ØCO]⁺ ratios in the charge exchange spectra of benzophenones obtained with Xe, Kr, CO, N₂ and Ar gave good correlations with sub-stituent constants in agreement with the same analysis. Good correlations were also noted for [YØCO]⁺/[ØCO]⁺ ratios with substituent constants for [M]⁺ ions of the benzophenones of the same excess energy (5.5 eV). [YØCO]⁺/[ØCO]⁺ ratios for benzils obtained by charge exchange with [CO]⁺ also showed good correlations with substituent constant. It is suggested that [Ø]⁺ and [YØ]⁺ ions from the benzophenones may be formed primarily by one step decompositions of the molecular ions, but that the [Ø]⁺ and [YØ]⁺ ions from the benzils are formed primarily by decomposition of [ØCO]⁺ and [YØCO]⁺ ions.