Control de congestión adaptativo en redes Infiniband

El uso de recursos compartidos en las redes de interconexión de alta performance puede provocar situaciones de congestión de mensajes que degradan notablemente las prestaciones, aumentando la latencia de trasporte y disminuyendo la utilización de la red. Hasta el momento las técnicas que intentan solucionar este problema utilizan la regulación de la inyección de mensajes. Esta limitación de la inyección traslada la contención de mensajes desde los conmutadores hacia los nodos fuente, incrementando el valor de la latencia promedio global, pudiendo alcanzar valores muy elevados. En este trabajo, proponemos una técnica de control de congestión para redes InfiniBand basada en un mecanismo de encaminamiento adaptativo que distribuye el volumen de comunicaciones entre diversas trayectorias alternativas quitando carga de la zona de congestión, lo que permite eliminarla. La experimentación realizada muestra la mejora obtenida en latencia y throughput, respecto al mecanismo de control de congestión original de InfiniBand basado en la regulación de la inyección. El mecanismo propuesto es totalmente compatible y no requiere que se modifique ningún aspecto de la especificación, debido a que se utilizan componentes de gestión definidos en el estándar InfiniBand.

Inputs to the hormonal control of egg development in Rhodnius prolixus

This paper re-examines existing data on the environmental inputs governing egg production in Rhodnius prolixus. Feeding has a direct effect on egg production such that the product of the unfed weight of the female times the weight of the blood meal is a good predictor of the number of eggs produced. Mating modifies this input, so that mated females produce more eggs. Egg production is governed by the corpus allatum, and indirect evidence suggests that the number of eggs producted by a female is a function of the length of time that juvenile hormone is secreted by the corpus allatum. The input which determines the times at which the corpus allatum is switched off originates in the stretch induced by the amount of the meal remaining in the crop, modified by the matedness status of the female. The precise nature of the sensors detecting stretch is not yet clear, but the integrity of the dorsal aorta is essential to the transmission of the information. These data are related to the survival strategy for Rhodnius.

A review of achievements of the national schistosomiasis control program in middle and upper Egypt areas

In Egypt the "national schistosomiasis control program" was formulated to control transmission by reduction of prevalence and intensity of current infections, and thereby achieve an acceptable level of schistosomiasis disease control. The program was implemented foremost in Middle Egypt (1977) and Upper Egypt (1980), collectively extending 800 km alongside of the River Nile and accommodate about 10.5 million people. Schistosoma haematobium has been essentially the prevailing species infection in both areas. The strategy of control entailed both area-wide mollusciciding with niclosamide, and selective population chemotherapy with metrifonate. Evaluation in 1986 showed that prevalence dropped from pre-control 29.4% in Middle Egypt and 26.3% in Upper Egypt to 6% and 7.8% respectively, together with a remarkable drop of infections among children. Also mean intensity attained low levels consistent of low grade infections. It is evident therefore that in these areas where an enhancement of schistosomiasis infections had been anticipated the employment of the twofold strategy effected a state of low-prevalence/low-intensity signifying a lowered reservoir of infection and a substantial interference with the potentials of transmission.

The Venezuelan experience in the control of Schistosomiasis mansoni

Schistosomiasis mansoni endemic zone of Venezuela is located in the valleys of the north central mountain region, with an extension of 15,000 km2 and inhabited by 5.1 million persons. The disease was discovered in 1906, but an organized Control Program was not established until 1943. Its basic activity has been the control of the snail vector, but prevention of man-water contact, prevention of snail infection, treatment of infected people and sanitary instruction, have also been carried out. Prevalence has diminished from 14.7% (1943-60) to 0.9% (1981-84). At present few active foci still persist, but a low transmission rate and low morbidity makes it difficult to know the exact number of infected people, which has been estimulated to be about 50,000.

Connexin-36 contributes to control function of insulin-producing cells.

Connexin-36 (Cx36) is a gap junction protein expressed by the insulin-producing beta-cells. We investigated the contribution of this protein in normal beta-cell function by using a viral gene transfer approach to alter Cx36 content in the insulin-producing line of INS-1E cells and rat pancreatic islets. Transcripts for Cx43, Cx45, and Cx36 were detected by reverse transcriptase-PCR in freshly isolated pancreatic islets, whereas only a transcript for Cx36 was detected in INS-1E cells. After infection with a sense viral vector, which induced de novo Cx36 expression in the Cx-defective HeLa cells we used to control the transgene expression, Western blot, immunofluorescence, and freeze-fracture analysis showed a large increase of Cx36 within INS-1E cell membranes. In contrast, after infection with an antisense vector, Cx36 content was decreased by 80%. Glucose-induced insulin release and insulin content were decreased, whether infected INS-1E cells expressed Cx36 levels that were largely higher or lower than those observed in wild-type control cells. In both cases, basal insulin secretion was unaffected. Comparable observations on basal secretion and insulin content were made in freshly isolated rat pancreatic islets. The data indicate that large changes in Cx36 alter insulin content and, at least in INS-1E cells, also affect glucose-induced insulin release.

Optimal feedback control rules sensitive to controlled endogenous risk-aversion

The objective of this paper is to correct and improve the results obtained by Van der Ploeg (1984a, 1984b) and utilized in the theoretical literature related to feedback stochastic optimal control sensitive to constant exogenous risk-aversion (see, Jacobson, 1973, Karp, 1987 and Whittle, 1981, 1989, 1990, among others) or to the classic context of risk-neutral decision-makers (see, Chow, 1973, 1976a, 1976b, 1977, 1978, 1981, 1993). More realistic and attractive, this new approach is placed in the context of a time-varying endogenous risk-aversion which is under the control of the decision-maker. It has strong qualitative implications on the agent's optimal policy during the entire planning horizon.