A general framework for updating belief distributions Bissiri, P. G.; Holmes, C. C.; Walker, S. G.
Journal of the Royal Statistical Society. Series B, Statistical methodology,
November 2016, Letnik:
78, Številka:
5
Journal Article
Recenzirano
Odprti dostop
We propose a framework for general Bayesian inference. We argue that a valid update of a prior belief distribution to a posterior can be made for parameters which are connected to observations ...through a loss function rather than the traditional likelihood function, which is recovered as a special case. Modern application areas make it increasingly challenging for Bayesians to attempt to model the true data-generating mechanism. For instance, when the object of interest is low dimensional, such as a mean or median, it is cumbersome to have to achieve this via a complete model for the whole data distribution. More importantly, there are settings where the parameter of interest does not directly index a family of density functions and thus the Bayesian approach to learning about such parameters is currently regarded as problematic. Our framework uses loss functions to connect information in the data to functionals of interest. The updating of beliefs then follows from a decision theoretic approach involving cumulative loss functions. Importantly, the procedure coincides with Bayesian updating when a true likelihood is known yet provides coherent subjective inference in much more general settings. Connections to other inference frameworks are highlighted.
Phloem sieve element (SE) occlusion has been hypothesized for decades to be a mechanism of resistance against phloem sap-feeding insects. Few studies have tested this hypothesis although it is likely ...a widespread phenomenon. This review focuses on SE occlusion by callose and P-proteins. Both are reversible, which would allow the plant to defend itself against phloem sap-feeders when SEs are penetrated and resume normal function when the insects give up and withdraw their stylets. Callose (β-1,3 glucans with some β-1,6 branches) serves many roles in plant physiology in many different tissues, each being under the control of different callose synthase genes; only callose deposited in SE sieve pores is relevant to SE occlusion. The amount of callose in sieve pores (and consequently how much it impedes sap flow) is determined by the balance in activity between callose synthase and β-1,3 glucanase. Sieve pore callose deposition has been shown to provide resistance to some phloem sap-feeders in a few studies, and in one, the difference in resistance between a susceptible and resistant rice variety was due to the ability or inability of the insect to upregulate the plants' β-1,3 glucanase that degrades the callose deposition. P-proteins occur only in dicotyledons and include a variety of proteins, not all of which are involved in SE occlusion. In some plants, P-proteins form distinct bodies in mature functional SEs. In papilionid legumes, these discrete bodies, called forisomes can expand and contract. In their expanded state, they effectively plug SEs and stop the flow of sap while in their contracted state, they provide negligible resistance to sap flow. Expansion of forisomes is triggered by an influx of Ca2+ into the SE. Penetration of a legume (Vicia faba) SE by a generalist aphid not adapted to legumes triggers forisome expansion which occludes the SE and prevents the aphid from ingesting sap. In contrast, a legume specialist aphid, Acyrthosiphon pisum, does not trigger forisome expansion and readily ingests sap from V. faba. P-protein bodies in SEs of non-legumes do not appear to be involved in SE occlusion. In most dicotyledons, P-proteins do not form discrete bodies, but rather occur as filamentous aggregations adhering to the parietal margins of the SE and in response to damage, are released into the lumen where they are carried by the flow of sap to the downstream sieve plate where they back up and clog the sieve pores. Their effectiveness at actually stopping the flow of sap is controversial. In one study, they seemed to provide little resistance to the flow of sap while in other studies, they provided considerable resistance. In response to injury in melon, they completely stop the flow of sap, and in an aphid-resistant melon, penetration of SEs by the melon aphid, Aphis gossypii, triggers P-protein occlusion which prevents the aphids from ingesting sap. The first P-protein described, PP1, occurs only in the genus Cucurbita, and although it has been often cited to function as a SE occlusion protein, experimental evidence suggests it does not play a significant role in SE occlusion. The most common strategy for phloem sap-feeders to mitigate P-protein occlusion seems to be avoid triggering it. A widely cited in vitro study suggested that aphid saliva can reverse P-protein occlusion, but a subsequent study demonstrated that saliva was ineffective at reversing P-protein occlusion in vivo. Lastly, SE callose deposition in wheat triggered by Russian wheat aphid has been hypothesized to create an artificial sink that benefits the aphid, but additional studies are needed to test that hypothesis.
The exquisite soft-tissue contrast of magnetic resonance imaging (MRI) has meant that the technique is having an increasing role in contouring the gross tumor volume (GTV) and organs at risk (OAR) in ...radiation therapy treatment planning systems (TPS). MRI-planning scans from diagnostic MRI scanners are currently incorporated into the planning process by being registered to CT data. The soft-tissue data from the MRI provides target outline guidance and the CT provides a solid geometric and electron density map for accurate dose calculation on the TPS computer. There is increasing interest in MRI machine placement in radiotherapy clinics as an adjunct to CT simulators. Most vendors now offer 70 cm bores with flat couch inserts and specialised RF coil designs. We would refer to these devices as MR-simulators. There is also research into the future application of MR-simulators independent of CT and as in-room image-guidance devices. It is within the background of this increased interest in the utility of MRI in radiotherapy treatment planning that this paper is couched. The paper outlines publications that deal with standard MRI sequences used in current clinical practice. It then discusses the potential for using processed functional diffusion maps (fDM) derived from diffusion weighted image sequences in tracking tumor activity and tumor recurrence. Next, this paper reviews publications that describe the use of MRI in patient-management applications that may, in turn, be relevant to radiotherapy treatment planning. The review briefly discusses the concepts behind functional techniques such as dynamic contrast enhanced (DCE), diffusion-weighted (DW) MRI sequences and magnetic resonance spectroscopic imaging (MRSI). Significant applications of MR are discussed in terms of the following treatment sites: brain, head and neck, breast, lung, prostate and cervix. While not yet routine, the use of apparent diffusion coefficient (ADC) map analysis indicates an exciting future application for functional MRI. Although DW-MRI has not yet been routinely used in boost adaptive techniques, it is being assessed in cohort studies for sub-volume boosting in prostate tumors.
Successive locations of individual large earthquakes (M
> 5.5) over years to centuries can be difficult to explain with simple Coulomb stress transfer (CST) because it is common for seismicity to ...circumvent nearest-neighbour along-strike faults where coseismic CST is greatest. We demonstrate that Coulomb pre-stress (the cumulative CST from multiple earthquakes and interseismic loading on non-planar faults) may explain this, evidenced by study of a 667-year historical record of earthquakes in central Italy. Heterogeneity in Coulomb pre-stresses across the fault system is >±50 bars, whereas coseismic CST is <±2 bars, so the latter will rarely overwhelm the former, explaining why historical earthquakes rarely rupture nearest neighbor faults. However, earthquakes do tend to occur where the cumulative coseismic and interseismic CST is positive, although there are notable examples where earthquake propagate across negatively stressed portions of faults. Hence Coulomb pre-stress calculated for non-planar faults is an ignored yet vital factor for earthquake triggering.
Ammonia-oxidizing archaea are ubiquitous in marine and terrestrial environments and now thought to be significant contributors to carbon and nitrogen cycling. The isolation of Candidatus ..."Nitrosopumilus maritimus" strain SCM1 provided the opportunity for linking its chemolithotrophic physiology with a genomic inventory of the globally distributed archaea. Here we report the 1,645,259-bp closed genome of strain SCM1, revealing highly copper-dependent systems for ammonia oxidation and electron transport that are distinctly different from known ammonia-oxidizing bacteria. Consistent with in situ isotopic studies of marine archaea, the genome sequence indicates N. maritimus grows autotrophically using a variant of the 3-hydroxypropionate/4-hydroxybutryrate pathway for carbon assimilation, while maintaining limited capacity for assimilation of organic carbon. This unique instance of archaeal biosynthesis of the osmoprotectant ectoine and an unprecedented enrichment of multicopper oxidases, thioredoxin-like proteins, and transcriptional regulators points to an organism responsive to environmental cues and adapted to handling reactive copper and nitrogen species that likely derive from its distinctive biochemistry. The conservation of N. maritimus gene content and organization within marine metagenomes indicates that the unique physiology of these specialized oligophiles may play a significant role in the biogeochemical cycles of carbon and nitrogen.
The association of genetic lesions detected by fluorescence in situ hybridization (FISH) with survival was analyzed in 1069 patients with newly presenting myeloma treated in the Medical Research ...Council Myeloma IX trial, with the aim of identifying patients associated with the worst prognosis. A comprehensive FISH panel was performed, and the lesions associated with short progression-free survival and overall survival (OS) in multivariate analysis were +1q21, del(17p13) and an adverse immunoglobulin heavy chain gene (IGH) translocation group incorporating t(4;14), t(14;16) and t(14;20). These lesions frequently co-segregated, and there was an association between the accumulation of these adverse FISH lesions and a progressive impairment of survival. This observation was used to define a series of risk groups based on number of adverse lesions. Taking this approach, we defined a favorable risk group by the absence of adverse genetic lesions, an intermediate group with one adverse lesion and a high-risk group defined by the co-segregation of >1 adverse lesion. This genetic grouping was independent of the International Staging System (ISS) and so was integrated with the ISS to identify an ultra-high-risk group defined by ISS II or III and >1 adverse lesion. This group constituted 13.8% of patients and was associated with a median OS of 19.4 months.
Flood modeling at the regional to global scale is a key requirement for equitable emergency and land management. Coupled hydrological‐hydraulic models are at the core of flood forecasting and risk ...assessment models. Nevertheless, each model is subject to uncertainties from different sources (e.g., model structure, parameters, and inputs). Understanding how uncertainties propagate through the modeling cascade is essential to invest in data collection, increase flood modeling accuracy, and comprehensively communicate modeling results to end users. This study used a numerical experiment to quantify the propagation of errors when coupling hydrological and hydraulic models for multiyear flood event modeling in a large basin, with large morphological and hydrological variability. A coupled modeling chain consisting of the hydrological model Hydrologiska Byråns Vattenbalansavdelning and the hydraulic model LISFLOOD‐FP was used for the prediction of floodplain inundation in the Murray Darling Basin (Australia), from 2006 to 2012. The impacts of discrepancies between simulated and measured flow hydrographs on the predicted inundation patterns were analyzed by moving from small upstream catchments to large lowland catchments. The numerical experiment was able to identify areas requiring tailored modeling solutions or data collection. Moreover, this study highlighted the high sensitivity of inundation volume and extent prediction to uncertainties in flood peak values and explored challenges in time‐continuous modeling. Accurate flood peak predictions, knowledge of critical morphological features, and an event‐based modeling approach were outlined as pragmatic solutions for more accurate prediction of large‐scale spatiotemporal patterns of flood dynamics, particularly in the presence of low‐accuracy elevation data.
Plain Language Summary
Floods are among the most devastating natural hazards, affecting multiple regions and millions of people each year. Accurate inundation predictions are vital information for land and emergency management. This objective can be achieved through a cascade of numerical models. However, each model is subject to uncertainties from different sources (e.g., input data, model structure, and parameters), and an understanding of how these uncertainties are propagated through each step of the modeling cascade is pivotal to improving inundation prediction accuracy. This study investigated the impact of uncertainties in streamflow predictions on the accuracy of floodplain inundation predictions. For this purpose, the Murray Darling Basin (Australia), a large basin that is affected by destructive floods, was used as a case study. The analysis illustrated the high sensitivity of floodplain inundation predictions to predicted streamflow peak values. Moreover, when attempting to model a long time series of low‐ and high‐flow periods, uncertainties in the inundation patterns increased over time and from upstream to downstream areas of the basin. These results demonstrated the need for accurate predictions of streamflow peak values and suggested that focusing on the modeling of each large flood event separately is a more effective strategy for reliable inundation predictions.
Key Points
A numerical experiment was used to improve understanding of uncertainty propagation in coupled hydrologic‐hydraulic models
Discrepancies in measured and simulated flow peaks lead to large uncertainties in predicted floodplain inundation volumes and extent
The challenges of multiyear continuous modeling are highlighted with an event‐based approach recommended
We present the result of a search of the Milagro sky map for spatial correlations with sources from a subset of the recent Fermi Bright Source List (BSL). The BSL consists of the 205 most significant ...sources detected above 100 MeV by the Fermi Large Area Telescope. We select sources based on their categorization in the BSL, taking all confirmed or possible Galactic sources in the field of view of Milagro. Of the 34 Fermi sources selected, 14 are observed by Milagro at a significance of 3 standard deviations or more. We conduct this search with a new analysis which employs newly optimized gamma-hadron separation and utilizes the full eight-year Milagro data set. Milagro is sensitive to gamma rays with energy from 1 to 100 TeV with a peak sensitivity from 10 to 50 TeV depending on the source spectrum and declination. These results extend the observation of these sources far above the Fermi energy band. With the new analysis and additional data, multi-TeV emission is definitively observed associated with the Fermi pulsar, J2229.0+6114, in the Boomerang pulsar wind nebula (PWN). Furthermore, an extended region of multi-TeV emission is associated with the Fermi pulsar, J0634.0+1745, the Geminga pulsar.
We monitored the chromospheric activity in the Ca II H and K lines of 13 solar-type stars (including the Sun): 8 of them over 3 years at the Canada-France-Hawaii Telescope (CFHT) and 5 in a single ...run at the Very Large Telescope (VLT). A total of 10 of the 13 targets have close planetary companions. All of the stars observed at the CFHT show long-term (months to years) changes in H and K intensity levels. Four stars display short-term (days) cyclical activity. For two, HD 73256 and Kappa super(1) Cet, the activity is likely associated with an active region rotating with the star; however, the flaring in excess of the rotational modulation may be associated with a hot Jupiter. A planetary companion remains a possibility for Kappa super(1) Cet. For the other two, HD 179949 and upsilon And, the cyclic variation is synchronized to the hot Jupiter's orbit. For both stars this synchronicity with the orbit is clearly seen in two out of three epochs. The effect is only marginal in the third epoch at which the seasonal level of chromospheric activity had changed for both stars. Short-term chromospheric activity appears weakly dependent on the mean K line reversal intensities for the sample of 13 stars. In addition, a suggestive correlation exists between this activity and the M sub(p) sin i of the star's hot Jupiter. Because of their small separation ( less than or equal to 0.1 AU), many of the hot Jupiters lie within the Alfven radius of their host stars, which allows a direct magnetic interaction with the stellar surface. We discuss the conditions under which a planet's magnetic field might induce activity on the stellar surface and why no such effect was seen for the prime candidate, tau Boo. This work opens up the possibility of characterizing planet-star interactions, with implications for extrasolar planet magnetic fields and the energy contribution to stellar atmospheres.