Early studies of the thermal state of the Cocos plate subducting beneath the northern Costa Rica margin identified two adjacent areas of higher (>100 mW/m 2) and lower (<30 mW/m 2) heat flow. Measurements on crust formed at the Cocos-Nazca spreading center

In order to evaluate the setting of the Humboldt-Rye Patch geothermal fi eld, we carried out a program of hyperspectral and light detection and ranging (LiDAR) imaging of the Humboldt River basin to test (1) whether fault patterns, surface mineral alteration, and mud volcanoes in the Humboldt-Rye Patch district offer the potential for additional geothermal exploration sites;(2) whether mud diapirism in this region could be caused by seismic shaking; and (3) whether significant improvements in exploration can be made using these remotesensing tools in addition to the more traditional techniques. In the southern (Rye Patch) region, a set of faults cuts the surface of the alluvial fans, and several faults cut shorelines of Lake Lahontan. These shorelines lie at an elevation of 1290 m, which corresponds with the elevation of the Lake 12,500 ± 500 yr ago. We fi nd no signs of surface mineral alteration in the Rye Patch area in spite of the existence of these faults and known alteration at depth. Farther north, in the Humboldt House region, we find abundant evidence of alteration products, including siliceous sinter, carbonate, montmorillonite, hematite, and jarosite. This alteration is widespread, and corresponds to young faulting in only one location. The LiDAR data show at least two mud volcanoes and a large fi eld of low-carbonate mounds. Some of these (apparently) diapiric features may have been associated with seismicity, and both active and paleoseismic events would have been suffi ciently close and energetic to have initiated liquefaction in this region. Such liquefaction events would have been more likely, however, during the high stands of Lake Lahontan, when the ground would have been saturated, consistent with reported ages on rocks correlated with the carbonate mounds. We propose further geothermal exploration based on these results.

Fluid flow along normal faults has created much of the geothermal energy that is currently being exploited in the Basin & Range. We used remote sensing (HyMap, ASTER) data and field-based methods in the Humboldt Block of the northwest Basin & Range to map fault zones and the surface distribution of minerals associated with hydrothermal fluid flow. The Humboldt Block lies on the Battle Mountain High heat flow area (>100mW/m 2 ), and has very high shallow water temperatures of around 200 o F. This area has undergone large amounts of extension from Oligocene to the present, accommodated along large, range-bounding normal faults. The western flank of the Humboldt Range is bounded by a normal fault that trends N-NE, dips W, and brings into contact Mesozoic sedimentary and volcanic rocks with Quaternary deposits.The structural setting, high heat flow, and high shallow water temperatures suggest significant geothermal potential for the Humboldt Block. We carried out the remote sensing study in two stages. We created fault maps by overlaying ASTER data onto a DEM, and mapping lithologic changes and structurally controlled lineations that were distinct from expected topographic patterns. This revealed two distinct fault patterns: N-NE trending faults and NW-SE trending faults. We then used HyMap (hyperspectral) data to identify and map minerals associated with hydrothermal fluid flow and their relation to regional structure. Sinter was mapped in the Humboldt River Valley along two distinct N-NE trends. Kaolinite was mapped in the Humboldt Range along the range front fault, and NW-SE trends.

<p>Combining quantum chemistry characterizations with generative machine learning models has the potential to accelerate molecular searches in chemical space. In this paradigm, quantum chemistry acts as a relatively cost-effective oracle for evaluating the properties of particular molecules while generative models provide a means of sampling chemical space based on learned structure-function relationships. For practical applications, multiple potentially orthogonal properties must be optimized in tandem during a discovery workflow. This carries additional difficulties associated with specificity of the targets and the ability for the model to reconcile all properties simultaneously. Here we demonstrate an active learning approach to improve the performance of multi-target generative chemical models. We first demonstrate the effectiveness of a set of baseline models trained on single property prediction tasks in generating novel compounds with various property targets, including both interpolative and extrapolative generation scenarios. For property ranges where accurate targeting proves difficult, the novel compounds suggested by the model are characterized using quantum chemistry to obtain the true values, and these new molecules closest to expressing the desired properties are fed back into the generative model for additional training. This gradually improves the generative models’ understanding of unknown areas of chemical space and shifts the distribution of generated compounds towards the targeted values. We then demonstrate the effectiveness of this active learning approach in generating compounds with multiple chemical constraints, including vertical ionization potential, electron affinity, and dipole moment targets, and validate the results at the B97X-D3/def2-TZVP level. This method requires no modifications to extant generative approaches, but rather utilizes their inherent generative and predictive aspects for self-refinement, and can be applied to situations where any number of properties with varying degrees of correlation must be optimized simultaneously.</p>