–

–

Game theory and artificial intelligence are deeply connected fields, particularly when model-
ing adversarial learning problems as Stackelberg games under uncertainty. In such settings,

defenders allocate resources randomly to prevent attackers from anticipating their moves. As
time evolves, the environment changes, and new strategies or actions may emerge. A suitable
subfield of game theory to model these dynamics is stochastic games, where the state captures

the set of available actions in a time-dependent process. Moreover, the attacker–defender inter-
action naturally exhibits a hierarchical bilevel structure: the defender must design randomized

strategies to protect targets, while the attacker responds by executing an action that maxi-
mizes its reward. This makes stochastic dynamic Stackelberg games a model of choice for such

applications [15], the Chilean party being very active in the domain [9, 10, 11].
In a previous collaboration [1], we have shown that defining an appropriate solution concept
for these games is far from trivial, and in some cases remains an open research question. Even
when a proper solution concept can be established, computing it becomes prohibitively expensive
for games with massive state and action spaces.
These challenges are not only of theoretical interest but also of high practical relevance.
This associate team will focus on two among the many possible applications.
Challenge 1: The first one is wildfire prevention. This theme has been widely examined in
the literature, particularly through models that optimize the design of firebreaks [16], [12] and
the the placement of surveillance cameras to limit fire spread. In contrast, the operational

dimension of prevention and firefighting—where resources must be dynamically allocated un-
der uncertainty—has received far less attention. Emerging distributed AI technologies with

autonomous agents, both centralized (e.g., coordinated command-and-control systems) and de-
centralized (e.g., drones and sensor swarms), could play a key role in addressing this challenge.

Challenge 2: Similarly, in the domain of distributed AI learning systems (e.g. Federated
Learning), a central authority is in charge of agregating the updates of distributed clients in the
process of minimizing a global metric. The presence of Byzantine (or faulty) agents can derail
convergence or induce a biais in this process. To mitigate this, robust agregation techniques
have been proposed in the literature based on identifying and discarding suspect contributions
[17, 18, 19]. But concrete attacks like ALIE [20] or FoE [21] tend to exhibit patterns. This lead
researchers to propose the first history-aware defenses [22, 23, 24]. Purely deterministic rules,
though, are likely to be defeated by powerful and well-informed agents. The attacker–defender
framework is crucial to design secure training environments against adversarial manipulations,
the principal issue being to decide if and when to exclude a suspected malicious agent.
Both these applications can be tackeled with tools of Game Theory. More applications
could be cited, like environmental conservation problems, such as poaching prevention [25],
environmental tax evasion [26] or border patrol security [9].

Game theory and artificial intelligence are deeply connected fields, particularly when model-
ing adversarial learning problems as Stackelberg games under uncertainty. In such settings,

defenders allocate resources randomly to prevent attackers from anticipating their moves. As
time evolves, the environment changes, and new strategies or actions may emerge. A suitable
subfield of game theory to model these dynamics is stochastic games, where the state captures

the set of available actions in a time-dependent process. Moreover, the attacker–defender inter-
action naturally exhibits a hierarchical bilevel structure: the defender must design randomized

strategies to protect targets, while the attacker responds by executing an action that maxi-
mizes its reward. This makes stochastic dynamic Stackelberg games a model of choice for such

applications [15], the Chilean party being very active in the domain [9, 10, 11].
In a previous collaboration [1], we have shown that defining an appropriate solution concept
for these games is far from trivial, and in some cases remains an open research question. Even
when a proper solution concept can be established, computing it becomes prohibitively expensive
for games with massive state and action spaces.
These challenges are not only of theoretical interest but also of high practical relevance.
This associate team will focus on two among the many possible applications.
Challenge 1: The first one is wildfire prevention. This theme has been widely examined in
the literature, particularly through models that optimize the design of firebreaks [16], [12] and
the the placement of surveillance cameras to limit fire spread. In contrast, the operational

dimension of prevention and firefighting—where resources must be dynamically allocated un-
der uncertainty—has received far less attention. Emerging distributed AI technologies with

autonomous agents, both centralized (e.g., coordinated command-and-control systems) and de-
centralized (e.g., drones and sensor swarms), could play a key role in addressing this challenge.

Challenge 2: Similarly, in the domain of distributed AI learning systems (e.g. Federated
Learning), a central authority is in charge of agregating the updates of distributed clients in the
process of minimizing a global metric. The presence of Byzantine (or faulty) agents can derail
convergence or induce a biais in this process. To mitigate this, robust agregation techniques
have been proposed in the literature based on identifying and discarding suspect contributions
[17, 18, 19]. But concrete attacks like ALIE [20] or FoE [21] tend to exhibit patterns. This lead
researchers to propose the first history-aware defenses [22, 23, 24]. Purely deterministic rules,
though, are likely to be defeated by powerful and well-informed agents. The attacker–defender
framework is crucial to design secure training environments against adversarial manipulations,
the principal issue being to decide if and when to exclude a suspected malicious agent.
Both these applications can be tackeled with tools of Game Theory. More applications
could be cited, like environmental conservation problems, such as poaching prevention [25],
environmental tax evasion [26] or border patrol security [9].

Postulación a Fondecyt Regular

Postulación a Fondecyt Regular

The phenomenon of a soft liquid drop eroding a hard stone surface over time, immortalized in the ancient proverb «dripping water wears away the stone,» presents a profound mechanical puzzle. While craters are common imprints of high-energy events, those formed by persistent, low-energy water dripping are exceptional. The impact energy of a single drop is far below the threshold required to plastically deform or fracture the material, yet erosion occurs. This project seeks to answer the fundamental question: How can water erode stone through dripping and create distinctive craters?
While recent advancements in drop-impact dynamics have revealed that an impacting drop generates propagating fronts of intense, singular pressure and shear, these theories were developed for ideal, non-porous surfaces and are insufficient to explain the erosion. Our preliminary experimental work—which has successfully reproduced water-dripping craters on gypsum targets while failing to erode non-porous materials—points to a crucial, previously overlooked element: the porous nature of the target material. We discovered that erosion and the formation of a distinct surface microstructure of pores commence only after the substrate becomes fully saturated with water. This key finding suggests that the complex interaction between the impact-induced flow and the internal, liquid-filled pore structure is the primary driver of the erosion mechanism.
This project will establish the first comprehensive experimental and theoretical framework for slow erosion in porous ma- terials by water dripping. We will investigate three potential and non-exclusive micro-mechanisms. The first is low-Reynolds accumulative erosion, where the impact pressure pumps liquid into the matrix, generating high shear stress along pore walls that slowly abrades material, a process whose rate is expected to be proportional to the wall shear stress. The second is the inter-pore propagation of pressure shocks; because the surface pressure front arrives at adjacent pore openings at slightly different times, large pressure gradients are generated within the saturated matrix, inducing mechanical fatigue and failure of inter-pore walls. The third is cavitation bursts, where the negative-pressure front trailing the initial impact shock— akin to an explosion’s blast wave—causes the formation and violent collapse of vapor bubbles. These collapses generate localized but highly destructive shock waves, a process potentially detectable via acoustic emissions.
Our methodology integrates a novel, multi-scale experimental approach with robust theoretical modeling. An automated, custom-built setup, featuring a syringe pump for precise drop control and a photo-gate for impact counting and synchroniza- tion, tracks crater evolution over tens of thousands of reproducible impacts. An automated translation stage will move the sample between the impact zone and a characterization chamber for on-the-run 3D shape reconstruction via high-resolution laser profilometry and for mass measurement via an integrated load cell. This will be complemented by a suite of characteriza- tion techniques, including high-speed imaging to capture rare ejecta events, microscopic surface imaging, and advanced bulk imaging (X-ray Micro-Tomography, Scanning Electron Microscopy or Nuclear Magnetic Resonance) to visualize the internal 3D pore network and wear propagation. Experiments will mainly utilize natural materials like gypsum and selenite, as well as custom-fabricated synthetic porous samples (e.g., PDMS). These transparent, engineered samples will allow for direct flow visualization via Particle Image Velocimetry (PIV) to isolate and study specific mechanisms in a controlled environment.
The theoretical work will couple established models for drop-impact pressure distributions with frameworks for flow in porous media, wall-shear erosion, and wave propagation. The goal is to develop predictive formulae for crater growth rates and their scaling with fluid and material properties, which can be validated against our extensive experimental data. By leveraging the research team’s expertise in drop-impact forces and tackling this 2,500-year-old question, this project will provide novel insights into fluid-solid interactions, wear on porous materials, and landscape evolution. It moves beyond prior studies, which used simplified substrates, to address the central role of porosity in this long-unsolved problem in continuum physics.

The phenomenon of a soft liquid drop eroding a hard stone surface over time, immortalized in the ancient proverb «dripping water wears away the stone,» presents a profound mechanical puzzle. While craters are common imprints of high-energy events, those formed by persistent, low-energy water dripping are exceptional. The impact energy of a single drop is far below the threshold required to plastically deform or fracture the material, yet erosion occurs. This project seeks to answer the fundamental question: How can water erode stone through dripping and create distinctive craters?
While recent advancements in drop-impact dynamics have revealed that an impacting drop generates propagating fronts of intense, singular pressure and shear, these theories were developed for ideal, non-porous surfaces and are insufficient to explain the erosion. Our preliminary experimental work—which has successfully reproduced water-dripping craters on gypsum targets while failing to erode non-porous materials—points to a crucial, previously overlooked element: the porous nature of the target material. We discovered that erosion and the formation of a distinct surface microstructure of pores commence only after the substrate becomes fully saturated with water. This key finding suggests that the complex interaction between the impact-induced flow and the internal, liquid-filled pore structure is the primary driver of the erosion mechanism.
This project will establish the first comprehensive experimental and theoretical framework for slow erosion in porous ma- terials by water dripping. We will investigate three potential and non-exclusive micro-mechanisms. The first is low-Reynolds accumulative erosion, where the impact pressure pumps liquid into the matrix, generating high shear stress along pore walls that slowly abrades material, a process whose rate is expected to be proportional to the wall shear stress. The second is the inter-pore propagation of pressure shocks; because the surface pressure front arrives at adjacent pore openings at slightly different times, large pressure gradients are generated within the saturated matrix, inducing mechanical fatigue and failure of inter-pore walls. The third is cavitation bursts, where the negative-pressure front trailing the initial impact shock— akin to an explosion’s blast wave—causes the formation and violent collapse of vapor bubbles. These collapses generate localized but highly destructive shock waves, a process potentially detectable via acoustic emissions.
Our methodology integrates a novel, multi-scale experimental approach with robust theoretical modeling. An automated, custom-built setup, featuring a syringe pump for precise drop control and a photo-gate for impact counting and synchroniza- tion, tracks crater evolution over tens of thousands of reproducible impacts. An automated translation stage will move the sample between the impact zone and a characterization chamber for on-the-run 3D shape reconstruction via high-resolution laser profilometry and for mass measurement via an integrated load cell. This will be complemented by a suite of characteriza- tion techniques, including high-speed imaging to capture rare ejecta events, microscopic surface imaging, and advanced bulk imaging (X-ray Micro-Tomography, Scanning Electron Microscopy or Nuclear Magnetic Resonance) to visualize the internal 3D pore network and wear propagation. Experiments will mainly utilize natural materials like gypsum and selenite, as well as custom-fabricated synthetic porous samples (e.g., PDMS). These transparent, engineered samples will allow for direct flow visualization via Particle Image Velocimetry (PIV) to isolate and study specific mechanisms in a controlled environment.
The theoretical work will couple established models for drop-impact pressure distributions with frameworks for flow in porous media, wall-shear erosion, and wave propagation. The goal is to develop predictive formulae for crater growth rates and their scaling with fluid and material properties, which can be validated against our extensive experimental data. By leveraging the research team’s expertise in drop-impact forces and tackling this 2,500-year-old question, this project will provide novel insights into fluid-solid interactions, wear on porous materials, and landscape evolution. It moves beyond prior studies, which used simplified substrates, to address the central role of porosity in this long-unsolved problem in continuum physics.

El borde costero de la región de O’Higgins enfrenta un desafío importante en cuanto a alcanzar un desarrollo
sostenible a nivel ambiental. Esto se debe a la exposición a contaminantes como pesticidas y metales
provenientes de actividades clave de la región, como lo son la agricultura y la minería. Esta contaminación
amenaza los ecosistemas del borde costero, la biodiversidad, la salud pública y las actividades económicas
fundamentales como la pesca artesanal y el turismo. A pesar de la magnitud del problema, existe escasa
información sobre los contaminantes presentes en este ambiente. La ausencia de sistemas de monitoreo
limita la caracterización de la presencia de contaminantes y la evaluación de sus efectos sobre los ecosistemas
costeros. Esta brecha dificulta la implementación de estrategias efectivas de gestión ambiental y pone en
riesgo la sostenibilidad de las comunidades costeras que dependen de estos recursos naturales.
El proyecto propone desarrollar una herramienta de biosurveillance adaptada a las características específicas
del borde costero regional, inspirada en el exitoso modelo francés BIOSURVEILLANCE que utiliza organismos
centinelas para evaluar la calidad del agua. La iniciativa busca establecer un sistema piloto de monitoreo que
combine análisis químicos de contaminantes con evaluación de biomarcadores inmunológicos en mejillones,
organismos filtradores que bioacumulan contaminantes y que sirven como reflejo del estado de salud del
ecosistema.
El objetivo general del proyecto es evaluar el estado de salud inmunológico de organismos centinela del borde
costero de la Región de O’Higgins y su relación con la presencia de contaminantes, desarrollando un piloto de
biosurveillance transferible a la gestión ambiental y acuicultura. Este proyecto cuenta con cuatro objetivos
específicos: (1) caracterizar la presencia y concentración de contaminantes persistentes (metales y pesticidas)
en agua, sedimentos y biota; (2) evaluar las respuestas inmunológicas en organismos centinela mediante
biomarcadores moleculares; (3) analizar la relación entre contaminantes específicos y respuestas
inmunológicas para identificar biomarcadores sensibles y robustos; y (4) desarrollar un piloto de
biosurveillance con potencial de transferencia a programas de gestión regional y nacional.
Para llevar a cabo este proyecto se implementará un sistema de encajonamiento de mejillones (Mytilus edulis
o M. chilensis) en tres sitios estratégicos del borde costero: Navidad, Pichilemu y Bucalemu. Estos sitios
representan puntos críticos de confluencia entre aportes continentales y ecosistema marino, permitiendo
evaluar efectos acumulativos de contaminación. Se realizarán dos campañas de muestreo (invierno y verano)
para capturar variación temporal. En cada sitio se colocarán individuos adultos en jaulas durante 15 días,
período tras el cual se recuperarán para análisis químicos y biológicos.
El componente químico incluirá análisis de metales mediante espectrometría (ICP-OES) y pesticidas mediante
cromatografía (LC-MS/MS) en matrices de agua, sedimentos y tejidos. Paralelamente, se caracterizarán
parámetros fisicoquímicos (pH, conductividad, temperatura, oxígeno disuelto, salinidad) para contextualizar
biogeoquímicamente los sitios.
El componente biológico evaluará biomarcadores de respuesta inmune mediante análisis de expresión génica
por qPCR de genes relacionados con estrés oxidativo (catalasa, superóxido dismutasa, glutatión S-transferasa,
metalotioneínas) en tejidos de mejillones. Estos biomarcadores son sensibles a la exposición a metales y
pesticidas, proporcionando indicadores tempranos de estrés ambiental.
La integración de datos químicos y biológicos se realizará mediante análisis estadísticos multivariados y
matrices de correlación utilizando lenguaje de programación (Phyton y R), identificando relaciones causales
entre contaminantes específicos y activación del sistema inmune. Esta aproximación permitirá seleccionar los
biomarcadores más apropiados según ubicación geográfica y estacionalidad, estableciendo las bases para un
sistema de alerta temprana aplicable a la gestión costera regional.
El proyecto tiene una duración de 18 meses y representa una iniciativa interdisciplinaria que integra Ciencias
del Medio Ambiente, Biología Marina y Matemática Aplicada, contribuyendo directamente a la sostenibilidad
ambiental del borde costero regional y generando conocimiento transferible para futuros programas de
monitoreo a escala nacional.

El borde costero de la región de O’Higgins enfrenta un desafío importante en cuanto a alcanzar un desarrollo
sostenible a nivel ambiental. Esto se debe a la exposición a contaminantes como pesticidas y metales
provenientes de actividades clave de la región, como lo son la agricultura y la minería. Esta contaminación
amenaza los ecosistemas del borde costero, la biodiversidad, la salud pública y las actividades económicas
fundamentales como la pesca artesanal y el turismo. A pesar de la magnitud del problema, existe escasa
información sobre los contaminantes presentes en este ambiente. La ausencia de sistemas de monitoreo
limita la caracterización de la presencia de contaminantes y la evaluación de sus efectos sobre los ecosistemas
costeros. Esta brecha dificulta la implementación de estrategias efectivas de gestión ambiental y pone en
riesgo la sostenibilidad de las comunidades costeras que dependen de estos recursos naturales.
El proyecto propone desarrollar una herramienta de biosurveillance adaptada a las características específicas
del borde costero regional, inspirada en el exitoso modelo francés BIOSURVEILLANCE que utiliza organismos
centinelas para evaluar la calidad del agua. La iniciativa busca establecer un sistema piloto de monitoreo que
combine análisis químicos de contaminantes con evaluación de biomarcadores inmunológicos en mejillones,
organismos filtradores que bioacumulan contaminantes y que sirven como reflejo del estado de salud del
ecosistema.
El objetivo general del proyecto es evaluar el estado de salud inmunológico de organismos centinela del borde
costero de la Región de O’Higgins y su relación con la presencia de contaminantes, desarrollando un piloto de
biosurveillance transferible a la gestión ambiental y acuicultura. Este proyecto cuenta con cuatro objetivos
específicos: (1) caracterizar la presencia y concentración de contaminantes persistentes (metales y pesticidas)
en agua, sedimentos y biota; (2) evaluar las respuestas inmunológicas en organismos centinela mediante
biomarcadores moleculares; (3) analizar la relación entre contaminantes específicos y respuestas
inmunológicas para identificar biomarcadores sensibles y robustos; y (4) desarrollar un piloto de
biosurveillance con potencial de transferencia a programas de gestión regional y nacional.
Para llevar a cabo este proyecto se implementará un sistema de encajonamiento de mejillones (Mytilus edulis
o M. chilensis) en tres sitios estratégicos del borde costero: Navidad, Pichilemu y Bucalemu. Estos sitios
representan puntos críticos de confluencia entre aportes continentales y ecosistema marino, permitiendo
evaluar efectos acumulativos de contaminación. Se realizarán dos campañas de muestreo (invierno y verano)
para capturar variación temporal. En cada sitio se colocarán individuos adultos en jaulas durante 15 días,
período tras el cual se recuperarán para análisis químicos y biológicos.
El componente químico incluirá análisis de metales mediante espectrometría (ICP-OES) y pesticidas mediante
cromatografía (LC-MS/MS) en matrices de agua, sedimentos y tejidos. Paralelamente, se caracterizarán
parámetros fisicoquímicos (pH, conductividad, temperatura, oxígeno disuelto, salinidad) para contextualizar
biogeoquímicamente los sitios.
El componente biológico evaluará biomarcadores de respuesta inmune mediante análisis de expresión génica
por qPCR de genes relacionados con estrés oxidativo (catalasa, superóxido dismutasa, glutatión S-transferasa,
metalotioneínas) en tejidos de mejillones. Estos biomarcadores son sensibles a la exposición a metales y
pesticidas, proporcionando indicadores tempranos de estrés ambiental.
La integración de datos químicos y biológicos se realizará mediante análisis estadísticos multivariados y
matrices de correlación utilizando lenguaje de programación (Phyton y R), identificando relaciones causales
entre contaminantes específicos y activación del sistema inmune. Esta aproximación permitirá seleccionar los
biomarcadores más apropiados según ubicación geográfica y estacionalidad, estableciendo las bases para un
sistema de alerta temprana aplicable a la gestión costera regional.
El proyecto tiene una duración de 18 meses y representa una iniciativa interdisciplinaria que integra Ciencias
del Medio Ambiente, Biología Marina y Matemática Aplicada, contribuyendo directamente a la sostenibilidad
ambiental del borde costero regional y generando conocimiento transferible para futuros programas de
monitoreo a escala nacional.