Virtual laboratory combining atmospheric models and measurements elected Center of Excellence
The name of the Center of Excellence led by Professor Vehkamaki is Virtuaalinen laboratorio ilmakehÃ¤n molekyylitason reaktioille ja faasimuutoksille (Virtual Laboratory for Molecular Level Reactions and Phase Changes).
Research measures and models atmospheric molecules that can condense into particles. Recent findings from the SMEAR station have proven that gases from plants, for example, can form fine particles, which in turn condense the droplets in clouds.
AI enables new modeling of molecular measurements
Research in the virtual lab uses both physical and chemical methods, as well as machine learning skills.
– The main idea is to build a compilation of virtual instruments and models. With the help of machine learning and artificial intelligence, we can combine measurements from different instruments and emulate the function of models – whatever happens before we have a measurement and understand what it means, said VehkamÃ¤ki.
By comparison, we can find out which measurements or calculations need to be done in more detail. Foresight is important because measurements and computational calculations at the molecular level take time and money, and they should not be undertaken unless you are sure they will be useful.
– Some computer operations lasted up to six years. Instruments are also expensive, so we have to use them appropriately. Our goal is to measure and model more efficiently, explains VehkamÃ¤ki.
The project includes developers of AI methods, such as associate professor Kai Puolamaki and researcher in the physics of machine learning materials, Associate Professor Patrick Rincke of Aalto University.
Doctoral and post-doctoral students will also be employed in the project.
– Each will have two supervisors with different skill profiles. This is how multidisciplinary researchers are educated from the start, says VehkamÃ¤ki.
Expertise arises from mutual understanding and different worlds
This is still a pilot project for a new type of research method. Such new ideas require years of development because there are no ready-made solutions; they must be developed from scratch.
– It is better to develop our research methods ourselves, because we cannot expect others to develop an AI that can be applied to our research problems, explains VehkamÃ¤ki.
The potential for combining measures is rarely used in research, although it can help us discover new methods for our work.
– Some research methods not currently in use might give us more precise results, but we cannot know for sure until we can compare the results in a controlled manner, explains VehkamÃ¤ki.
She wants to remind us that the exploration of frontline research directions can never be fully controlled, but rather that new opportunities emerge through trial and error.
– As we grow together, we discover new and innovative solutions. Funding from the Academy allows us to test new ideas and develop together over the long term, explains VehkamÃ¤ki.
Besides VehkamÃ¤ki, Arkke Eskola, Juha Kangasluoma, Theo KurtÃ©n, Kai PuolamÃ¤ki and Mikko SipilÃ¤ are also members of the CoE of the University of Helsinki.