Scientists demonstrate chemical reservoir computation using the formose reaction

Researchers at the Institute for Molecular Materials at Radboud University in the Netherlands have demonstrated that complex self-organizing chemical reaction networks can perform a variety of computational tasks, including nonlinear classification and prediction of complex dynamics.

The field of molecular computing is Computational capabilities Chemistry and Biological SystemsIn these systems, Chemical reaction or Molecular Processes It acts as a reservoir computer, transforming inputs into higher dimensional outputs.

research, Published in Naturewas led by Professor Wilhelm Hack of Radboud University.

Researchers have harnessed the potential of chemical and biological networks for complex computing power, but implementing molecular computing presents engineering and design challenges.

Instead of trying to engineer Molecular Systems To perform specific computational tasks, Professor Huck and his team are investigating how naturally complex chemical systems can exhibit novel computational properties.

“I am very interested in the chemical driving forces that guided the origin of life. In this context, we are looking for mechanisms by which chemical evolution shapes the properties of complex reaction mixtures. This work prompted me to think about how molecular systems can process information,” he explained to Phys.org.

Formose reaction

The formose reaction is a chemical reaction that synthesizes sugars from formaldehyde in the presence of a catalyst, calcium hydroxide. This reaction was chosen because of its unique properties.

Professor Huck explains: “Although chemistry may seem complex to an outsider, most reaction sequences are rather linear. The formose reaction is a unique example of a self-organising reaction network with a highly non-linear topology, involving numerous positive and negative feedback loops.”

In other words, the reactions are not simple, but involve the generation of multiple intermediate compounds that can react further to form new compounds. These dynamic reactions result in a wide variety of chemical species and are inherently nonlinear.

In addition, the network contains positive feedback loops that amplify the reaction outcome and negative feedback loops that attenuate the reaction outcome.

The networks are known as “self-organizing” because they evolve naturally, without the need for external intervention, and respond to chemical inputs to produce diverse outputs.

Computational power emerges from the inherent properties of the network rather than being explicitly programmed, making the computations extremely flexible.

Reservoir computer implementation

The researchers used a continuous stirred tank reactor (CSTR) to carry out the Formose reaction, controlling the input concentrations of four reactants (formaldehyde, dihydroxyacetone, sodium hydroxide, and calcium chloride) to regulate the behavior of the reaction network.

The output molecules are identified using a mass spectrometer, which can track up to 106 molecules. Calculations can be made using this setup, with the reactant concentrations becoming inputs to the functions that need to be calculated.

But first we need to train the system to find the outcome of this calculation, which we do by using a set of weights.

“We need to find a set of weights that transform the mass spectrometer trace into the correct calculated value. This is a linear regression problem and is easy to compute. Once complete, the reservoir computer calculates the result of this function for any new input,” explains Professor Huck.

The weights are coefficients that determine how much influence each input has on the output. This training step is essential as it allows the reservoir to learn and predict how changes in the inputs affect the output, allowing it to predict the output for a new set of inputs.

Computational capabilities

The researchers used the Reservoir computer to perform several tasks. The first task was to perform a nonlinear classification task. The Reservoir computer can emulate all Boolean logic gates and can also handle more complex classifications such as XOR, checker, circle, and sine functions.

The team also showed that they could predict the behavior of a complex metabolic network model of E. coli, accurately capturing both linear and nonlinear responses to varying inputs over a range of concentrations.

Additionally, the system demonstrated the ability to predict the future state of a chaotic system (Lorenz attractor), accurately predicting two out of three input dimensions several hours ahead.

The team also found that some of the chemical species in the system exhibited short-term memory, retaining information about past inputs.

They also provided a proof-of-concept of a complete chemical readout using a colorimetric reaction, showing how the state of the system can be interpreted without electronic measuring devices.

In other words, the color change caused by a chemical reaction can be used to interpret the state of the system, eliminating the need for electronic measuring devices.

The origin of life, neuromorphic computing, and beyond

This new approach to molecular computing has the potential to bridge the gap between the information-processing capabilities of artificial systems and living cells.

More scalable Flexible approach It has expanded to molecular computing, opening up the possibility of creating autonomous chemical systems that can process information and react to their environment without external electronic control.

Professor Huck expressed his team’s interest in this field, saying, “Can we incorporate reservoir computing into chemical systems that sense their environment, process that information and take appropriate action?”

“To do this, we need to combine the reservoir with other elements that can translate the brain’s chemical output into some form of mechanical response or interaction with living cells.”

The work also has intriguing implications for the origin of life: the novel computational properties of this relatively simple chemical system may provide insight into how early biological systems developed information-processing capabilities.

Professor Huck said this was his main motivation for studying reservoir calculations.

The research team also sees potential in neuromorphic computing, which mimics the neural structure and function of the human brain to improve computational efficiency and processing power.

“We are very interested in exploring the technical limits of Formosus’s computational capabilities. reservoir The computer is currently being researched in collaboration with IBM Zurich. Reservoir Computing is Neuromorphic Computing“They are attracting attention because they are expected to consume less power than conventional computers,” Professor Hack explained.

© 2024 Science X Network