How they managed to unravel the mystery of nitrogenasis on classical algorithms.
Chemists at the California Institute of Technology have calculated one of the most difficult parts of the enzyme, without which life on Earth would not have gained access to atmospheric nitrogen. The work concerns nitrogenase, a protein that converts molecular nitrogen into ammonia. For theoretical chemistry, this is an important result in itself, but the dispute around it went far beyond biology: the calculation was performed on conventional computers, although nitrogenase for many years was given as a task where quantum machines should get a decisive advantage.
Garnet Chan, one of the leading specialists in quantum chemistry, has long been engaged not so much in the future of computing as with specific chemical tasks. He is interested in how to describe complex reactions in living systems. At the same time, it was Chan’s works that were at the center of a discussion on whether it is necessary to wait for full-fledged fault-tolerant quantum computers to deal with the heaviest molecular systems.
The new result strengthens Chan’s position. In early January, he and five other chemists from Caltech reached an important stage in the description of nitrogenase. They managed to calculate the state of the active center of the enzyme by classical methods, that is, without a quantum computer. For the area where the thesis of the almost inevitable superiority of quantum computing has been sounded for many years, this is a noticeable argument.
Nitrogenase refers to key molecules of terrestrial biology. Along with photosynthesis, nitrogen fixation is among the processes, without which a complex life would not be able to develop in the usual form. The atmosphere is almost 80% nitrogen, but it is in the air as a molecule N2. Two atoms are bound by a strong triple bond, so living organisms can’t just take this gas and embed it in proteins, DNA or other substances.
Before the advent of nitrogenazase, life depended on rare events that broke the bond in the nitrogen molecule. One such source was lightning: high-energy discharges turned part of atmospheric nitrogen into compounds suitable for biological processes. About three billion years ago, the early prokaryotes got a different path. Nitrogenase has learned to tear N2 and turn inert gas into ammonia, which can be used to build living matter.
The enzyme solves this problem through the active center FeMo-co. This is a cluster of iron and molybdenum atoms, where the main chemistry takes place. It is he who makes nitrogenase so difficult for calculations. Each iron atom carries four or five unpaired electrons, and their state depends not only on its own position, but also on the behavior of other particles in the cluster.
This is how the problem of electronic correlation arises. In simple models, electrons can sometimes be considered almost independently, but FeMo-co does not allow such rudeness. Particles are connected by quantum dependencies, so the change in one part of the system is reflected in another. It is necessary to understand the general electronic structure and energy of the cluster, and the number of possible configurations grows explosively.
Historically, people have long been interested not in the mechanism of nitrogenase, but ammonia as a practical product. In the 19 century, a reliable source of available nitrogen was guano from the islands off the coast of Peru, and this resource was valued so highly that it had conflicts between states. In 1909, Fritz Haber and Carl Bosch created an industrial process of fixing nitrogen, which allowed mass-produced ammonia for fertilizers. But the scientific mystery remains: soil bacterium does under normal conditions what industry performs in a rigid chemical mode.
Quantum computers entered this dispute because of the very nature of molecular calculations. An ordinary computer works with bits that take a value of zero or unit. Quantum uses qubits that can be in superposition and get confused with each other. For tasks with a suitable mathematical structure, such a machine is theoretically able to give a huge gain in speed.
Chemistry has long been considered one of the most natural uses of quantum computing. Molecules obey the laws of quantum mechanics, which means that a computer with quantum states looks like a suitable tool for their modeling. Nitrogenase was an unofficial test for this idea after a meeting of Microsoft in 2011, where they discussed possible applications of quantum machines. Chan was then engaged in the enzyme for more than ten years and talked about its complexity.
In 2017, Microsoft researchers published статьюan article where they called nitrogenase a convincing test for quantum computers. The logic was understandable: if the active center of the enzyme contains a strongly confusing system of electrons, the usual methods may not have enough power. Chan from the very beginning was skeptical of this conclusion and believed that classical algorithms have not yet exhausted possibilities.
The new work does not describe the entire mechanism of nitrogenase from beginning to end. The researchers focused on a more basic question: what is the energy of the basic state of FeMo-co. The main state is an electronic configuration with minimal energy, from which the further description of the reaction begins. If this level is calculated incorrectly, all subsequent stages will also be unreliable.
The difficulty is that the active center contains a cluster of seven iron atoms, and each has several unpaired electrons. Their backs can take different directions, orbitals change, and the behavior of each particle depends on the neighbors. For FeMo-co, there are more than 78 000 plausible electronic configurations. The main condition is not one picture, but a balanced combination of many options.
Schrödinger’s equation in principle describes how all these configurations are formed into the final state and what energy the system should have. In practice, a direct exact solution for such a number of interacting electrons is almost inaccessible. The problem remains heavy for both classical machines and future quantum computers. In both cases, you need to start with a reasonable approximation - guesses about which configurations make the main contribution.
The difference between approaches begins. The classic algorithm tries to gradually consider additional configurations and prove that a huge part of the remaining can be safely discarded, because they almost do not change energy. A quantum computer, if a sufficiently powerful machine already existed, could imagine an initial guess as a quantum state and then bring it to the desired structure in time. Proponents of this path believe that this is where the advantage will appear.
Chan and his colleagues are arguing with this confidence. In their opinion, the quantum method still rests on the need for a good initial approximation, and the future equipment does not have an obvious advantage at this stage. In addition, the classical methods in recent years have greatly advanced. The new work shows that even in the FeMo-co level system, important configurations can be selected without a quantum iron.
The team used two independent methods of compressing a complex quantum state. In the first approach, the researchers began with approaching and gradually changing the behavior of small groups of electrons. They then checked whether larger changes made a noticeable contribution to energy. Such an analysis showed which configurations need to be left, and which can be excluded without loss of accuracy.
The second method is related to the direction that Chan developed most of his career. The initial state was divided into parts and limited the amount of information that can take place between them. After that, the researchers increased the limit and looked when further complication ceases to change the result. The idea was not to describe everything at once, but to find a fairly compact view for an extremely related electronic system.
Both methods gave the same energy assessment of the energy of the main state of FeMo-co. It also coincided with what scientists were waiting for experimental observations. This allowed the authors to conclude that they have found the true ground state of the used model of the active center.
The result does not cover the entire task of nitrogenase. To describe the work of the enzyme completely, you need to calculate the sequence of intermediate chemical states and understand how the system passes through the nitric conversion reaction into ammonia. It’s much more difficult than finding rest energy. But now chemists have a more reliable start for the next stage.
The dispute about quantum computers after this work will not end either. Some researchers believe that one solitary example does not prove the tolerability of the method to other systems. James Whitfield of Dartmouth College points out that long-term work on one task does not answer the question of whether entire classes of molecular problems can also be effectively solved.
Proponents of quantum computing also believe that the main advantage may not be in the calculation of one energy of the main state, but in modeling the development of the system over time. It is the dynamics of the reaction, the sequence of transitions and the change of the electronic state that can be much less convenient for classical algorithms.
Chan does not argue with the fact that quantum computers may be important for chemistry. If a powerful fault-tolerant machine already existed, he would like to use such a tool. His position is different: chemists should not consider the most complex tasks inaccessible before the appearance of the ideal quantum equipment. The story with FeMo-co shows that classical methods are still able to shift the boundary of the possible.
For nitrogenase, this result is especially important because the enzyme ceases to be only an abstract example in the dispute about future computers. Each new calculation brings scientists closer to understanding how the microbial molecular machine makes ammonia under a mild environment, while the Gaber-Bosch process was needed for a similar reaction. Now the question sounds more practical: how far it will be possible to move in classical methods before quantum computers become a real working tool of chemists.
Chemists at the California Institute of Technology have calculated one of the most difficult parts of the enzyme, without which life on Earth would not have gained access to atmospheric nitrogen. The work concerns nitrogenase, a protein that converts molecular nitrogen into ammonia. For theoretical chemistry, this is an important result in itself, but the dispute around it went far beyond biology: the calculation was performed on conventional computers, although nitrogenase for many years was given as a task where quantum machines should get a decisive advantage.
Garnet Chan, one of the leading specialists in quantum chemistry, has long been engaged not so much in the future of computing as with specific chemical tasks. He is interested in how to describe complex reactions in living systems. At the same time, it was Chan’s works that were at the center of a discussion on whether it is necessary to wait for full-fledged fault-tolerant quantum computers to deal with the heaviest molecular systems.
The new result strengthens Chan’s position. In early January, he and five other chemists from Caltech reached an important stage in the description of nitrogenase. They managed to calculate the state of the active center of the enzyme by classical methods, that is, without a quantum computer. For the area where the thesis of the almost inevitable superiority of quantum computing has been sounded for many years, this is a noticeable argument.
Nitrogenase refers to key molecules of terrestrial biology. Along with photosynthesis, nitrogen fixation is among the processes, without which a complex life would not be able to develop in the usual form. The atmosphere is almost 80% nitrogen, but it is in the air as a molecule N2. Two atoms are bound by a strong triple bond, so living organisms can’t just take this gas and embed it in proteins, DNA or other substances.
Before the advent of nitrogenazase, life depended on rare events that broke the bond in the nitrogen molecule. One such source was lightning: high-energy discharges turned part of atmospheric nitrogen into compounds suitable for biological processes. About three billion years ago, the early prokaryotes got a different path. Nitrogenase has learned to tear N2 and turn inert gas into ammonia, which can be used to build living matter.
The enzyme solves this problem through the active center FeMo-co. This is a cluster of iron and molybdenum atoms, where the main chemistry takes place. It is he who makes nitrogenase so difficult for calculations. Each iron atom carries four or five unpaired electrons, and their state depends not only on its own position, but also on the behavior of other particles in the cluster.
This is how the problem of electronic correlation arises. In simple models, electrons can sometimes be considered almost independently, but FeMo-co does not allow such rudeness. Particles are connected by quantum dependencies, so the change in one part of the system is reflected in another. It is necessary to understand the general electronic structure and energy of the cluster, and the number of possible configurations grows explosively.
Historically, people have long been interested not in the mechanism of nitrogenase, but ammonia as a practical product. In the 19 century, a reliable source of available nitrogen was guano from the islands off the coast of Peru, and this resource was valued so highly that it had conflicts between states. In 1909, Fritz Haber and Carl Bosch created an industrial process of fixing nitrogen, which allowed mass-produced ammonia for fertilizers. But the scientific mystery remains: soil bacterium does under normal conditions what industry performs in a rigid chemical mode.
Quantum computers entered this dispute because of the very nature of molecular calculations. An ordinary computer works with bits that take a value of zero or unit. Quantum uses qubits that can be in superposition and get confused with each other. For tasks with a suitable mathematical structure, such a machine is theoretically able to give a huge gain in speed.
Chemistry has long been considered one of the most natural uses of quantum computing. Molecules obey the laws of quantum mechanics, which means that a computer with quantum states looks like a suitable tool for their modeling. Nitrogenase was an unofficial test for this idea after a meeting of Microsoft in 2011, where they discussed possible applications of quantum machines. Chan was then engaged in the enzyme for more than ten years and talked about its complexity.
In 2017, Microsoft researchers published статьюan article where they called nitrogenase a convincing test for quantum computers. The logic was understandable: if the active center of the enzyme contains a strongly confusing system of electrons, the usual methods may not have enough power. Chan from the very beginning was skeptical of this conclusion and believed that classical algorithms have not yet exhausted possibilities.
The new work does not describe the entire mechanism of nitrogenase from beginning to end. The researchers focused on a more basic question: what is the energy of the basic state of FeMo-co. The main state is an electronic configuration with minimal energy, from which the further description of the reaction begins. If this level is calculated incorrectly, all subsequent stages will also be unreliable.
The difficulty is that the active center contains a cluster of seven iron atoms, and each has several unpaired electrons. Their backs can take different directions, orbitals change, and the behavior of each particle depends on the neighbors. For FeMo-co, there are more than 78 000 plausible electronic configurations. The main condition is not one picture, but a balanced combination of many options.
Schrödinger’s equation in principle describes how all these configurations are formed into the final state and what energy the system should have. In practice, a direct exact solution for such a number of interacting electrons is almost inaccessible. The problem remains heavy for both classical machines and future quantum computers. In both cases, you need to start with a reasonable approximation - guesses about which configurations make the main contribution.
The difference between approaches begins. The classic algorithm tries to gradually consider additional configurations and prove that a huge part of the remaining can be safely discarded, because they almost do not change energy. A quantum computer, if a sufficiently powerful machine already existed, could imagine an initial guess as a quantum state and then bring it to the desired structure in time. Proponents of this path believe that this is where the advantage will appear.
Chan and his colleagues are arguing with this confidence. In their opinion, the quantum method still rests on the need for a good initial approximation, and the future equipment does not have an obvious advantage at this stage. In addition, the classical methods in recent years have greatly advanced. The new work shows that even in the FeMo-co level system, important configurations can be selected without a quantum iron.
The team used two independent methods of compressing a complex quantum state. In the first approach, the researchers began with approaching and gradually changing the behavior of small groups of electrons. They then checked whether larger changes made a noticeable contribution to energy. Such an analysis showed which configurations need to be left, and which can be excluded without loss of accuracy.
The second method is related to the direction that Chan developed most of his career. The initial state was divided into parts and limited the amount of information that can take place between them. After that, the researchers increased the limit and looked when further complication ceases to change the result. The idea was not to describe everything at once, but to find a fairly compact view for an extremely related electronic system.
Both methods gave the same energy assessment of the energy of the main state of FeMo-co. It also coincided with what scientists were waiting for experimental observations. This allowed the authors to conclude that they have found the true ground state of the used model of the active center.
The result does not cover the entire task of nitrogenase. To describe the work of the enzyme completely, you need to calculate the sequence of intermediate chemical states and understand how the system passes through the nitric conversion reaction into ammonia. It’s much more difficult than finding rest energy. But now chemists have a more reliable start for the next stage.
The dispute about quantum computers after this work will not end either. Some researchers believe that one solitary example does not prove the tolerability of the method to other systems. James Whitfield of Dartmouth College points out that long-term work on one task does not answer the question of whether entire classes of molecular problems can also be effectively solved.
Proponents of quantum computing also believe that the main advantage may not be in the calculation of one energy of the main state, but in modeling the development of the system over time. It is the dynamics of the reaction, the sequence of transitions and the change of the electronic state that can be much less convenient for classical algorithms.
Chan does not argue with the fact that quantum computers may be important for chemistry. If a powerful fault-tolerant machine already existed, he would like to use such a tool. His position is different: chemists should not consider the most complex tasks inaccessible before the appearance of the ideal quantum equipment. The story with FeMo-co shows that classical methods are still able to shift the boundary of the possible.
For nitrogenase, this result is especially important because the enzyme ceases to be only an abstract example in the dispute about future computers. Each new calculation brings scientists closer to understanding how the microbial molecular machine makes ammonia under a mild environment, while the Gaber-Bosch process was needed for a similar reaction. Now the question sounds more practical: how far it will be possible to move in classical methods before quantum computers become a real working tool of chemists.
