In 1963 I applied for academic positions in the United States from Birmingham, England, where I was finishing two postdoctoral years in Rudolf Peierls’ wonderful department. In those early post-Sputnik halcyon days you then applied for assistant professorships and collected the offers that came in reply. Peierls had suggested that I send such a letter to Walter Kohn. Disappointingly, he offered only a two-year postdoc. But I had never been to California and Peierls clearly had a high regard for the man, so this otherwise noncompetitive proposal was tempting. I asked the people at Cornell, who did come through with a faculty position, whether I could defer it for two years to do a postdoc in La Jolla. They said two would be too many but one would be OK. Walter said one was fine with him, so in August 1963 I showed up in La Jolla.

Walter was not there. He was finishing a sabbatical in Paris. But there were several wonderful postdocs, the Physics Department was still located right on the beach, Dorothy and I found a house in Del Mar on a cliff overlooking the Pacific for $128 a month, and life was good. (I had written a letter from Birmingham to an acquaintance from college who was then a postdoc in La Jolla, asking what it was like there. All I remember from his reply was “Volleyball is standard on the beach at noon.”)

Eventually the moment of truth arrived. The boss returned. The holiday threatened to end. Walter invited me to his office to say hello. It was immediately clear that this was a kind, charming, witty man. After we had exchanged pleasantries, he told me about a little theorem he and Pierre Hohenberg had proved back in Paris. The proof was one of those clever three-line arguments that wouldn't have occurred to me if I had thought about it for a hundred years, but was utterly simple and transparent when Walter laid it out in front of me.

He asked me to think about how to generalize the theorem from the ground state to thermal equilibrium. I returned to my office to consider it, and quickly realized that a new variational principle for the free energy that I had formulated in Birmingham for an utterly unrelated purpose, seemed to be tailor-made for generalizing the Hohenberg–Kohn theorem to nonzero temperature.