Select a Resistor

587KΩ +/- what tolerance?

With enough money you can have a custom part made for you if you really do need this atypical part value. I greatly doubt that your circuit requires this atypical specific part. You can get a 590 k ohm resistor or as Del suggested use two more standard value resistors in parallel or series to achieve this atypical value.

These suggestions though do not address your un-punctuated question. How anyone properly selects an alternative passive component is that a sketch of all of the relevant components must be drawn so that a knowledgeable person can do a circuit analysis. The analysis will tell the tolerance of the resistor value, power handling, thermal drift concerns and possibly a few more esoteric aspects of the resistor to be replaced. Additionally the visual examination will determine the component packaging required to allow the replacement to mechanically fit the location.

Since these ideas are new to you, you should not attempt to find a replacement. I suspect that you actually have a typographic error and that the real component value is a readily available part. The engineer that designed the circuit needs to clarify what is actually needed.

A 720kΩ and a 3.3MΩ connected in parallel gets within 1%, as does 820kΩ and 2.2MΩ, and 630KΩ and 8.1MΩ. Any good?

Are there color stripes or dots on this resistor? If so, what are they? How do you know it is 587K Ω?

Outside a tier2 or 3 standards lab I have not encountered such odd values. Real circuits may be calculated with it, but it is a rookie mistake.

Because: nobody can build it in real life, nobody can test it, nobody can maintain it. Other than that, it is a sure sign, that he student selected a wrong topology for the task.

The right topology: you decide to compare 2 nearly equal values. Both inputs of the opamp are connected via 100k / 1k dividers. That takes advantage some facts. Opamps may differ in parameters by a wide margin, but the 2 inputs on the same chip are quite similar. 5% resistors coming from the same batch are usually well within 1% to each others. Their temperature coefficient tracks even 10x better.Most opamps have a separate offset trim input to improve on this further. At that point, we are at a few parts per thousand tracking, reliably in a real life circuit. And that translates straight to the precision of the measurements.

I am not in the habit of commenting on my ratings.

But, this is idiotic. I gave the guy a way out of the box, he painted himself. While he being an awful conversation partner.

Frequently enough, a guy needs a pointer to his problem. NOT to his question asked from simple ignorance.

Pay attention, what is needed.