After you've taken an optimization class, if you're not careful, everything starts looking like an optimization problem. If you want to frame dating as one, that's certainly doable.

The first part of formulating the problem is determining what the objective function is. Clearly this varies for different people and across different phases of life; you can view this as setting different tradeoff parameters for the various parts of the objective, some of which might include...

• short-term personal happiness -- having someone to talk to and hang out with to and cuddle with etc.; in the case of rich significant others, there are the perks of nice gifts and nice vacations
• convenience -- having someone to go to events together with, like weddings or formal dinners; or more casual things, if many of your friends are in relationships and fun activities for your social circles are often couple activities
• reputation -- on the positive side, this could be "dating upwards" and social mobility; on the negative side, people might earn reputations of being “easy” or noncommittal for dating around too much, or appear undesirable for dating after long periods of singleness
• practice -- getting better at dating and being in a relationship, in ways like being thoughtful towards a significant other, learning to put yourself second, being able to back off from silly arguments, meeting and getting along with a significant other's friends, etc.
• accordance with social norms -- aka not stealing a best friend's significant other or pursuing someone in a relationship
• conformance to familial expectations and pressures -- parents often have strong notions of what constitutes "acceptable" for their children
• long-term happiness and emotional security -- related to the institution of marriage, and to potential child-rearing
• long-term financial security -- self-explanatory

Depending on the person, some of these may actually be hard constraints. (You can think of this as having formed the Lagrangian and included the constraints into the objective.)

I actually think this problem is best formulated as a dynamic programming problem (in the optimization sense, not the algorithmic sense).

To give some background here, a stochastic system at time t consists of three parts:

1. the state of the system x_t
2. the noise w_t, or random disturbances from the environment to the state
3. the control decision u_t

Given the state of the system and what you know about the distribution of noise, you want to make control decisions at each time step to optimize the objective over a (finite, in this case) time horizon.

To map our dating problem to the stochastic formulation: Considering

1. the current set of social / dating / friend relationships that concern you -- how many of your close friends are in relationships; how happy you are with your current relationship status; who are your potentials; etc.
2. plus the potential of change in these dynamics -- who knows, maybe a new girl will move in next door; the guy you like will break up with his girlfriend; etc.
3. you make decisions that maintain or attempt to change the status quo -- breaking up with someone; flirting with someone; proclaiming your love to someone; etc.

that you expect will maximize your rewards accumulated over the time horizon. These rewards are the different parts of the objective function, evaluated at each time interval, weighted in a way that corresponds to how important each is to you (cf: tradeoff parameters).